docNHS CFHEP 001 Extension: Final report
download 
Report Transcription
NHS CFHEP 001 Extension: Final report
The Impact of eHealth on the Quality and Safety of Healthcare An updated systematic overview & synthesis of the literature Final report for the NHS Connecting for Health Evaluation Programme (NHS CFHEP 001) Aziz Sheikh, Susannah McLean, Kathrin Cresswell, Claudia Pagliari, Yannis Pappas, Josip Car, Ashly Black, Akiko Hemmi, Ulugbek Nurmatov, Mome Mukherjee, Brian McKinstry, Rob Procter and Azeem Majeed March 2011 Table of contents List of authors Acknowledgements Abbreviations Executive summary SECTION 1: BACKGROUND 1. Introduction SECTION 2: METHODS AND FORMATIVE WORK 2. Methods 3. The National Programme for Information Technology, NHS Connecting for Health and the Information Revolution 4. Exploring, describing and integrating the fields of quality, safety and eHealth SECTION 3: FINDINGS Descriptive overview of results 5. Health information exchange and interoperability 6. Electronic health record 7. Computer-assisted history taking systems 8. Computerised decision support systems 9. Case study: Computerised decision support in mammography screening 10. ePrescribing systems 11. Case study: ePrescribing for oral anticoagulation therapy in primary care with warfarin 12. Case study: eHealth-based approaches to reducing repeat drug exposure in patients with known drug allergies 13. Telehealthcare 14. Case study: Telehealthcare for asthma 15. Case study: Internet-based interventions for smoking cessation 16. Human factors 17. Importance of organisational issues in the implementation and adoption of eHealth innovations 18. Case study: Lessons in relation to the design, implementation and adoption of the NHS Care Record Service in secondary care SECTION 4: CONCLUSIONS, DISCUSSION PRIORITIES 19. Conclusions, discussion and recommendations 20. Future research priorities AND FUTURE RESEARCH SECTION 5: APPENDICES AND GLOSSARY Appendix 1: Search strategy Appendix 2: Quality assessment form Appendix 3: PRISMA flow diagrams Appendix 4: Included systematic reviews Appendix 5: Data extraction and critical appraisal scores of included reviews Appendix 6: Programmes for eHealth Masterclasses Appendix 7: Academic outputs Glossary 2 List of authors and contact details Ashly D. Black, Research Assistant, Health Unit, Department of Primary Care and Social Medicine, Imperial College London Dr Josip Car, Director of eHealth Unit, eHealth Unit, Department of Primary Care and Social Medicine, Imperial College London Kathrin Cresswell, Research Associate, eHealth Research Group. Centre for Population Health Sciences, The University of Edinburgh Dr Akiko Hemmi, Research Fellow, eHealth Research Group, Centre for Population Health Sciences, The University of Edinburgh Professor Azeem Majeed, Professor of Primary Care, Department of Primary Care and Social Medicine, Imperial College London Dr Susannah McLean, Clinical Research Fellow, eHealth Research Group, Centre for Population Health Sciences, The University of Edinburgh Dr Brian McKinstry, Reader, eHealth Research Group, Centre for Population Health Sciences, The University of Edinburgh Mome Mukherjee, Senior Research Analyst, eHealth Research Group, Centre for Population Health Sciences, The University of Edinburgh Dr Ulugbek Nurmatov, Clinical Research Fellow, eHealth Research Group, Centre for Population Health Sciences, The University of Edinburgh Dr Claudia Pagliari, Senior Lecturer in Primary Care & Chair of eHealth Group Centre for Population Health Sciences, The University of Edinburgh Dr Yannis Pappas, Deputy Director eHealth Unit, Department of Primary Care and Social Medicine, Imperial College London Professor Rob Procter, Professor of e-Social Science, Manchester eResearch Centre, The University of Manchester Professor Aziz Sheikh, Professor of Primary Care Research & Development, Centre for Population Health Sciences, The University of Edinburgh Correspondence to: A Sheikh, Centre for Population Health Sciences, The University of Edinburgh, Teviot Place, Edinburgh EH8 9AG, Email: [email protected] 3 Acknowledgements Identifying, retrieving, collating, synthesising and interpreting the vast body of evidence that we have drawn on in compiling this review has been a challenging task and would not have been possible without the help of a number of people who we here take great pleasure in acknowledging. Throughout the process of undertaking this work we have had helpful support from colleagues at the NHS Connecting for Health Evaluation Programme, namely Professor Richard Lilford, Jo Foster, Nathalie Maillard and Lee Priest. Richard, Jo and Lee have kindly also represented the funders on our External Steering Group which, under the able chairmanship of Professor Denis Protti, and with the helpful support of Professor David Bates, Professor Jeremy Wyatt, Dr Maureen Baker, Mr Antony Chutter and Ms Alison Turner provided thoughtful and constructive advice. Marshall Dozier, The University of Edinburgh’s librarian kindly helped with developing the search strategy and we were aided in the selection of references by Dr Tomislav Bokun, Matko Marlais and Chuin Ying Ung. Dr Chantelle Anandan contributed to an earlier iteration of this report. We have had the welcome opportunity to present our findings to colleagues at a number of meetings both in the UK and internationally and we are grateful for the feedback that has been received, which has either explicitly or implicitly informed our thinking. In addition, Dr Chris Burton, Dr Sangeeta Dhami, Dr Bernard Fernando, Professor Dave Fitzmaurice, Professor MariePierre Gagnon, Dr Hilary Pinnock, Professor Michael Sharpe and Dr Paul Taylor kindly provided critical feedback on earlier versions of draft chapters, as also did members of the External Steering Group. Our thanks also to the three external reviewers who provided constructive suggestions on an earlier version of this report. Our thanks are also due to the many colleagues–both from the UK and abroad–whom we have had the pleasure to work with on the various 4 Cochrane systematic reviews, in particular Dr Marta Civljak, Dr Lindsay Stead, Dr Joe Liu and David Cahndler who worked with us on exemplar systematic reviews incoroporated into this report. Our thanks also to the many colleagues who have supported us in putting on the series of eHealth Masterclasses. Anna Wierzoch provided administrative support and also proof-read the final manuscript. To all of these individuals, who so generously gave us their time, we wish to record our sincerest thanks. Finally, we are grateful to publishers to allow reproduction of figures, boxes, tables and text referred to in individual chapters. Our thanks are in this latter respect particularly due to The Cochrane Collaboration. 5 Abbreviations ADE ADR AHR AHRQ AI AMR AQLQ ATC BA BCSH BMI BNF BP BS7799 BWH C C&B CAD CAHTS CASP CBMR CBPR CBPRS CCT CI CT CCR CDR CDSS CFHEP CHI CHIN CIS CMR COPD COTS CP CPOE CPR CPT CRDB CRS CS CSA CSC CTS DCR Adverse drug events Adverse drug reaction Automated Health Records Agency for Healthcare Research and Quality Artificial intelligence Automated Medical Records Asthma Quality of Life Questionnaire Anatomical Therapeutic Chemical Classification Index Before after (pre-post) study British Committee for Standards in Haematology Body Mass Index British National Formulary Blood Pressure British Standard for Information Security Management Brigham and Women’s Hospital Case study Choose and Book Computer-aided detection or diagnosis Computer-assisted history taking systems Critical Appraisal Skills Programme Computer-Based Medical Record Computer-Based Patient Record Computer-Based Patient Record System Controlled Clinical Trial Confidence Interval Controlled Trial Continuity of care record Clinical Data Repository Computerised Decision Support System See NHS CFHEP Community Health Index number Community Health Information Network Clinical information system Computerised medical record Chronic Obstructive Pulmonary Disease Commercial off-the-shelf Cerebral Palsy Computerised Provider Order Entry Computerised patient record Current Procedural Terminology Care Record Development Board see NHS CRS Cross-sectional study Clinical Spine Application Computer Sciences Corporation Computer Telephony System Detailed Care Record 6 DH DIN DIRC dm+d DMR DSS ECR ECRI ECS EHCR EHR EM eMAR EMIS EMR EPHR EPR EPS ERP ETD ETP EV FAA FDA GEP-HI GINA GMS GP GPRD GPSoC HbA1c HCI HF HFE HIE HIEI HIMSS HIS HL7 HRO HT HTA ICBI ICD ICHPPC ICNP ICPC ICR ICU IEHR Department of Health Doctor Independent Network Dependable Interdisciplinary Research Collaboration Dictionary of Medicines and Devices Digital Medical Record Decision support system Electronic Care Record Emergency Care Research Institute Emergency Care Summary Electronic Health Care Records Electronic Health Record Experimental methods Electronic medication administration record Egton Medical Information Systems Electronic medical record Electronic Personal Health Records Electronic Patient record Electronic Prescription Service Enterprise resource planning Education Training and Development Electronic Transmission of Prescriptions economic evaluations Federal Aviation Authority Food and Drug Administration Guidelines for Best Evaluation Practices in Health Informatics Global Initiative for Asthma General Medical Services General Practitioner General Practice Research Database GP Systems of Choice Glycated Haemoglobin Human-computer interaction Human factors Human factors engineering Health information exchange Health information exchange and interoperability Healthcare Information and Management Systems Society Hospital information system Health Level Seven High Reliability Organizations hypothesis testing Health Technology Assessment Internet and Computer Based Interventions International Classification of Diseases International Classification of Health Problems in Primary Care International Classification of Nursing Practice International Classification of Primary Care Integrated Care Record Intensive care unit Interoperable electronic health record 7 IM&T Information management and technology InPS In Practice System INR International Normalised Ratio IQAP Information Quality Assurance Programme IP Internet protocol ISAAC International Study of Asthma and Allergies in Children ISB Information Standards Board ISO International Organization for Standardization ISIP Integrated Service Improvement Programme IT Information technology JACHO Joint Commission on Accreditation of Healthcare Organizations KLAS Keystone Library Automation System LAN Local Area Network LHR Longitudinal health record LIS Laboratory information system LOINC Logical observation identifiers names and codes LPfIT London Programme for Information Technology LSPs Local Service Providers MA Meta-analysis ME Medical Error MeSH Medical Subject Headings MHRA Medicines and Healthcare Products Regulatory Agency MRC Medical Research Council MRHA Medicines and Healthcare Products Regulatory Agency N narrative NANDA North American Nursing Diagnoses N3 National Network for the NHS N3SP N3 Service Provider NASP National Application Service Provider NHLBI National Heart Lung and Blood Institute NHS National Health Service NHS BSP NHS Breast Screening Programme NHS CFH NHS Connecting for Health NHS CFHEP NHS Connecting for Health Evaluation Programme NHS CRS NHS Care Records System NIC Iowa Nursing Intervention Classification NICE National Institute for Health and Clinical Excellence NIH National Institute for Health NIHR National Institute of Health Research NISP National Infrastructure Service Provider NKS National Knowledge Service NLOP National Programme for IT Local Ownership Programme NOC Iowa Nursing Outcome Classification NPfIT National Programme for Information Technology NPSA National Patient Safety Agency NPT Near patient test NS Not Specified NSF National Service Framework OAT Oral anticoagulation therapy OBS Output-based specification 8 OCCO OITIM OSCHR OSI OXMIS P PA PACS P&P PACQLQ PAQLQ PAS PDA PBR PC PCEHR PCR PCT PDA PDF PDS PEHR PHQ PHR PICS PMCS PMR PMRI PMS POC PRIMIS PRO PSIS QI QMAS QOF QoS R RCT RDBT RFID RHIOs RID RTC Rx SBCCE SC SCI SCR SDS Office of the Chief Clinical Officer Organisation Information Technology/System Innovation Model Office for Strategic Coordination of Health Research Open System Iinterconnection Oxford Medical Information Systems Prospective Predictive analysis Picture Archiving and Communications System Pen and paper Paediatric Asthma Carer’s Quality of Life Questionnaire Paediatric Asthma Quality of Life Questionnaire Patient administration systems Personal Digital Assistant Payment by Results Personal computer Person-Controlled Electronic Health Record Patient care record Primary Care Trust Personal digital assistant Portable Document Format Personal Demographic Service Personal electronic health record Patient Health Questionnaire Personal health record Prescribing Information and Communication System Predictive modelling of cost savings Patient (carried) medical records Patient medical record information Personal medical services Point-of-care Primary Care Information Services Patient reported outcomes Personal Spine Information Service Qualitative interview Quality Management and Analysis Subsystem Quality and Outcomes Framework Quality of Service retrospective Randomised controlled trial NHS CFH Requirements, Design, Build and Test Radiofrequency identification Regional Health Information Organizations Regional Implementation Director Roadmap for Transformational Change ePrescribing Santa Barbara County Care Exchange Secondary Care Scottish Care Information Summary Care Record Spine Directory Service 9 SE Software engineering SHAs Strategic Health Authorities SI Service implementation SIGN Scottish Intercollegiate Guidelines Network SMD Standard mean difference SMS Short message service SNOMEDCT Systematized Nomenclature of Medicine Terminology-Clinical Terms SOAP Subjective information, objective information, assessments and plan SR Systematic review STARE-HI Statement on Reporting of Evaluation Studies in Health Informatics SUS Secondary Uses Service TGC Tight Glycemic control TMS Transaction Messaging Service TRaMS Training Messaging Service TS Time series UK United Kingdom UKCRN United Kingdom Comprehensive Research Network US United States of America VATAM Validation of Telematics Applications in Medicine VME Vancomycin resistant Enterococcus faecium VPN Virtual private network WAN Wide area network WHO HEN World Health Organization’s Health Evidence Network 10 Executive summary INTRODUCTION • Increasing life expectancy, improved survival in people with acute and long-term conditions and a greater array of available treatment options are combining to place an increasing burden on healthcare systems internationally. • This picture is particularly true of the economically developed world where high salaries for healthcare professionals and ever-increasing public expectations contribute to the challenges facing governments trying to contain spending on healthcare provision and planning. Similar pressures are also emerging in economic transition and developing countries. • There is now a substantial body of research, both domestic and international, identifying considerable shortcomings in the current provision of healthcare. • Key issues emerging from this literature are substantial variations in the quality of healthcare, the considerable risks of iatrogenic harm and inefficiencies in healthcare provision; this latter issue has become a particularly acute cause for concern in the light of the economic downturn and the resulting pressures on healthcare budgets. • These failings contribute in a major way to the high rates of potentially avoidable morbidity, mortality and healthcare expenditure. • There have been substantial developments in information technology (IT), hardware and software capabilities over recent decades and there is now considerable potential to apply these technological developments in relation to aspects of healthcare provision (the application of IT in this way will henceforth in this report be referred to by the term “eHealth”). • Of particular international interest is the development and deployment of an array of eHealth applications, with a view to improving the quality, safety and efficiency of healthcare delivery. 11 • Whilst these eHealth applications have considerable potential to aid professionals in delivering healthcare and patients in self-management, it is not widely appreciated that use of these new technologies may also introduce significant new risks to patients. • Also of concern is that even when potentially useful interventions are developed and deployed, they frequently fail to live up to their potential when deployed in the “real world”; a major factor contributing to this paradox is the failure of these technological fixes to integrate effectively with existing work patterns or to be adequately conceptualised as a change management reform which can then be used to reshape delivery of care. • Given that England’s National Health Service (NHS) is engaged in one of the largest eHealth-based modernisation programmes in the world, it is appropriate and timely to critically review the international eHealth literature with a view to identifying lessons that can usefully be learned with respect to the development, design, deployment, integration and evaluation of eHealth applications. AIMS AND OBJECTIVES • We were originally commissioned by the Patient Safety Research Programme and then by the NHS Connecting for Health Evaluation Programme (NHS CFHEP) to produce a systematic overview of the literature examining the effectiveness of IT (eHealth) applications to improve the quality and safety of healthcare. • We focused in our initial phase of work on evidence relating primarily to: o the storage and retrieval of medical information o tools to support healthcare professionals in making clinical decisions o ways of promoting the effective development, deployment and use of eHealth applications in routine healthcare settings. • In the second phase of work, this was expanded to: o include evidence on delivering healthcare from a distance 12 o undertaking de novo systematic reviews in key eHealth areas o co-ordinating a series of “eHealth Masterclasses” aimed at policymakers, academics and practitioners o developing a database of the critically appraised and critiqued eHealth literature o updating and integrating this with the previous body of work on storing and managing healthcare information, supporting clinical decision making and promoting the successful adoption of eHealth applications. • This report focuses on providing a synopsis of the main findings from the synthesis of this body of work: findings from the new Cochrane systematic reviews that we have initiated and/or contributed to are integrated, where appropriate, into chapters and a summary of these outputs is detailed along with the programmes for the Masterclasses in the report’s appendices; the final database of eHealth systematic reviews will, with the agreement of the funders, be submitted later this year. An application to support the publication of this database through NHS Evidence will be submitted in April 2011. METHODS AND FORMATIVE WORK Methods • We conducted a systematic search and critique of the empirical literature on eHealth applications and their impact on the quality and safety of healthcare delivery and synthesised this with relevant theoretical, technical, developmental and policy-relevant literature with a view to producing an authoritative and accessible overview of the field. • Whilst we drew on established Cochrane review principles to systematically search for, critique and synthesise the literature, this approach needed to be adapted in several respects in order to produce a meaningful overview of the literature (see below). • Searching the literature was complicated by the lack of internationally (or indeed in some cases nationally) agreed terminology relating to 13 eHealth applications, the lack of agreed definitions of quality and safety, and consequently poor indexing of these constructs in databases of published literature. • In order to undertake a thorough review of the literature, we therefore needed to undertake initial developmental work to formulate a comprehensive search strategy. • Using the set of comprehensive Medical Subject Headings (MeSH) and free text search terms developed, we systematically searched major medical databases over a 13-year period (1997–2010) to identify systematic reviews, technical reports and health technology assessments, and randomised controlled trials investigating the effectiveness of eHealth applications. The specific databases searched were: o MEDLINE o EMBASE o The Cochrane Database of Systematic Reviews o Database of Abstracts of Reviews of Effects o The Cochrane Central Register of Controlled Trials o The Cochrane Methodology Register o Health Technology Assessment Database and NHS Economic Evaluation. • We, in addition, searched key national and international databases to identify unpublished work and research in progress. • The systematic reviews have then been subjected to critical review using the Critical Appraisal Skills Programme (CASP) approach, adapted for use with eHealth applications. • The de novo systematic reviews that have contributed to this review have been conducted using established Cochrane review methods. • To provide a broader appreciation of the context of this work and furthermore to aid conceptual development and interpretation of findings, we supplemented this systematic search for empirical evidence with a more emergent approach to identify relevant background and theoretical literature in relation to the essential concepts underpinning this overview—namely: eHealth; quality; safety; 14 and efficiency. This involved drawing on our personal databases of relevant papers, identifying seminal papers and reports as well as searching the grey literature. • The overall body of literature identified was too diverse to make any quantitative synthesis of the literature meaningful, nor indeed was it possible. Rather, we chose to qualitatively synthesise the literature drawing on the relevant preliminary conceptual work to guide this narrative synthesis. • Our overall assessment of the volume and strength of evidence in relation to key findings are summarised in this Executive summary using a modified version of the World Health Organization’s Health Evidence Network (WHO HEN) system for public health evidence, which grades evidence into three main categories: o strong (consistent, good quality, plentiful or generalisable) o moderate (consistent and good quality) o limited to none (inconsistent or poor quality). NHS Connecting for Health and the National Programme for Information Technology • The NPfIT remains one of the most comprehensive, ambitious and expensive eHealth-based overhauls of healthcare delivery ever undertaken. • This Programme has its origins in the 1998 Department of Health strategy Information for Health, which committed the NHS to lifelong electronic health records for everyone with round-the-clock, on-line access to patient records and information about best clinical practice for all NHS clinicians. The current Programme, launched in 2002, was to be a 10-year initiative aiming initially to create the infrastructure, tools and environment through which it is possible to deliver: o a longitudinal electronic patient record (from “cradle to grave”) accessible to multiple users throughout the NHS; this (i.e. NHS Care Records Service or NHS CRS) together with the dedicated NHS broadband (National Network for the NHS or N3) and the 15 national database on which these records will be held (the Spine); represented the backbone of the Programme o a service through which prescriptions can be transferred electronically from the general practitioner and other prescribers to pharmacists (Electronic Prescriptions Service) and eventual integration of this with NHS CRS (Electronic Transmission of Prescriptions or ETP) o an electronic appointment booking service enabling general practitioners to electronically book hospital appointments (Choose and Book). • The Programme was however subsequently expanded to include amongst other things: o a Picture Archiving and Communication System (PACS) o GP2GP, which is a system that enables transfer of patient records between GP practices o Quality Management Analysis System (QMAS), which automates assessment of GP practice performance against criteria included in the new GP contract o ePrescribing. • Whilst these represented the headline deliverables of the Programme, our scoping of the field has identified a number of other related eHealth projects or applications which are also closely inter-connected with the delivery of the Programme. • Originally managed directly by the Department of Health, oversight of NPfIT was transferred in 2005 to a newly created arm’s length body, namely NHS CFH. • Foremost amongst the roles of NHS CFH was responsibility for nationally procuring systems and services that will be needed to ensure delivery of NPfIT. • Given the extremely high level of public expenditure, the Programme attracted considerable public, professional, legal, financial, political and international scrutiny. 16 • It is thus probably no exaggeration to say that in addition to it being the most comprehensive, ambitious and expensive eHealth reform programme in the world, it is also the most influential in that its success or failure is likely to have major domestic and international ripples for many years to come. • NHS CFH has more recently been subsumed into the Department of Health’s Informatics Directorate. • Following the election of the new coalition government in 2010, the scope and ambition of NPfIT/NHS CFH has been reduced with a rapid move to less centralised and more local procurement and implementation of IT into NHS healthcare settings. Exploring, describing and integrating the fields of quality, safety and eHealth • eHealth remains a relatively new and rapidly evolving field and so many of the concepts, terms and applications are still in a state of flux. • There is furthermore no agreed definition of eHealth, with some researchers using this to relate primarily to the area of consumer informatics, whereas others use it more generically to refer to any of the ways in which IT can be employed to aid the deliver of healthcare. For the purposes of this review, we considered it important to use an inclusive definition and chose to use Eysenbach’s definition, as adapted by Pagliari, for the basis for our work: ‘eHealth is an emerging field of medical informatics, referring to the organisation and delivery of health services and information using the Internet and related technologies. In a broader sense, the term characterizes not only a technical development, but also a new way of working, an attitude, and a commitment for networked, global thinking, to improve healthcare locally, regionally, and worldwide by using information and communication technology.’ • Whilst the number of eHealth applications is potentially vast, these can nonetheless be divided into three broad domains relating to key activities they support: o storing, managing and sharing data 17 o informing and supporting clinical decision-making o delivering expert professional and or consumer care remotely. • The effective commissioning, development, deployment and routine use of eHealth applications is a cross-cutting area that impacts on each of these three domains. Quality • There are no internationally agreed definitions of healthcare quality. • Most frameworks of quality currently in use do, however, incorporate the following key dimensions of care: o effectiveness of treatments o appropriateness of means of delivery o acceptability o efficiency o equity. Safety • Whilst there are no internationally agreed definitions of patient safety, adaptations of the National Patient Safety Agency’s definition of Patient Safety Incidents are increasingly being used. This, in its original form, states that: ‘A patient safety incident is any unintended or unexpected incident which could have harmed or did lead to harm for one or more patients being cared for by the NHS.’ • There are a number of patient safety taxonomies currently in existence, however, our scoping of this literature found that the Joint Commission on Accreditation of Healthcare Organizations (JCAHO) Patient Safety Event Taxonomy is the most comprehensive and clinically relevant in that it incorporates five key primary areas: o impact of medical error o type of processes that failed o domain, i.e. the setting in which an incident occurred o cause or factors leading to the safety incident 18 o prevention and mitigation factors to reduce risk of recurrence and or improve outcomes in the case of a further incident. Integrating eHealth, quality and safety • Integrating the fields of eHealth, quality and safety clearly demonstrates the numerous ways in which technology has the potential to improve the efficiency of many facets of healthcare delivery. This can occur through, for example, helping clinicians readily to access comprehensive information on their patients, aiding monitoring of their conditions and the treatments being issued, reducing inappropriate variability in healthcare delivery, and proactively identifying and alerting clinicians to threats to patient safety. • This integration of these domains, however, also highlighted the many ways in which introduction of new eHealth applications could inadvertently increase risks. MAIN FINDINGS • Our initial searches retrieved a total of 67 systematic reviews. Updating and expanding this review during the second phase of work has identified an additional 95 systematic reviews. We thus in total identified 162 systematic reviews published during the period 1/1/199731/10/2010. • We have initiated 20 eHealth-related systematic reviews during the course of the second phase of work, the majority of which have or are now nearly completed. The key findings from this body of work have also been incorporated into the review. • The volume of primary and secondary literature is large, rapidly expanding, poorly collated and of very variable quality; as a result the literature surrounding eHealth poses major challenges to synthesis and interpretation. • In synthesising the available evidence, we used the following generic approach in relation to different eHealth applications and their related considerations: 19 o clarifying definitions, description and scope for deployment o drawing on conceptual maps to reflect on the potential benefits and risks of each application o identifying the empirically demonstrated benefits and risks, using exemplar subject areas and or detailed case studies on issues that are of direct or potential future relevance to the UK o based on a synthesis of the above, highlighting the policy, clinical and research implications for the individual areas of interest with a view to realising the potential that eHealth has to offer. • Our findings’ chapters are grouped together in relation to the four main foci of this report, namely the domains of: o managing, storing and transmitting data o supporting clinical decision-making o supporting the remote delivery of care o the cross-cutting issue of the socio-techno-cultural dimensions of developing and implementing eHealth applications. Data storage, management and retrieval Health information exchange and interoperability • Effective and efficient sharing of clinical information is essential to the future development of modern healthcare systems, which are increasingly characterised by the involvement of many specialist healthcare providers, often working from different sites, contributing to the care of individual patients. • The ideal in this respect is for professionals and patients themselves to have the ability simultaneously to access and seamlessly transfer, contribute to and integrate clinical data from disparate sources. • The potential gains in relation to improving the quality, safety and overall efficiency of healthcare delivery are potentially enormous, as suggested by a recent US-based economic analysis. • Most UK healthcare settings are however currently characterised by relatively low levels of health information exchange and interoperability 20 (HIEI) capability, this being particularly true of the hospital sector, where paper-based records are still the main means of recording and communicating clinical information. • NPfIT has however substantially increased the potential for HIEI, for example, through N3 and the Spine; deployment of the NHS Care Record Service (NHS CRS) aims to result in the creation of summary and detailed electronic health records that have the potential to be shared, to varying degrees, across healthcare settings and between providers. • NHS CFH has insisted that new eHealth applications must be Health Level Seven (HL7) compliant—a voluntary but nonetheless widely used standard for interoperability—thereby assuring a degree of ability to exchange information between systems. However, while this is undoubtedly welcome, none of the headline NPfIT applications will achieve the ideal levels of HIEI, with the result that patient safety may needlessly be compromised. • The current empirical evidence-base that supports the role of HIEI in improving organisational efficiency, practitioner performance or indeed any clinical patient outcomes.is, however, weak. • Although improving HIEI to the optimal level to allow seamless transfer and access to data in all settings, will probably result in societal costsavings in the longer run, this will inevitably require considerable upfront investment in hardware and software • An important paradox to further developments in this area is that whilst increasing levels of HIEI are clearly desirable for many reasons, greater availability of data also inevitably increases the risk of threats to breaches of patient confidentiality and data security. • Key outstanding issues that face healthcare systems in realising the potential for seamless exchange of information include the need to develop and deploy standard coding structures across all care settings (e.g. using Systematized Nomenclature of Medicine-Clinical Terms (SNOMED-CT)), facilitate integration of the increasing amounts of patient-generated data (e.g, through HealthSpace, home sensors 21 and/or telemetry devices), and improve secure audited access to electronic records to minimise the risks of breaching confidentiality. • Making progress in optimising the ability to share and integrate data seems particularly relevant given the expressed government objective of creating an “information revolution” within the NHS. Electronic health record • The electronic health record (EHR) represents the backbone of all major international eHealth developments currently taking place internationally, including NHS CFH’s NPfIT. • The ultimate goal is to have available comprehensive longitudinal health information for all members of the population, with the potential for accessing and contributing to these records by multiple users working across a range of healthcare settings; no country has however yet to achieve this comprehensively and if successful the NHS CRS will be one of the first in the world to come close to this aspiration. • Electronic health records range from simple storage devices to those with varying degrees of added functionality, including the ability to electronically prescribe (ePrescribing) and access to computerised decision support systems (CDSSs), which are active knowledge systems, which use individual patient data to generate case-specific advice. • The main potential advantages of EHRs relate to improved legibility and, particularly if structured templates are used, comprehensiveness of recording information, access by multiple users that is not geographically-bound (if interoperable), the ability to incorporate professional support tools, and time and cost savings. • There are, however, important potential risks associated with the EHR, these in the main relating to disruptions to work flow, professional frustration and dissatisfaction, data security considerations; there is in addition, the concern that clinically important information may be overlooked, particularly in contexts where there is parallel recording of data using both electronic and paper records. 22 • The empirically demonstrated benefits relating to introduction of EHRs are currently limited to improved legibility, time savings for some professionals (nurses), and the facilitation of higher order functions such as audit, secondary analysis of routine data and performance management. • Time taken for doctors to enter and retrieve data has in contrast been found to increase; studies have furthermore found that the time disadvantage for clinicians to record and retrieve information did not attenuate with increased familiarity and experience with using EHRs. • There is moderate evidence that these can help improve patient outcomes, particularly in relation to provision of preventive care. • Standardised and widely accepted measures of data quality in EHRs are lacking and their development should be a priority. • Important potential national future development for EHRs include better human-machine interfaces which allow, for example, the use of voice recognition to speed up data input and the ability to incorporate multimedia files such as heart sounds, retinal screens and audio- or videorecordings of consultations. Computer-assisted history taking systems • Most computer-assisted history taking systems (CAHTS) are designed for use by healthcare professionals, although some elicit information directly from the patient, as in the case of pre-consultation interviews. • Computer-assisted history taking systems can be used in a variety of clinical settings and have, when eliciting data directly from patients, proven particularly useful in identifying potentially sensitive information such as alcohol consumption, sexual health and psychiatric illnesses (e.g. suicidal thoughts). • Computer-based questionnaires are particularly useful for gathering important background data prior to the consultation, which can then allow more time for focusing on key aspects of the health problems in the actual consultation. These systems may also save money by reducing administrative costs. 23 • Speech software and speech completed response CAHTS allow adaptability for those with particular needs such as non-English speaking patients, patients with hearing impediments and those who are illiterate. • These can also prove helpful in the context of delivering high volume care in specific contexts such as the screening of patients in the context of pandemic illness (e.g. H1N1-pandemic influenza). • There is moderate evidence that data collected electronically tend to be more accurate and contain fewer errors than data captured manually with traditional pen and paper techniques; such data are also more legible. • The current generation of computers is however not adept at detecting non-verbal behaviour; these systems should therefore be seen as not a substitute, but rather an adjunct to the clinical history. • There have as yet been no comparative studies that have formally assessed the effectiveness and cost-effectiveness of different CAHTSs. • It is important for the DH’s Informatics Directorate to carefully consider the considerable potential efficiency gains to be made from incorporating CAHTS functionality—particularly if this involves direct entry of data by patients—into future iterations of the informatics strategy. HealthSpace could facilitate this as could a number of other modalities such as touch-screen or voice-recognition equipped computers available in waiting rooms. This will, however, need to be introduced within a clear evaluative context. Supporting professional decision-making Computerised decision support systems • There are strong theoretical reasons for believing that improved access to relevant clinical information for healthcare professionals, at the point of care, can translate into improvements in healthcare quality, patient safety and organisational efficiency. 24 • Defined as software applications that use individual patient data, computerised decision support systems (CDSSs) utilise a repository of clinical information (knowledge-base) and an inference mechanism (logic) to generate patient-specific output. These applications are highly variable in sophistication, output and the extent to which they can integrate with other clinical information systems. • Computerised decision support systems have the potential to improve clinical decision-making by providing practitioners with real time patient-specific and evidence-based support and by providing individually tailored feedback. • Although numerous evaluations of CDSSs have taken place, very little consistent and generalisable evidence exists on their ability to improve patient outcomes; the available evidence suggests however that practitioner performance can be improved, particularly in relation to relatively simple tasks such as aspects of preventive care (e.g. vaccination reminders) and monitoring of those with long-term conditions (e.g. blood pressure monitoring in people with hypertension). • The use of computerised reminders for provision of preventive care has been empirically demonstrated to be of benefit. However, trials have tended not to focus on patient outcomes; this may be because for most preventive care interventions, the numbers needed to be studied and/or and the time needed to demonstrate an effect on patient outcomes are prohibitively large/long. • Commercial CDSSs are largely unregulated in the US and UK due to exclusion from the Federal Drug Administration and Medicine and Healthcare Products Regulatory Agency respectively. • Without formal quality and safety assurances in relation to CDSS applications, the potential risks to patient safety need to be seriously considered as they may in certain situations inadvertently introduce new errors. • As CDSSs work best when interfacing with or integrated within existing clinical information systems, and the evidence of benefit is clearest and 25 risk of harm is least for provision of preventive healthcare, the DH should consider introducing a range of computerised health promotion tools into primary care and with the roll-out of the EHR into secondary care. • Research should focus on understanding the contexts in which these decision support tools are most likely to prove effective and this should be a priority consideration for the DH as it introduces new eHealth applications with built-in decision support functionality. ePrescribing • There is considerable variation in the quality of prescribing. Medicines management errors are common, costly and an important source of iatrogenic harm. • ePrescribing can be defined as the use of computing devices to enter, modify, review and output or communicate prescriptions. ePrescribing systems are highly variable in genealogy, functionality, configurability and the extent to which they integrate with other systems. • ePrescribing has the potential to greatly improve the quality and safety of prescribing, through facilitating cost-conscious evidence-based prescribing and in particular reducing errors associated with knowledge gaps and routine tasks such as repeat prescribing. • There is moderate evidence that practitioner performance is improved through better access to these guidelines. Patient outcomes are, however, less well studied and, when assessed, most studies have not been able to demonstrate a clinical benefit. This is likely to be due to the prohibitive numbers needed to demonstrate a clear impact on infrequent, but serious outcomes. • Evidence of benefit from ePrescribing systems has in the main been demonstrated from evaluations of “home-grown” systems in a few centres of excellence. Most systems in use are however commercially procured and these systems typically lack the sophistication, clinical relevance and sense of ownership associated with the tailored homegrown systems. 26 • Poorly designed ePrescribing systems and a failure to appreciate the socio-techno-cultural issues associated with their introduction can introduce unexpected new risks to patient safety. • In the UK, the Medicines and Healthcare Products Regulatory Agency does not consider ePrescribing systems to be a medical device and does not therefore require these systems to be quality assured. This is an important policy failing that needs to be addressed. • Further research into the design features, knowledge-bases and underlying algorithms, clinical relevance of output, interoperability of ePrescribing systems and socio-technical factors that enhance use is needed in order to replicate the benefits of ePrescribing that have been demonstrated in US centres of excellence. This research needs to focus on commercial systems. • There is merit in considering the introduction of ePrescribing systems into NHS hospitals, but in an evaluative context. Remote delivery of care Telehealthcare • Current systems of healthcare are inadequate to deal with rapidly increasing populations, many of whom have one or more long-term conditions. • Telehealthcare can be defined as “personalised healthcare delivered over a distance.” It thus involves transfer of data from the patient to a professional using IT, which is then followed by personalised feedback from the professional to the patient. • It has the potential to improve access to care, promote preventive care, and help manage the health and social care needs of, amongst others, those with long-term conditions. • Whilst potentially attractive, these new models of care can however prove very costly and may also cause a considerable change to consultation dynamics and workflow processes; these should therefore be introduced only after a careful assessment of effectiveness, costeffectiveness and safety considerations. 27 • Most patients have in general a positive view of telehealthcare, so long as it is offered in addition to, rather than as a substitution for, the faceto-face consultation. Qualitative studies have highlighted that the needs of the carer should also be taken in account when designing telehealthcare systems. • Overall, there is for people with long-term conditions such as asthma, chronic obstructive pulmonary disease and diabetes now moderate evidence that telehealthcare can reduce hospital admissions in those with more severe disease without increasing mortality; the evidence in relation to improving quality of life in those with milder disease is however less encouraging. • There is a need for more cost-effectiveness studies which take into account the full range of perspectives, i.e. that of the patient, healthcare services and society. • Important potential pitfalls of telehealthcare that need to be considered include user interface issues such as older people not understanding how to operate systems, technical breakdown, and safety concerns such as data loss and breaches of confidentiality. • The use of telehealthcare often results in a reshaping of the doctorpatient relationship and attention should be paid to the opportunities for humanising this interaction. Workflows also need to be considered to minimise the risk of unintended disruptions to normal working routines. Socio-technical dimensions of designing, developing, and deploying eHealth applications Human factors • The aim of human factors (also known as ergonomics) is to bring to bear an understanding of how people’s conduct is shaped by their technological, organisational and social environments to processes of design and innovation so that new technologies are compatible with their users’ goals, capabilities and needs. • When new technologies are introduced into the workplace, success will typlically depend on addressing a diverse range of human factors 28 issues, including: the nature of existing work practices and performance requirements (including efficiency, dependability and safety); user skills; possibilities for the re-design of work (including user perspectives and preferences); and training requirements (including formal, informal and on-the-job). • The healthcare sector has been slow to incorporate human factors into the design and evaluation of eHealth applications, despite the mounting complexity of care delivery systems and evidence of resulting risk to patients from poorly designed systems. • A well-designed user interface is, for example, as important as functionality and reliability in applications such as ePrescribing. Confusion and frustration with an application interface are enough to impede users’ acceptance, with an adverse subsequent knock-on effect on implementation. User-centred design and evaluation for usability are critical for acceptance. • There is a lack of consensus in the literature regarding how best to incorporate human factors into the design, development and evaluation of eHealth applications..The DH needs to establish best practice guidelines for human factors in eHealth applications, including standards for usability evaluation, and ensure that a human factors audit and usability findings are available to inform assessment of new eHealth applications prior to approval for use in the NHS. • The EHR, ePrescribing and CDSS systems are examples of complex applications for which it is clearly essential that user needs have been adequately considered in all stages of design, development and deployment. • Critical feedback from users of new eHealth applications should not only be facilitated, but must also be actively encouraged so as to ensure that new applications are fit-for-purpose and minimise risks to patient safety. • Embedding human factors principles and thinking is resource intensive; the DH needs to ensure that adequate time, resources and 29 prioritisation are given to this so as to maximise the chances of success of its various eHealth initiatives. Effective implementation and adoption of eHealth applications • Many technological innovations fail to realise their potential and this unfortunately has also been true with respect to the history of eHealth applications. • Major factors contributing to these failures—which may in some cases be spectacular—include the lack of appreciation and attention paid to human factors’ issues during product development and socio-technical factors that subsequently enable innovations to integrate with and embed themselves into healthcare organisations. • In planning IT deployments, health systems need to be clear as to whether the aim is to replicate existing practices or to use the intervention as a basis for organisational reform. • There is a burgeoning change management literature, dating back to the influential Diffusions of Innovation theory and stretching to the more recent Diffusion of Innovations in Health Service Organisations work. Much of this is however descriptive and so the predictive ability of these models of change management is as yet unknown. • Using an adaptation of the above and other theories to render them more relevant to the planned dissemination of eHealth applications, we have developed and updated our Infusion of eHealth Innovations in Health Services Organisations Model to undertake a detailed case study seeking to learn lessons from NHS CFH’s approach to promoting the NHS CRS in secondary care. CONCLUSIONS AND FUTURE RESEARCH PRIORITIES • In undertaking the first phase of work, we made four main methodological contributions to this nascent field of enquiry, namely: o development of a comprehensive search strategy for identifying high quality primary and secondary literature investigating the impact of eHealth on the quality and safety of healthcare 30 o development of integrated conceptual maps of eHealth, quality and safety, which have, as demonstrated in this project, the ability to draw attention to the major potential benefits associated with use of different eHealth applications o adapting a tool for critically appraising systematic reviews of eHealth applications o development of a framework with which to consider the planned implementation of eHealth innovations into complex health service organisations. • The second phase of work has allowed us to initiate and undertake: o a suite of Cochrane systematic reviews on various aspects of eHealth, in particular focusing on telehealthcare o hold a series of national Masterclases, which have enabled academics, policymakers, clinicians and industry representatives to come together and discuss and debate how best to maximise the health gains associated with eHealth. • The formative work for this project and the review of technical reports and a variety of review documents clearly demonstrate that eHealth applications have great potential to improve the quality of healthcare delivery; even more importantly, perhaps, there is considerable potential to improve the safety profile of medicine through elimination of both latent and active errors, and through promoting real-time systems checks. • The major finding from reviewing the empirical evidence—which is of variable quality—however, is that there is still as yet only limited evidence demonstrating that these technologies actually improve patient outcomes. • The reasons for this are multi-faceted; these include: o an assumption that the benefits associated with these applications are obvious, thus resulting in many (and indeed perhaps the majority of) interventions being implemented without due consideration to evaluation 31 o using proxy outcome measures as opposed to those that are actually clinically important o poorly theorised interventions and studies o overestimating likely effect sizes o methodological limitations, particularly the failure to appropriately use cluster designs in studies that are at risk of contamination, poor outcome definition and measurement, and approaching these interventions as the equivalent of simple interventions, which they are typically clearly not. o naïve assumptions that these technologies will be equally effective in all contexts and settings o inappropriately short time-frames to study likely health gains o the failure to involve end-users at a sufficiently early stage in the design and deployment process such that they can actually influence factors that are likely to increase acceptability of the interventions to clinicians o a failure to pay adequate attention to the socio-technical factors that are likely to be important in relation to the diffusion, implementation and adoption of these applications. • Despite these substantial gaps in the evidence-base—many of which can be filled—we are, on the basis of the theoretical work and empirical evidence reviewed, cautiously optimistic that a number of the eHealth applications that have been introduced into the NHS are likely to result in significant medium- to long-term benefits to organisation efficiency, professional practice and in time patient care. • We would in particular encourage the DH to prioritise the implementation of ePrescribing into hospitals, ideally with CDSS functionality in an integrated way within the NHS CRS within secondary and tertiary care settings, and also to improve ePrescribing functionality in primary care. • There is also a need to ensure that adequate expertise exists to ensure that the considerable body of data now being generated–whether 32 through routine healthcare data sources and/or the increasing array of telehealthcare initiatives–are appropriately interrogated and exploited. • Realising the benefits is however likely to be crucially dependant on menaingfully facilitating end-user input throughout the commissioning, design, development and implementation process as this will maximise the chances that aspects of these technologies that are actually clinically useful are developed; this will also promote a sense of ownership and buy-in into the technological innovation and any accompanying service redesign. • Appreciating the structural, organisational and, to a lesser extent, professional challenges that need to be overcome in the deployment should not, however, be transformative overlooked, particularly when complex technologies such as the NHS CRS are introduced. The need for training in the use of new technologies and on-the-job support also needs much greater emphasis than tends to be the case. • End-user consultation and feedback needs to be viewed as an ongoing process and should therefore continue after deployment to ensure that problems are identified early, as are possible solutions, which can be incorporated into system upgrades. • We have throughout this report identified a number of areas in relation to specific technologies where further research is warranted. There are however a number of broader research considerations that need to be prioritised, which we emphasise here, namely: o the need for further conceptual development and then international consensus building on use of standardised terminologies to facilitate future primary and secondary work o the development of a methodological toolkit to facilitate evaluation of eHealth applications throughout all aspects of the development and deployment lifecycle of these technologies. o the need to encourage researchers to, where appropriate, combine rigorous quantitative and simultaneously conducted qualitative assessments when evaluating the effectiveness of new eHealth applications to allow a detailed appreciation of 33 relevant contextual factors that might help better understand the reasons for the success or failure of the intervention to emerge and also, where found to be successful, to allow an assessment of the likely generalisability of the intervention to be made o for consensus building on the contexts in which proxy measures are appropriate surrogates for clinically important end-points. • Given the considerable gulf between the potential advantages associated with eHealth applications and the actual empiricallydemonstrable benefits we make the following recommendations: o our major policy recommendation is the need for more considered and cautious claims of the likely benefits and possible risks associated with the introduction of new technology into healthcare settings, and the need to be cognisant of the fact that technical, human and organisational factors are all likely to be important success factors, particularly in the context of implementing disruptive innovations o our major research recommendation is to ensure that independent evaluation into the risks and benefits of eHealth continues: given the scale of the investment taking place, it is vital that every opportunity is taken to ensure that this public money produces the desired outcomes. • In conclusion, reviewing this large body of evidence has demonstrated that eHealth is an area of considerable policy interest internationally with considerable potential to impact on care delivery. But despite emerging evidence of positive impacts on aspects of professional performance, there is as yet little clear evidence that these developments are translating into improvements in patient outcomes. Demonstrable benefits from many of the IT-related initiatives ushered in through the NPfIT are, we suspect, most likely to be seen from the reuse of digital data for the purposes of, amongst other things, policy planning, local audit and governance activities, patient empowerment and research. Recent IT developments are likely therefore, over time, to help consolidate the UK’s acknowledged expertise and leadership in 34 the secondary uses of data, which aligns closely with the DH’s new policy priority of initiating an “information revolution”. SECTION 1 BACKGROUND 35 Chapter 1 Introduction Summary • Increasing life expectancy, improved survival of people with acute and long-term conditions and a greater array of available treatment options are placing an increasing burden on health systems internationally. • There is now a substantial body of research, both domestic and international, identifying considerable shortfalls in the current provision of healthcare. • Key issues emerging from this literature are substantial variations in the quality of healthcare and the considerable risks of iatrogenic harm. • These failings make a significant contribution to the high rates of potentially avoidable morbidity and mortality, and rising healthcare expenditure. • There have been substantial developments in information technology hardware and software capabilities over recent decades and there is now considerable potential to apply these technological developments in relation to aspects of healthcare provision. • Of particular international interest is the deployment of eHealth applications—that is the use of information technology in healthcare contexts—with a view to improving the quality, safety and efficiency of healthcare. • Whilst these eHealth technologies have considerable potential to aid professionals in delivering healthcare, the use of these new technologies may also introduce significant new unanticipated risks to patients. • Also of concern is that even when high quality interventions are developed, they frequently fail to live up to their potential when deployed in the “real world”; the reasons underpinning these failures remain poorly understood, but include amongst other things, professional resistance to change, poorly designed applications and the resulting disruption to clinical practices and care provision. 36 Summary contd… • Given that the NHS has embarked on one of the largest eHealth-based modernisation programmes in the world, it is appropriate and timely to critically review the international eHealth literature with a view to identifying lessons that can be learned with respect to the future development, design, deployment and evaluation of eHealth applications. 1.1 Background to this study and policy context The Institute of Medicine has defined quality as the ‘…degree to which health services for individuals and populations increase the likelihood of desired health outcomes and are consistent with current professional knowledge.’1 The quality of services provided by the National Health Service (NHS) in England varies widely, and there is often a large gap between the optimal standard of services, when judged based on professionally agreed criteria, and the actual quality of practice.2 3 This quality gap can have serious health consequences and major implications for the intimately inter-related notion of patient safety.4 These deficiencies can manifest, for example, as misdiagnosis,5 medication errors,6 7 increased rates of complications in patients with long-term diseases,8 hospital admissions for adverse drug reactions,9 10 and harm in already hospitalised patients.11 This gap between optimal and actual levels of healthcare also has large financial costs for the healthcare system, national governments, and society, as well as affecting patients' quality of life and life expectancy.10 12 13 Problems with the quality of care can have a considerable impact on healthcare costs; for example, through increased hospital admission rates for unscheduled care and/or prolonged hospital stay.10 Problems with the safety of healthcare can similarly have major cost implications; for example, adverse drug reactions in the UK are estimated to result in approximately 250,000 hospital admissions 37 per year9 10 at a total cost of about £500 million (€700 million; $1,008 million).9 10 Mounting evidence of the disease burden associated with variations in standards of care, especially with regard to medical errors, has informed a number of recent national policy and strategic reports on patient safety.14-23 These documents have highlighted the pressing need for healthcare policy to focus more clearly on developing systems—and strategies for their implementation—that facilitate the delivery of safe and effective high quality healthcare. This response being catalysed by the identification of a series of failings, both with respect to policy and approaches to strategically implement these healthcare reforms.24 There are also other relevant factors, including: the fact that economicallydeveloped (and indeed also transition and developing) countries are facing increases in the proportions and number of older people in their populations; increasing public expectations about rapid access to high quality patientcentred health services; and an expectation from governments, professionals and the public alike that people should be able to live with the minimum possible impact from disease and disability on their quality of life and day-today activities.15 Capitalising on the information technology (IT) revolution is increasingly seen as pivotal to redesigning healthcare systems so that they are able to deliver safe, effective and convenient healthcare. Such objectives lay at the heart of the NHS’s Information Strategy and led to the creation of structures such as the National Patient Safety Agency (NPSA),25 the National Reporting and Learning System (NRLS),26 NHS Connecting for Health (NHS CFH) and its National Programme for Information Technology (NPfIT) and the more recent creation of the Department of Health’s (DH) Informatics Directorate (see Chapter 3 for a detailed discussion of NHS CFH, NPfIT and the DH Informatics Directorate). 38 1.2 Agenda for system redesign: quality and safety in healthcare The Department of Health (DH) estimates that 10 per cent of patients admitted to NHS hospitals are unintentionally harmed, this high rate being similar to that in other developed countries. If lessons had been learnt from previous incidents, around 50 per cent of these patient safety incidents could have been avoided.17 In its report An Organisation with a Memory, an expert committee, chaired by the then Chief Medical Officer Sir Liam Donaldson, identified four conditions that the NHS must meet to learn effectively from failures and offer the best possible protection to patients in the future.14 These relate to the development of: • a more open culture, in which clinical errors or failures of service delivery can be reported and discussed by NHS staff • unified mechanisms for reporting and analysing patient safety incidents in healthcare delivery • mechanisms to ensure that, where lessons are identified, the necessary changes to improve the delivery of healthcare are put into practice • a much wider appreciation of the value of system-based approaches in preventing, analysing and learning from errors. These requirements called for complex and comprehensive solutions as no single “magic bullet” will be sufficient.15 As an example of the role of IT in these endeavours, evidence-based technological solutions may provide significant support for the development of standardised incident reporting and analysis mechanisms, as well as providing a platform for automated interventions to ensure that lessons learned from past mistakes are incorporated into daily practice.26 Such interventions also demonstrate the system-based approach called for in a number of policy documents. 1.3 Importance of technology in designing safe healthcare systems Enthusiasm about the potential that IT offers for improving health services has resulted in unprecedented investments into IT by the NHS (and other health systems internationally). Technology-based health services and clinical 39 systems come in many different forms, have myriad aims, and can be implemented in numerous ways. Information technologies (sometimes also referred to as information and communication technology (ICT)), health informatics or eHealth applications, the latter being our preferred term in this report) may be used to record, collate and share information on patients (e.g. electronic health records (EHRs)). eHealth applications may also be used to support evidence-based practice both in general terms (e.g. guideline-linked reminders)27 or more specifically individual patients (e.g. through advice on the management of advice on avoiding drug-drug interactions).28 29 eHealth applications can also serve to facilitate care from a distance30 (e.g. telehealthcare), patient self-care (e.g. through consumer informatics applications), epidemiological and quasi-experimental research (e.g. using databases that collate data derived from electronic health records)31 and healthcare management activities such as quality improvement initiatives.32 33 As discussed in subsequent chapters, many of these technology-based approaches are vigorously being pursued both in the NHS and internationally.34 35 eHealth solutions range from computer-assisted history taking systems (CAHTS),36-42 computerised ePrescribing,44-52 decision computerised support medication systems administration (CDSSs),43 records,53 automated pharmacy systems and bar-coding to point-of-care personal digital assistants, robot-assisted surgery, telehealthcare and “smart” homes.54-60 Because of the often complex nature of these systems, there are important considerations relating to appropriately evaluating their impact and we reflect on these issues in more detail in Chapters 2 and 20. The potential offered by these eHealth applications is considered in Chapter 4 and the empirical evidence in relation to a number of key exemplary eHealth applications is summarised in Chapters 6–18. The desire to realise the potential of technology to improve the quality and safety of healthcare often leads to the introduction of new IT solutions, on a large or even national scale, with typically only limited evidence in support of the systems’ overall effectiveness and safety. Of particular relevance here is 40 that the hardware-software development mismatch often creates unrealistic expectations about the likely impact of eHealth systems. Processing power doubles approximately every 18 months and storage capacity and network bandwidth also increase rapidly within relatively short time periods. In contrast, productivity in software development has been increasing much more slowly. The Chaos studies performed by the Standish Group have, for example, shown that the majority of software projects (across all industries) are delivered late and over budget.61 Specifically, software applications often become a bottleneck for healthcare organisations trying rapidly to improve their processes, this despite substantial investments in IT.62 To meet the expectations of the market, systems’ developers have thus far been more concerned with meeting design briefs and satisfaction than with assessing the efficacy and effectiveness of systems in terms of producing actual health gains. Consequently, many eHealth innovations remain either unstudied or under-studied. Indeed, when evaluated, the evidence on the question of their effectiveness is often inconsistent, frequently reflecting, as we will see, a lack of attention to issues pertaining to context, which is so crucial to success.63-68 Researchers have paid even less attention to patient safety (i.e. the potential for harmful effects), as demonstrated by recent work in the UK evaluating the safety profile of general practice (GP) computing software in providing decision support for prescribing.45 69 Other areas of relevance in the context of both clinical effectiveness and safety are user acceptance, interoperability, workflow integration, compatibility with legacy applications, and system maturity.70 71 Emerging eHealth systems have the potential to reduce errors and enhance patient safety by, amongst other things, improving the legibility of clinical communications, enabling shared access to health records, improving access to care, reducing reliance on human memory and prompting evidence-based prescribing.33 72-75 For example, automated monitoring and routine feedback have been shown to reduce the rate at which hospitalised adults with renal insufficiency received excessive doses of medication.76 41 However, such systems also have the potential to introduce risks and compromise quality as demonstrated in Chapter 4, where we map the key health risks associated with eHealth solutions. For example, a study examining the types of errors facilitated by a leading ePrescribing system found that it regularly increased the risk of a variety of errors, with users describing up to 22 different types of failures.77 Similarly, high rates of adverse drug events have been found in a highly computerised hospital.78 Although improved case finding (in part resulting from IT use) may partially explain these high rates, the overall rate of adverse drug events (related to problems in drug selection, dosage, and monitoring) per 1,000 patient-days was 5 to 19 times higher than that reported in previous studies. These examples illustrate the point that, as new IT solutions are conceived and implemented, there is also a need to develop mechanisms proactively to study the potential for these systems to inadvertently cause harm.79-84 As with many innovative ventures, not all ideas ultimately prove beneficial. Even when safe, some 50 to 70 per cent of eHealth projects fail.85 These failures do not in the majority of cases represent technological problems, but rather human and organisational factors. Problems may arise due to developers misjudging the functioning of health systems or not envisaging the full range of consequences. Even when the ideas prove effective in controlled research settings, they often challenge prevailing thinking and behaviour in a way that renders them difficult to implement.86 87 There is ample evidence to show that the past failures of technological innovations with respect to improving health outcomes (or to drive improvements in other industries)88-90 have not necessarily been due to their clinical ineffectiveness, but rather to social, technological and cultural issues relating to their implementation and adoption.91 In particular, the human elements affecting the success of technological implementation and adoption should not be underestimated, whether these relate to organisational issues, e.g. managing complexity92— complex systems involve many gaps between people, stages, and processes;93 and or availability, accessibility and user attitudes towards, or usability of the technology.70 87 94-100 Furthermore, integration of IT systems into clinical settings fundamentally may change not only how clinicians view 42 their daily work practices, but also the very process of medical reasoning itself, introducing new elements into care pathways, often with unpredictable effects.101 102 Any significant adverse events due to a failure of new eHealth solutions have the potential to lead to disastrous effects in the way in which these programmes are perceived by professionals and the public alike. It is therefore vital to ensure that implementation and adoption considerations are given at least equal importance as questions of IT effectiveness and safety. Successfully integrating new IT solutions into the healthcare workflow is crucially dependant on engagement of clinicians and patients right from the start of the development and ongoing evaluation of these new applications. In addition to providing important insights into user requirements, such engagement leads to a sense of ownership and buy-in that is critical to the success of the often radical programmes to transform health services that are currently underway in different parts of the world.87 103 These socio-techno- cultural considerations are reviewed in Chapters 16-18. In summary, key challenges currently facing the DH’s/NHS CFH’s ambitious NPfIT and the NHS quality and patient safety agendas include ensuring that new eHealth solutions developed for and introduced into the NHS represent a genuine clinical advantage, that their deployment is, wherever possible, supported by a secure evidence-base, and that evaluative mechanisms are in place to assess their effects with regard to both quality and safety of health service delivery. As a recent Institute of Medicine report highlighted, patient safety must be indistinguishable from the delivery of quality care23 and must score top on our agendas.104 1.4 Aims of this review This report, which builds on our scoping work,33 completed and ongoing systematic reviews,105-110 both quantitative and qualitative studies on eHealth approaches to enhancing quality and minimising threats to patient safety, aims to provide a systematic overview of the impact of eHealth on the safety and quality of healthcare. In so doing, we seek to identify, critically appraise 43 and then synthesise evidence to develop a theoretically informed framework for understanding the potential and empirically demonstrated benefits and risks associated with eHealth and to describe and understand approaches that can be used to guide the successful implementation and adoption of effective interventions. We also aim to identify areas of eHealth in need of further study, highlighting possible approaches that might successfully be employed to ensure a well-rounded, context-bound appreciation of the overall impact of IT interventions in healthcare. 1.5 Structure of this report After this introductory chapter, Section 2 begins with a detailed description of our methods (Chapter 2) and this is followed by the findings of our attempts at understanding the history, development and policy foci of the DH/NHS CFH and NPfIT (Chapter 3) and describing and mapping the fields of eHealth, quality and safety (Chapter 4). In Section 3 we present our main findings, beginning with the essential underpinning issues of health information exchange capability and interoperability (Chapter 5), and then moving on to reviews of key eHealth applications, namely EHRs (Chapter 6), CAHTS (Chapter 7), CDSSs (Chapter 8), ePrescribing (Chapter 10) and telehealthcare (Chapter 13). We then turn to the cross-cutting issues of how to design safe, effective and usable systems (Chapter 16) and how then to ensure their effective implementation and adoption into routine models of care (Chapter 17). In each of these chapters we begin by considering key definitional issues and the relevant theoretical literature with a view to having a firm foundation from which to interpret the empirical evidence; the chapters conclude with a reflection of key policy, practice and research implications arising from this work. We then reflect on this evidence in considerably more detail in the context of applying it to five exemplary case studies looking at important clinical and policy questions in the remaining chapters in this section (Chapters 9, 11, 12, 14, 15 and 18). 44 Our conclusions and policy recommendations from this review are presented in Chapter 19. Section 4 concludes with our reflections on the quality of the evidence reviewed and overarching suggestions for future research which should, if pursued, help to improve both the quality and quantity of empirical evidence for eHealth applications and their successful deployment (Chapter 20). The final section contains a number of appendices relating to the methods employed (Appendices 1-3), detailed findings (Appendices 4 and 5) and outputs from this work (Appendices 6 and 7). This section also includes a glossary of key terms. 1.6 Conclusions As a result of the work carried out for this report, we have, we believe, made a number of key conceptual and methodological advances in relation to techniques for identifying, appraising, synthesising and interpreting the evidence. Furthermore, our work has laid the foundation for the creation of a comprehensive, critically appraised, and carefully interpreted database of systematic reviews, which should, once fully developed, be of international interest to all those with an interest in eHealth and its impact on healthcare delivery. We hope that the theoretical and empirical evidence presented in this report, when viewed in the context of ongoing policy developments, provides a reliable and accessible overview of a vast corpus of hitherto disparate knowledge and through so doing helps to inform on-going deliberations on how to ensure that eHealth realises its enormous potential to impact positively on both the quality and safety of healthcare delivery. References 1. Institute of Medicine. Crossing the Quality Chasm: A New Health System for the 21st Century. Washington, DC: National Academy Press, 2001. 2. Klein R. Britain's National Health Service revisited. N Engl J Med 2004;350:937-942. 3. Martin RM, Sterne JAC, Gunnell D, Ebrahim S, vey Smith G, Frankel S. NHS waiting lists and evidence of national or local failure: analysis of health service data. BMJ 2003; 326: 188. 45 4. Institute of Medicine. To Err Is Human: Building a Safer Health System. . Washington, DC: National Academy Press, 2000. 5. Chadwick D, Smith D. The misdiagnosis of epilepsy. BMJ 2002; 324: 495-96. 6. Wong IC, Ghaleb MA, Franklin BD, Barber N. Incidence and nature of dosing errors in paediatric medications: a systematic review. Drug Saf 2004; 27:66170. 7. Snijders C, van Lingen RA, Molendijk A, Fetter WPF. Incidents and errors in neonatal intensive care: a review of the literature. Arch Dis Child Fetal Neonatal Ed 2007; 92: F391-98. 8. Alexander KP, Chen AY, Roe MT, Newby LK, Gibson CM, len-LaPointe NM et al. Excess dosing of antiplatelet and antithrombin agents in the treatment of non-ST-segment elevation acute coronary syndromes. JAMA 2005; 294: 310816. 9. Pirmohamed M, James S, Meakin S, Green C, Scott AK, Walley TJ et al. Adverse drug reactions as cause of admission to hospital: prospective analysis of 18 820 patients. BMJ 2004; 329: 15-19. 10. Davies EC, Green CF, Mottram DR, Rowe PH, Pirmohamed M. Emergency readmissions to hospital due to adverse drug reactions within 1 year of the index admission. Br J Clin Pharmacol 2010; 70: 749-55. 11. Davies EC, Green CF, Taylor S, Williamson PR, Mottram DR, Pirmohamed M. Adverse drug reactions in hospital in-patients: a prospective analysis of 3695 patient-episodes. PLoS One 2009; 4: e4439. 12. Institute of Medicine. Preventing Medication Errors. 2007. 13. Institute of Medicine. Patient Safety: Achieving a New Standard for Care. Washington, DC: National Academy Press, 2003. 14. Department of Health. An Organisation with a Memory: Report of an expert group of learning from adverse events in the NHS. London: HMSO, 2000. 15. Department of Health. Equity and Excellence: Liberating the NHS. London: HMSO, 2010. 16. Department of Health, Chief Pharmaceutical Officer. Building a Safer NHS for Patients: Improving Medication Safety. London: HMSO, 2004. 17. National Audit Office. Department of Health. A Safer Place for Patients: Learning to Improve Patient Safety. London: Stationery Office, 2005. 18. National Audit Office. Safety, Quality, Efficacy: Regulating Medicines in the UK. London: The Stationery Office, 2003. 19. National Audit Office. A Comparison of International Practices in the Management and Control of Hospital-acquired Infections. HC 876. London, National Audit Office, 2003. 20. Department of Health. Safety First: A Report for Patients, Clinicians and Healthcare Managers. London: HMSO, 2006. 21. NPSA. Safety in Doses. London: NRLS, 2007. 22. Department of Health. Saving Lives. London: HMSO, 1999. 23. WHO. World Alliance for Patient Safety: Forward Programme 2008-2009. WHO, 2008. 24. Berwick DM. Not again! BMJ 2001;322:247-248. 25. NHS NPSA. National Patient Safety Agency. NHS National Patient Safety Agency Available from: http://www.npsa.nhs.uk/ (last accessed 1/1/11) 26. NHS NPSA. Reporting Incidents. NHS NPSA Available from: http://www.npsa.nhs.uk/patientsafety/reporting (last accessed 1/1/11). 27. Shiffman RN, Liaw Y, Brandt CA, Corb GJ. Computer-based guideline implementation systems: a systematic review of functionality and effectiveness. J Am Med Inform Assoc 1999; 6: 104-14. 28. Holdsworth MT, Fichtl RE, Raisch DW, Hewryk A, Behta M, Mendez-Rico E et al. Impact of computerized prescriber order entry on the incidence of adverse drug events in pediatric inpatients. Pediatrics 2007;120: 1058-66. 46 29. Kuperman GJ, Bobb AM, Payne T, Avery AJ, Gandhi TK, Burns G et al. Medication-related clinical decision support in computerized provider order entry systems: a review. J Am Med Inform Assoc 2007. 14; 29-40. 30. Hersh WR, Hickam DH, Severance SM, Dana TL, Pyle KK, Helfand M. Diagnosis, access and outcomes: update of a systematic review of telemedicine services. J Telemed Telecare 2006; 12 :S3-31. 31. Pringle M. Using computers in general practice research. Practitioner 1986; 230: 635-39. 32. Pagliari C, Sloan D, Gregor P, Sullivan F, Detmer D, Kahan JP et al. What is eHealth (4): a scoping exercise to map the field. J Med Internet Res 2005; 7: e9. 33. Pagliari C, Sloan D, Gregor P, Sullivan F, Kahan JP, Detmer D et al. Literature Review and Conceptual Map of the Field of eHealth. Final report to the SDO programme. 2004. 34. Department of Health. Health Informatics Review. London: HMSO, 2010. 35. Department of Health. Liberating the NHS: An Information Revolution. London: HMSO, 2010. 36. Lilford RJ, Bingham P, Fawdry R, Setchell M, Chard T. The development of online history-taking systems in antenatal care. Methods Inf Med 1983; 22: 18997. 37. Lilford RJ, Glyn-Evans D, Chard T. The use of a patient-interactive microcomputer system to obtain histories in an infertility and gynecologic endocrinology clinic. Am J Obstet Gynecol 1983; 146: 374-79. 38. Lilford RJ, Bingham P, Bourne GL, Chard T. Computerized histories facilitate patient care in a termination of pregnancy clinic: the use of a small computer to obtain and reproduce patient information. Br J Obstet Gynaecol 1985; 92: 33340. 39. Lilford RJ, Chard T, Bingham P, Carrigan E. Use of a microcomputer network for history taking in a prenatal clinic. Am J Perinatol 1985; 2: 143-47. 40. Lilford RJ. Comparisons between written and computerised patient histories. Br Med J 1987; 295: 503. 41. Lilford RJ, Guthrie K, Kelly M. History-taking by computer. Baillieres Clin Obstet Gynaecol 1990; 4: 723-42. 42. Pringle M. Using computers to take patient histories. BMJ 1988; 297: 697-98. 43. Bates DW, Kuperman GJ, Wang S, Gandhi T, Kittler A, Volk L et al. Ten commandments for effective clinical decision support: making the practice of evidence-based medicine a reality. J Am Med Inform Assoc 2003; 10: 523-30. 44. Avery AJ, Sheikh A, Hurwitz B, Smeaton L, Chen YF, Howard R et al. Safer medicines management in primary care. Br J Gen Pract 2002; 52: S17-22. 45. Avery AJ, Savelyich BS, Sheikh A, Cantrill J, Morris CJ, Fernando B et al. Identifying and establishing consensus on the most important safety features of GP computer systems: e-Delphi study. Inform Prim Care 2005; 13: 3-12. 46. Barber N, Franklin B, Cornford T, Klecun E, Savage I. Safer, Faster, Better? Evaluating Electronic Prescribing. 2006. Available from: http://www.who.int/patientsafety/information_centre/reports/PS019_Barber_Fin al_report.pdf (last accessed 1/1/11). 47. Barber N. Designing information technology to support prescribing decision making. Qual Saf Health Care 2004; 13: 450-54. 48. Barber N, Cornford T, Klecun E. Qualitative evaluation of an electronic prescribing and administration system. Qual Saf Health Care 2007; 16: 271-78. 49. Bates DW, Evans RS, Murff H, Stetson PD, Pizziferri L, Hripcsak G. Policy and the future of adverse event detection using information technology. J Am Med Inform Assoc 2003; 10: 226-28. 47 50. Bates DW, Evans RS, Murff H, Stetson PD, Pizziferri L, Hripcsak G. Detecting adverse events using information technology. J Am Med Inform Assoc 2003; 10: 115-28. 51. Bates DW. Computerized physician order entry and medication errors: finding a balance. J Biomed Inform 2005; 38: 259-61. 52. Bates DW. Preventing medication errors: a summary. Am J Health Syst Pharm 2007; 64: S3-S9. 53. Franklin BD, O'Grady K, Donyai P, Jacklin A, Barber N. The impact of a closedloop electronic prescribing and administration system on prescribing errors, administration errors and staff time: a before-and-after study. Qual Saf Health Care 2007; 16: 279-84. 54. Bates DW, Cohen M, Leape LL, Overhage JM, Shabot MM, Sheridan T. Reducing the frequency of errors in medicine using information technology. J Am Med Inform Assoc 2001; 8: 299-308. 55. Schiff GD, Kim S, Krosnjar N, Wisniewski MF, Bult J, Fogelfeld L et al. Missed hypothyroidism diagnosis uncovered by linking laboratory and pharmacy data. Arch Intern Med 2005; 65: 574-77. 56. Bates DW, Gawande AA. Improving safety with information technology. N Engl J Med 2003; 348: 2526-34. 57. Patterson ES, Cook RI, Render ML. Improving patient safety by identifying side effects from introducing bar coding in medication administration. J Am Med Inform Assoc 2002; 9: 540-53. 58. Kaushal R, Barker KN, Bates DW. How can information technology improve patient safety and reduce medication errors in children's health care? Arch Pediatr Adolesc Med 2001; 155: 1002-07. 59. Garg AX, Adhikari NKJ, McDonald H, Rosas-Arellano MP, Devereaux PJ, Beyene J et al. Effects of computerized clinical decision support systems on practitioner performance and patient outcomes: a systematic review. JAMA 2005; 293: 1223-38. 60. Durieux P. Electronic medical alerts -- so simple, so complex. N Engl J Med 2005; 352: 1034-036. 61. The Standish Group. CHAOS: A recipe for success. http://www.standishgroup.com/ (last accessed 1/1/11). 62. Stead WW, Kelly BJ, Kolodner RM. Achievable steps toward building a national health information infrastructure in the United States. J Am Med Inform Assoc 2005; 12: 113-20. 63. Kaushal R, Bates DW. Information technology and medication safety: what is the benefit? Qual Saf Health Care 2002; 11: 261-65. 64. Wears RL, Berg M. Computer technology and clinical work: still waiting for Godot. JAMA 2005; 293: 1261-63. 65. Koppel R, Metlay JP, Cohen A, Abaluck B, Localio AR, Kimmel SE et al. Role of computerized physician order entry systems in facilitating medication errors. JAMA 2005; 293: 1197-1203. 66. Kuperman GJ, Gibson RF. Computer physician order entry: benefits, costs, and issues. Ann Intern Med 2003; 139: 31-39. 67. Adhikari N. Medical informatics in the intensive care unit: overview of technology assessment. J Crit Care 2003; 18: 41-47. 68. Berg M. Health Information Management: Integrating Information Technology in Health Care Work. London, England: Routledge; 2004. 69. Morris CJ, Savelyich BSP, Avery AJ, Cantrill JA, Sheikh A. Patient safety features of clinical computer systems: questionnaire survey of GP views. Qual Saf Health Care 2005; 14: 164-68. 70. Robertson A, Cresswell K, Takian A, Petrakaki D, Crowe S, Cornford T, et al. Implementation and adoption of nationwide electronic health records in 48 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. secondary care in England: qualitative analysis of interim results from a prospective national evaluation. BMJ 2010; 341: c4564. Fox J, Thomson R. Clinical decision support systems: a discussion of quality, safety and legal liability issues. Proc AMIA Symp 2002; 265-69. Protti D. Comparison of information technology in general practice in 10 countries. Healthc Q 2007;10: 107-16. Sullivan F, Wyatt JC. How decision support tools help define clinical problems. BMJ 2005; 331: 831-33. Sullivan F, Wyatt JC. How informatics tools help deal with patients' problems. BMJ 2005; 331: 955-57. Sullivan F, Wyatt JC. How computers help make efficient use of consultations. BMJ 2005; 331: 1010-12. Nash IS, Rojas M, Hebert P, Marrone SR, Colgan C, Fisher LA et al. Reducing excessive medication administration in hospitalized adults with renal dysfunction. Am J Med Qual 2005; 20: 64-69. Koppel R, Metlay JP, Cohen A, Abaluck B, Localio AR, Kimmel SE et al. Role of computerized physician order entry systems in facilitating medication errors. JAMA 2005; 293: 1197-1203. Nebeker JR, Hoffman JM, Weir CR, Bennett CL, Hurdle JF. High rates of adverse drug events in a highly computerized hospital. Arch Intern Med 2005; 165: 1111-16. Strom BL, Schinnar R, Aberra F, Bilker W, Hennessy S, Leonard CE, et al. Unintended effects of a computerized physician order entry nearly hard-stop alert to prevent a drug interaction: a randomized controlled trial. Arch Intern Med 2010; 170: 1578-83. Battles JB, Lilford RJ. Organizing patient safety research to identify risks and hazards. Qual Saf Health Care 2003; 12: 2ii-7. McNutt RA, Abrams R, Aron DC, for the Patient Safety Committee. Patient safety efforts should focus on medical errors. JAMA 2002; 287: 1997-2001. Overhage JM, Tierney WM, Zhou XH, McDonald CJ. A randomized trial of "corollary orders" to prevent errors of omission. J Am Med Inform Assoc 1997; 4: 364-75. Gillam M, Rothenhaus T, Smith V, Kanhouwa M. Information technology principles for management, reporting, and research. Acad Emerg Med 2004; 11: 1155-61. Bates DW, Teich JM, Lee J, Seger D, Kuperman GJ, Ma'Luf N et al. The impact of computerized physician order entry on medication error prevention. J Am Med Inform Assoc 1999; 6: 313-21. Lorenzi NM, Smith JB, Conner SR, Campion TR. The success factor profile for clinical computer innovation. Medinfo 2004; 11: 1077-80. Stryer D, Clancy C. Patients' safety. BMJ 2005; 330: 553-54. Greenhalgh T, Robert G, Bate P, Kyriakidou O, Macfarlane F, Peacock R. How to Spread Good Ideas: A systematic review of the literature on diffusion, dissemination and sustainability of innovations in health service delivery and organisation. Report for the National Co-ordinating Centre for NHS Service Delivery and Organisation R & D. 1-424. London, The National Co-ordinating Centre for NHS Service Delivery and Organisation R & D (NCCSDO), 2004. Neumann PG. Computer-Related Risks. New York, NY: ACM Press; 1995. Eason K. Local sociotechnical system development in the NHS National Programme for Information Technology. J Inform Technol 2007; 22: 257–64. Rochlin GI. Trapped in the Net: The Unanticipated Consequences of Computerization. Princeton, NJ: Princeton University Press; 1997. Tenner E. Why Things Bite Back: Technology and the Revenge of Unintended Consequences. New York, NY: Vintage Books; 1997. 49 92. Plsek PE, Wilson T. Complexity science: Complexity, leadership, and management in healthcare organisations. BMJ 2001; 323: 746-49. 93. Cook RI, Render M, Woods DD. Gaps in the continuity of care and progress on patient safety. BMJ 2000; 320: 791-94. 94. Lorenzi NM, Riley RT, Blyth AJC, Southon G, Dixon BJ. Antecedents of the people and organizational aspects of medical informatics: Review of the Literature. J Am Med Inform Assoc 1997; 4: 79-93. 95. Silver MP, Geis MS, Bateman KA. Improving health care systems performance: a human factors approach. Am J Med Qual 2004; 19: 93-102. 96. Stryer D, Clancy C. Patients' safety. BMJ 2005; 330: 553-54. 97. Ball MJ, Douglas JV. Redefining and improving patient safety. Methods Inf Med 2002; 41: 271-76. 98. Short D, Frischer M, Bashford J. Barriers to the adoption of computerised decision support systems in general practice consultations: a qualitative study of GPs' perspectives. Int J Med Inf 2004; 73: 357-62. 99. Kaplan B, Farzanfar R, Friedman RH. Personal relationships with an intelligent interactive telephone health behavior advisor system: a multimethod study using surveys and ethnographic interviews. Int J Med Inf 2003; 71: 33-41. 100. Catwell L, Sheikh A. Information technology (IT) system users must be allowed to decide on the future direction of major national IT initiatives. But the task of redistributing power equally amongst stakeholders will not be an easy one. Inform Prim Care 2009; 17: 1-4. 101. Patel VL, Kaufman DR. Science and practice: a case for medical informatics as a local science of design. J Am Med Inform Assoc: 5: 489-92. 102. Hunt DL, Haynes RB, Hanna SE, Smith K. Effects of computer-based clinical decision support systems on physician performance and patient outcomes: a systematic review. JAMA 1998; 280: 1339-46. 103. Morrison Z, Robertson A, Cresswell K, Crowe S, Sheikh A. Comparing and contrasting three approaches to nationwide implementations of electronic health records: England, the USA and Australia. J Hlthcare Eng (in press). 104. Wilson T, Pringle M, Sheikh A. Promoting patient safety in primary care. BMJ 2001; 323: 583-84. 105. Wilson T, Sheikh A. Enhancing public safety in primary care. BMJ 2002; 324: 584-87. 106. Car J, Sheikh A. Email consultations in health care: 2--acceptability and safe application. BMJ 2004; 329: 439-42. 107. Car J, Sheikh A. Email consultations in health care: 1--scope and effectiveness. BMJ 2004; 329: 435-38. 108. McLean S, Chandler D, Nurmatov U, Liu J, Pagliari C, Car J, Sheikh A. Telehealthcare for asthma. Cochrane Database Syst Rev 2010; CD007717. 109. Civljak M, Sheikh A, Stead LF, Car J. Internet-based interventions for smoking cessation. Cochrane Database Syst Rev 2010; CD007078. 110. Gagnon MP, Légaré F, Labrecque M, Frémont P, Pluye P, Gagnon J, et al. Interventions for promoting information and communication technologies adoption in healthcare professionals. Cochrane Database Syst Rev 2009; CD006093. 50 SECTION 2 BACKGROUND & FORMATIVE WORK 51 Chapter 2 Methods Summary • Drawing on our own systematic reviews and those conducted by others, we conducted a systematic search and critique of the empirical literature on eHealth applications and their impact on the quality and safety of healthcare delivery; this body of work was then synthesised with relevant theoretical, technical, developmental and policy relevant literature with a view to producing an authoritative and accessible overview of the field. • Whilst we drew on established Cochrane review principles to systematically search for, critique and synthesise the literature, this approach needed to be adapted in several respects in order to produce a meaningful “overview” of the literature. • Searching the literature was complicated by the lack of internationally (or indeed in some cases nationally) agreed terminology relating to eHealth applications, the lack of agreed definitions of quality and safety, and consequently poor indexing of these constructs in databases of published work. • In order to undertake a comprehensive review of the literature we therefore needed to undertake initial developmental work to formulate highly sensitive search strategies. • Using this comprehensive set of search terms, we systematically searched: MEDLINE, EMBASE, The Cochrane Database of Systematic Reviews, Database of Abstracts of Reviews of Effects, The Cochrane Central Register of Controlled Trials, The Cochrane Methodology Register, Health Technology Assessment Database, NHS Economic Evaluation Database over a 13-year period (beginning with 1997 to October 2010) to identify systematic reviews, technical reports and health technology assessments, and randomised controlled trials investigating the impact of eHealth applications on the quality and safety of healthcare. We also searched databases of unpublished work to identify relevant ongoing and unpublished work. 52 Summary continued… • To provide a broader appreciation of the context to this work and furthermore to aid development and interpretation of findings, we built our interpretation of evidence around relevant theoretical literature. We used a more emergent approach to identify relevant background literature in relation to the core concepts underpinning this overview—namely eHealth, quality, safety and the National Programme for Information Technology. This involved drawing on our personal databases of relevant papers, identifying seminal papers and reports, and searching the grey literature. • We integrated the findings from this broader theoretical and empirical evidence-base in relation to each of the domains of enquiry. • The overall body of literature identified was too diverse to make any meaningful quantitative synthesis of the literature desirable, nor was it possible. Rather, we chose to synthesise the literature qualitatively drawing on the relevant preliminary conceptual work to guide this narrative synthesis. • Our assessment of individual reviews is summarised in Appendix 6 using the modified Critical Appraisal Skills Programme criteria; the overall assessment of the volume and strength of evidence in relation to key findings is summarised in the Executive Summary using a modified version of the World Health Organization’s Health Evidence Network system for public health evidence, which grades evidence into three main categories: o strong (consistent, good quality, plentiful or generalisable) o moderate (consistent and good quality) o limited to none (inconsistent or poor quality). 53 2.1 Outline of methods Research in the field of eHealth is exponentially expanding in breadth and increasing in volume thus rendering it impossible for any individual to read, critically appraise and synthesise the state of current knowledge, let alone remain up-to-date in this dynamic area. Systematic reviews have become essential sources of information for policy-makers and practitioners who need to remain abreast of important new developments in their fields of interest. These also serve another important function, namely helping to identify areas where there are important gaps in the literature and where further primary studies are required. Our aim in conducting this systematic overview of the literature was to survey the relevant theoretical, technical and empirical literature to produce a comprehensive summary of the evidence for eHealth applications to improve the quality and safety of healthcare delivery, and identify key issues in relation to the design, development, implementation and adoption of eHealth applications into routine healthcare settings. More specifically, in keeping with our brief we were initially interested in two broad areas of eHealth applications, these relating to technologies concerned with improving the storage and management of patient data and tools that support professional decision-making; the scope of enquiry was extended in the second phase of our work to include tools to support the delivery of healthcare from a distance (see Chapter 4). Given the extensive body of eHealth literature, we considered various approaches to searching, appraising, interpreting and synthesising the evidence. Considering the aim of this report and in order to make our findings accessible to policy-makers, managers and end-users,1 We throughout attempted to generate new insights and pragmatic recommendations suitable for effective policy-making in healthcare. Based on our recent experiences of conducting systematic reviews in the subject area, we considered it important to take a broad approach to identifying potentially relevant work in this systematic overview.2 This 54 approach has the advantage of allowing the strength (i.e. consistency and generalisability) of research findings to be assessed across a wide range of different settings, study populations, and behaviours, thereby reducing the risks of drawing erroneous conclusions. Established systematic review methods were used to identify and critically appraise studies and then abstract and synthesise findings. We explicitly sought to factor into our assessment important contextual factors. Detailed methods for the de novo systematic reviews we conducted in the second phase of work are detailed in the relevant publications (see Appendix 7). 2.1.1 Literature reviews of other sources of evidence Standard systematic review methods (e.g. Cochrane), which focus on identifying comparable studies addressing specific research topics and combining their results to establish general effect sizes, were too narrow for the purposes of meeting the objective of research brief.3-5 While evidence collated from such reviews appropriately carried greater weight in relation to assessing questions of effectiveness, our interpretation of evidence was also informed by other research architectures or designs, primarily randomised controlled trials (RCTs) and other experimental designs at relatively low risk of bias,6 but also case series, instructive case reports, and conceptual papers to inform safety considerations. Systematic reviews of RCTs are, on account of their ability to control for known and unknown confounders, the “gold standard” evidence source in relation to studying the effectiveness of interventions. Whilst RCTs and systematic summaries of these are ideally suited for studying drug treatments, they are typically far less appropriate for studying safety issues, a major problem being the prohibitively large RCTs that are typically needed to study adverse events. A meaningful study of safety considerations must therefore embrace a broader literature than that needed to establish efficacy or effectiveness. Randomised controlled trials are furthermore unable to shed detailed insights on whether systems will be used or indeed how they will be used—factors which greatly influence the effectiveness (as opposed to efficacy) of interventions when implemented in routine practice. As the 55 commissioning brief advised a focus on both quality and safety issues of IT deployment in healthcare services, we therefore needed to develop a novel combination of approaches to knowledge acquisition and evidence appraisal to ensure that we produced a valid summary of the literature. We began evidence synthesis by drawing primarily on high quality evidence from systematic reviews. Where these did not provide adequate answers to the questions of interest, we employed a “snowball” approach to identifying additional potentially relevant literature, starting our additional searches for and examination of evidence from RCTs and, as necessary and feasible, then also considering other study designs (e.g. controlled before-after designs, time-motion, descriptive and qualitative studies), as well as relevant theoretical and technical papers.6 7 Since the field is rapidly developing, a key aim of these additional searches was to ensure that any important primary evidence that has emerged since the included reviews were conducted was not overlooked. Many of the included studies went beyond the usual measures of system performance or doctors’ behaviour by focusing on “fit” of the system with other aspects of professional and organisational life.5 We carefully considered issues relating to the measurement of error.8 Patient safety research initiatives can be considered in three different stages: (1) identification of risks and hazards; (2) design, implementation, and evaluation of patient safety practices; and (3) vigilance to ensure that a patient safety environment and culture is maintained. Clearly, different research methods and approaches are needed at each of the different stages of the continuum. No single method can be universally applied to identify risks and hazards in patient safety. Rather, multiple approaches using combinations of these methods increased identification of risks and hazards in terms of potential injury or harm to patients.9 2.2 Planning phase In terms of developing an overall search strategy, we considered several options, such as employing individual strategies for each of the different 56 eHealth areas of interest (discussed below). However, instead we decided to develop a single all-encompassing search strategy in order to improve overall efficiency by avoiding overlapping multiple retrieval of publications in these closely related areas. We began by building on the findings of our eHealth scoping report to the NHS Service Delivery & Organisation10 and our existing multi-disciplinary programme of work on the impact of eHealth on the quality and safety of health services. Our search strategy for bibliographic databases was developed in accordance with the recommendations for search strategies in the Cochrane Handbook for Systematic Reviews of Interventions11 and encompassed broad, comprehensive Medical Subject Headings (MeSH) and free text terms in order to achieve high sensitivity in our search for possibly eligible studies. Initially, we also planned to employ computerised knowledge mapping methods and analysis of non-standard data sources using free text terms. However, due to the volume of literature from established databases, this step was, in agreement with our External Steering Group, not performed due to project time constraints (the Delphi method for identifying research priorities was also omitted for the same reason). 2.2.1 Rationale for an overview of reviews The success of the evidence-based medicine movement has resulted in a proliferation of systematic reviews of the literature, this in turn necessitating the need for overviews of systematic reviews to allow policy-makers, clinicians and increasingly patients to obtain an accurate overview of the literature in a broad field of medicine. These overviews, also sometimes known as “umbrella” reviews seek to collate and synthesise evidence from multiple systematic reviews on the effectiveness of different approaches to addressing a specific or set of related problems or, in contrast, the effectiveness of related interventions for a range of conditions, into one accessible document. 57 Assessing and comparing the effectiveness of different approaches to dealing with a particular problem is clearly of considerable potential importance to healthcare professionals and patients. Of note is that the same or similar interventions may sometimes be used to treat different conditions and whilst an overview of this evidence is likely to be of only limited interest to clinicians or consumers, such an overview can, however, be of considerable interest to policy-makers, basic scientists and technical developers as well as academics working in related areas. Such overviews of the high quality evidence are now routinely being undertaken by a range of commercial (e.g. Clinical Evidence), professional (e.g. the British Thoracic Society and Scottish Intercollegiate Guideline Network Asthma Guideline), and regulatory bodies (e.g. the National Institute for Health and Clinical Excellence (NICE).12 We sought to draw upon and build on the approaches that have been developed by such organisations. 2.2.2 Detailing the scope and foci of the review and mapping the available evidence Getting the question(s) “right” is critical to the success of any systematic review process.13 In the present case, whilst the review question was clearly formulated by the commissioner of our work, this was very broad. The mapping exercise described below was employed to guide and refine the focus of the review and assess the volume of potentially relevant literature. In the context of reviews of the effectiveness of interventions, there is general agreement that a well-formulated question involves the following key components: (1) participants who are the focus of the interventions (in our case both clinicians as users of eHealth applications and patients as those for whom these applications are being used by clinicians); (2) the interventions; and (3) the outcomes. As our review was also concerned with the factors that influence the design, implementation and use of technological interventions, our question included components related to (4) the context in which the intervention was developed and implemented (see Chapter 1). 58 Shortly after the commissioning of this report, a systematic review was published investigating the impact of IT on the quality, efficiency and costs of healthcare. This was commissioned by the US Agency for Healthcare Research and Quality (AHRQ) and undertaken by Southern California Evidence-based Practice Center, Santa Monica, CA and the RAND Corporation.13 Their searches initially covered the time period (January 1995 to January 2004) and involved systematically searching MEDLINE, the Cochrane Central Register of Controlled Trials, the Cochrane Database of Abstracts of Reviews of Effects, and the Periodical Abstracts Database. These searches yielded a total of 257 eligible studies, which were then synthesised qualitatively (see Box 2.1 for key findings). A recent update of this review has confirmed the overall impressions, with little discernible evidence of progress.14 Box 2.1: Key findings from the Agency for Healthcare Research and Quality’s report • IT has been shown to improve quality by increasing adherence to guidelines, enhancing disease surveillance, and decreasing medication errors. • Much of the evidence on quality improvement relates to primary and secondary preventive care. • The major demonstrated efficiency benefit has been decreased utilisation of healthcare. • Evidence of the impact of these technologies on the professional time is inconclusive. • There are only limited data on the cost-effectiveness of eHealth interventions. • Most of the high quality evidence on eHealth applications comes from four benchmark research institutions. • Little evidence is available on the effectiveness of multi-functional commercially developed systems. • There is scant evidence available on interoperability and consumer health informatics. • A major shortcoming of the evidence is its limited generalisability. Adapted from reference 13, permission requested from the Annals of Internal Medicine 59 This AHRQ review necessitated a need to reconsider the scope of our proposed review as there was little point in simply duplicating this work. This issue was raised with our External Steering Group (comprising of representatives of the funders and independent expert advisors) and based on the advice received, we refocused our efforts to expand and update this review, but also to frame the findings within a relevant theoretical framework to aid interpretation and assessment of the generalisability of findings. As the AHRQ report did not cover the dimension of patient safety or implementation issues, we retained our focus on these issues. The field of eHealth spans a vast and complex range of applications, technologies and issues for implementation. An important component of our formative work has been the development and refinement of conceptual frameworks, as an aid to managing and meaningfully interpreting the relevant literature. A key objective of this exercise has been to triangulate the health informatics topic areas with the particular needs and concerns of the Department of Health’s Informatics Directorate/NHS Connecting for Health (NHS CFH). As noted in the commissioning brief, patient safety considerations are often overlooked in the development and implementation of IT in healthcare. This aspect of our work built on the eHealth model developed by our team (Pagliari C, unpublished work) and involved identifying and synthesising existing frameworks for understanding dimensions of quality and patient safety taxonomies, including that developed by the World Health Organization (WHO),15 to develop a framework to understand the interplay between eHealth applications of particular relevance to the DH/NHS CFH, quality and patient safety. This represented a helpful conceptual framework within which to organise thinking when considering development, deployment and commissioning of IT in healthcare applications. Based on this initial mapping of these fields, we were able to refine the scope of the review. For more details of the considerations that have informed this planning phase, please see Chapters 3 and 4. Whilst mapping the fields of eHealth, quality and safety, it became clear that these areas are still undergoing significant conceptual and technical 60 development. For instance, there are major problems with the indexing of this literature in the major biomedical databases. Identifying the relevant literature therefore necessitated a considerable amount of developmental work to identify relevant MeSH and free text terms. This paid dividends though, in that it enabled us to identify a considerable body of relevant literature that was overlooked by the AHRQ review. Cognisant of the key areas of interest to the DH/NHS CFH’s National Programme for Information Technology (NPfIT), we focused in the initial phase of work on issues that are of particular interest to the Programme. These related to the storing, managing and sharing of data, tools to support professional decision-making and delivering care from a distance, and the cross-cutting issue of how to develop and implement technological interventions so as to maximise the chances that they will be successfully used. Our interest was not only on what works, but also on why things work or conversely, do not work.16 We thus also explored the attributes of effectiveness of interventions and critical features for success of an intervention. In addition, we sought to understand how this varies according to context. 2.3 Search methods for identification of evidence The search strategy and subsequent decisions about inclusion and exclusion of studies needs to be tailored according to the review questions, so it was important that the questions were clearly articulated. Despite having well defined areas (see Section 2.4.3), searching the literature proved highly complex and laborious. This resulted from the inherent immaturity of these disciplines relative to others, the resultant flux that these disciplines are currently in and the poorly developed indexing of core concepts and technologies. Inevitably, however, the comprehensiveness, sensitivity and specificity of any search strategy for complex overviews of this kind have to be influenced by pragmatic considerations such as the time and resources available for the project. 61 While overviews of reviews use different methods from systematic reviews in the sense that they summarise existing reviews rather than find and collate original studies, our search strategy was developed in the same fashion as a search strategy for a systematic review. Our focus was on identifying SRs, whether exclusively of RCTs or also including studies employing other designs. 2.3.1 Time period Although computer applications in medicine have existed since the 1960s the eHealth field, as we know it today, has emerged relatively recently and has been characterised by considerable change over recent years. For this reason, we confined our searches initially to the period 1997–2007, and then extended this to 2010. It was considered likely that information on major historical considerations and seminal studies conducted before this time period would have been identified by SRs appearing in this timeframe, thereby allowing a deeper awareness of the evolution of the field to emerge. Also of relevance was that a large amount of information from 1993 to the start of the time period of interest would have been captured in our earlier report on eHealth.10 2.3.2 Sensitivity and specificity There is a tension between sensitivity and specificity in any search strategy and especially in one that addresses a broad question. A sensitive search strategy identifies all relevant information, whereas one that is specific ensures that only articles of relevance are captured. Sensitivity and specificity are influenced by a number of factors, these include: the time period covered; the search terms used; and the combination in which these are used. There is thus a trade-off between conducting an exhaustive search (with the additional resources that this requires) and undertaking a search that may miss some studies, but is unlikely to alter the overall strength of the evidence or findings.12 Given concern that previous reviews may have missed large amounts of relevant literature, we decided, as agreed with our External Steering Group, to develop a highly sensitive search strategy because, even if 62 it was not possible to review all the relevant material identified in the course of this project, the identification of this literature would nonetheless represent an important resource to the wider academic community. Our experiences and those of others2 reveal that quality improvement research, particularly if related to enhancing safety, is poorly indexed within bibliographic databases; as a result, broad search strategies using free text and allied MeSH headings needed to be used. We built on taxonomies we developed for our recent reviews, these including MeSH headings such as: Equipment Safety; Information Systems (all sets, i.e. Hospital, Clinical Laboratory, Clinical Pharmacy, Operating Room, Ambulatory Care); Medical Informatics; Public Health Informatics; Clinical Informatics; Decision Support; and Decision Aids. Communication, Telehealthcare. Other search terms included, Electronic Clinical Medical Computing, Medical Records Systems and These were then appropriately combined with terms that focused searches on Quality of Healthcare, Safety and Errors. We started our search using MEDLINE, which is considered the best starting point to “get into” the literature.11 An initial list of MeSH terms was compiled by our multi-disciplinary team relevant to the eHealth, quality and patient safety fields. These terms were extracted from: • personal databases of relevant published literature already available from our teams previous work • the AHRQ’s search strategy13 14 • an in-depth manual MeSH tree exploration by two researchers (TB & AB with input from JC, CP and the rest of the team); possible relevant terms were explored with regard to their potential relevance by reading the scope note which describes the MeSH term • the keyword or index terms listings in our Reference Manager 11 (RefMan) database of SRs in the fields of eHealth, quality, patient safety, and organisational and implementation issues. 63 The individual MeSH terms were independently scored by two reviewers according to the contribution these made to the “netting” of the references related to the key objectives of our project, this contribution being assessed using a scoring algorithm (Table 2.1). Consensus was achieved by involving a third researcher and the resultant file of relevant MeSH terms was then circulated for comments and discussion to all other members of the team. Table 2.1: Scoring algorithm for subjective exercise of scoring eHealth Medical Subject Heading terms possibly relevant for our project Score Scope note 1 Stand alone MeSH term (not necessary to combine it with other terms) of high priority for the project 2 MeSH term requires combining with other MeSH or free text terms to result in high priority for the project 3 Combining MeSH term with other MeSH or free text terms still results in Low priority for the project 4 MeSH term falls outside of project scope or remit (to exclude) Whilst searching for relevant literature using MeSH terms we realised that most of the eHealth field is also poorly indexed in bibliographic databases. We therefore considered it important to also devise a comprehensive list of free text terms so as to maximise the sensitivity of our search. In order to create a broad, comprehensive list of free text terms to identify as much as possible of the relevant literature, two researchers extracted potential free text terms from: • our existing database of systematic reviews • online glossaries • webpages relating to patient safety and or eHealth (e.g. NHS CFH, National Patient Safety Agency, AHRQ and the Virginia National Centre for Patient Safety). 64 In addition, members of the team added terms which they considered to be important based on their own previous research experiences. Finally, free text terms were circulated for comments and discussion. 2.3.3 Databases searched We searched the following eight electronic databases: • MEDLINE (from 1 January 1997 to August 2007) • EMBASE (from 1 January 1997 to August 2007) • The Cochrane Database of Systematic Reviews (from 1 January 1997 to August 2007) • Database of Abstracts of Reviews of Effects (from 1 January 1997 to August 2007) • The Cochrane Central Register of Controlled Trials (from 1 January 1997 to August 2007) • The Cochrane Methodology Register (from 1 January 1997 to to August 2007) • Health Technology Assessment Database (from 1 January 1997 to August 2007) • NHS Economic Evaluation (from 1 January 1997 to August 2007). These searches were then extended to cover the period until 31 October 2010. We also searched databases of research in progress or unpublished work: • The National Research Register (http://www.nihr.ac.uk/databases) • ClinicalTrials.gov (http://clinicaltrials.gov) • Current Controlled Trials (http://www.controlled-trials.com). In order to make our search strategy sufficiently sensitive to avoid missing any high quality evidence we used the Boolean operator ‘OR’ between all relevant MeSH and free text terms relating to eHealth. We also did this with relevant quality, safety, organisational and implementation terms. In the final stage, we combined those two searches with Boolean operator ‘AND’ (IT search 65 strategy ‘AND’ patient safety and quality, organisational and implementation issues search strategy). Although this approach inevitably yielded a large number of articles that ultimately proved irrelevant, this formative phase allowed us to develop a highly sensitive search strategy that helped to overcome a key limitation that systematic reviewers have hitherto faced, i.e. suboptimal indexing of papers on eHealth. Appendix 1 details our final search strategy. 2.4 Criteria for considering reviews 2.4.1 Types of studies Systematic reviews, health technology assessments and RCTs were the main study designs of interest. We did, however, also draw on studies using other study designs such as descriptive or qualitative studies to understand broader contextual considerations, particularly in relation to assessing the acceptability of interventions. 2.4.2 Types of participants In keeping with the commissioning brief, we were interested in reviews focused on reports of studies evaluating the impact of IT used by any type of a healthcare professional (e.g. doctor and nurse) or allied professional (e.g. receptionists and administrators). Reviews that reported studies focusing exclusively on the use of IT by patients and or their carers were not eligible for inclusion. 2.4.3 Types of interventions To be eligible for inclusion, studies needed to address eHealth applications. We therefore included studies assessing the following interventions: • use of computers in information exchange • electronic health records • computer-based history taking systems • electronic booking systems and electronic referral systems • computerised decision support systems • artificial intelligence in healthcare (relevant to computerised decision support systems) 66 • computerised reminders in clinical practice • computer-aided detection or diagnosis in medical imaging • decision support for ePrescribing and other orders • telehealthcare • human and organisational factors related to computing in healthcare. In addition, studies needed to address at least one of the following: impact on safety; quality; or organisational, implementation or adoption considerations. Consequently, although studies aiming to describe the trends in the literature informed our work they were excluded as they did not assess impact on the parameters of interest to the report. Although within the broad field of eHealth, we excluded reports that were outside the focus of the commissioning brief, these included studies of: • computer-assisted and any other type of IT enhanced surgery • computer-assisted therapy predominantly directed at or used by patients independently or under supervision of a healthcare professional (e.g. smoking cessation, cognitive behaviour therapy17) • telemedicine (i.e. inter-professional communication)18 • eLearning (the use of electronic technology and media to deliver, support and enhance learning and teaching)19 • consumer health informatics (patient-oriented eHealth) • information literacy20-22 • point-of-care testing without CDSS or with CDSS directed at patients.23 • public health surveillance systems24 • references not having an a priori IT focus or objective, but that reported on IT post retrieval of references.25 26 We subsequently found that many studies employed systematic review methodology to simply describe the past and or current state of the literature for medical or health informatics in general,27 or in particular geographical setting28 or describe eHealth applications for a particular clinical domain without assessing patient, practitioner or organisational impact.29 30 Similarly, 67 we found a number of studies using systematic review methodology to assess study quality,31 32 such as use and validity of outcome measurement,33 validity of economic analysis34 35 or appropriate use of statistical analysis.36 We found that systematic reviews had also been conducted to provide a taxonomic description of a particular eHealth application37-39 such as who is using the application40 or how language regarding a field is being used.41 Some systematic reviews did not provide sufficient information on included studies to determine relevance.42 43 Finally, some highly relevant (non-systematic) literature reviews had to be excluded from data abstraction and critical appraisal due to one or more inclusion criteria not being met.44-54 2.4.4 Types of outcomes The focus of this review were studies that were concerned with assessing the impact of eHealth applications on a wide range of quality of care and or patient safety indicators, as well as factors related to implementation and adoption of eHealth applications. 2.5 Methods of the review Two reviewers independently assessed the potential relevance of all titles and abstracts identified from the electronic searches. Full text copies of all articles judged to be potentially relevant from the titles and abstracts were retrieved. Two reviewers then independently assessed these retrieved articles for inclusion. Consensus on the final list of included studies was reached, with any disagreements about particular reviews being resolved by discussion. Reference lists of included reviews were manually searched. 2.6 Criteria for considering articles for this review In the process of search strategy development we considered a number of different methodological filters exploring the types of studies these yielded and the volume of this literature, e.g. the Cochrane Effective Practice and Organisation of Care Group filter. In the end, we decided that filters produced by SIGN55 gave the best balance between sensitivity and specificity. Using these filters, we also uncovered a number of studies using other designs and these were categorised as described above. Although we did not apply any 68 language restrictions to electronic searches, we did not undertake critical appraisal or data extraction on studies not published in English. Similarly critical appraisal and data extraction were not performed on HTAs. 2.6.1 Selection criteria for systematic reviews In order to select relevant SRs we applied methodological filters for SRs devised by SIGN.55 This approach has been validated and is used in the development of a number of evidence-based guidelines (see Appendix 1 for methodology filters). In order to be considered a SR, the publication needed to: • explicitly state that the publication is a SR in the title or text; or • search the literature in a systematic way in order to answer one or more clear questions or describe a given topic; and • apply inclusion and or exclusion criteria or perform quality assessment 2.7 Quality assessment We reviewed a range of quality assessment tools, but found that these were all less than ideal for our purposes, the specific issues of relevance being the narrow range of methodological approaches being considered and the failure to pay sufficient attention to contextual considerations. We therefore adapted available instruments56 to produce an instrument that was suitable for assessment of SRs of eHealth (see Appendix 2) and this was then used to assess the methodological quality of all included SRs. The quality of all eligible studies was assessed by two independent reviewers. 2.8 Data extraction We developed a customised data extraction form to meet our specific needs (see Appendix 2). Where appropriate, quantitative and qualitative data were extracted from the references identified, however, in view of the breadth of literature to be summarised and the limited timescale of the project, this took the form of descriptions of the results of secondary research, e.g. SRs of the effectiveness or cost-effectiveness of CDSSs, rather than attempts to combine the results of individual studies for statistical analysis. 69 2.9 Synthesis and interpretation of identified quality and safety, introduction, implementation and sustainability issues and features of eHealth systems into conceptual frameworks Multiple factors may combine to influence the success of eHealth applications and for this reason such interventions can be difficult to evaluate.57 While compiling available evidence, we aimed to explore the frontiers of knowledge on quality and safety of eHealth applications by evaluating functional capabilities and characteristics of eHealth solutions according to evidence from studies. Where possible though, we also used information from computer simulation58 and conceptual models59 that assessed the effects that could be expected from each capability in the proposed clinical environment.60-62 This additional consideration of conceptual models was important as we know that reporting of errors in healthcare is poor63 and many may conceivably be envisaged to happen even if not reported. Therefore, throughout the process of collection of evidence we aimed to identify (and build new where non-existent) conceptual frameworks for evaluating eHealth solutions based on their functional capabilities, functional characteristics and simulations.64-66 We commented on the relevance of the evidence-base within the specific context of the NHS67 aiming to incorporate findings, wherever possible, into an appreciation of the skills, knowledge, experience, attitudes and values of people (clinicians, healthcare managers, leaders, and patients); as well as the characteristics of tools (such as adaptiveness), environmental factors, tasks, goals and their inter-relationships.68 Importantly, we reported the evidence that is mounting about features that are independent predictors of improved clinical practice69 70 and predictors of successful implementation.70-75 For example, flexibility in incorporating information from diverse sources and adaptability to varied practice settings are likely to be the quality criteria by which CDSSs are judged in the future.76 We also aimed to consider broader factors, for example, whether external incentives are being used sufficiently in promotion of eHealth innovations.77 70 Our multi-disciplinary group, representing the disciplines of clinical care, informatics, patient safety, epidemiology, social science, psychology, medical education and health policy, sought, in regular Project Steering Group meetings, to engage with the data being generated and to contextualise the findings within broader considerations relating to healthcare systems’ reforms, developing technologies and policy deliberations on ways of encouraging professionals to embrace innovations to improve the quality and safety of primary and secondary medical care. We followed three main steps in conducting a narrative synthesis of these data: (1) developing a preliminary synthesis of the findings of included studies; (2) exploring relationships in the findings; and (3) assessing the robustness of the synthesis produced.7 Depending on the findings from this last step, we would, if necessary, undertake further searches for evidence as described above. We aimed as far as possible for transparency78 in this essentially qualitative approach to seeking “saturation” in finding new evidence or new dimensions to evidence. In evaluating the overall strength of evidence, we considered the World Health Organization’s Health Evidence Network (WHO HEN) system for public health evidence79 (adapted from Greenhalgh et al.).80 This classifies evidence as: • Strong: consistent, good quality, plentiful or generalisable • Moderate: consistent and good quality • Limited to none: inconsistent or of poor quality. Further, we judged key elements defined by the AHRQ in rating the strength of findings on the basis of considerations relating to: • Quality: the aggregate of quality ratings for individual studies, predicated on the extent to which bias was minimised • Quantity: number of studies, sample size or power, and magnitude of effect • Consistency: for any given topic, the extent to which similar findings are reported using similar and different study designs. 71 This classifying of the strength of evidence is presented in the Executive Summary. While covering a vast territory of eHealth, we selected and developed exemplary case studies to develop a far richer, more context specific account of evidence in relation to key technologies and issues that have the potential to be of particular interest to the DH/NHS CFH (Chapters 9, 11, 12 and 18), which involved engaging with the primary evidence in considerably more detail. We have furthermore also conducted Cochrane systematic reviews (see for example, Chapters 14 and 15) and are actively engaged in a number of other related systematic reviews (see Appendix 7), the findings from which have contributed to this overview of the literature. References 1. Sheldon TA. Making evidence synthesis more useful for management and policy-making. J Health Serv Res Policy 2005; 10 Suppl1: 1-5. 2. Grimshaw J, McAuley LM, Bero LA, Grilli R, Oxman AD, Ramsay C et al. Systematic reviews of the effectiveness of quality improvement strategies and programmes. Qual Saf Health Care 2003; 12: 298-303. 3. McAuley L, Pham B, Tugwell P, Moher D. Does the inclusion of grey literature influence estimates of intervention effectiveness reported in meta-analyses? Lancet 2000; 356: 1228-31. 4. Pham B, Platt R, McAuley L, Klassen TP, Moher D. Is there a "best" way to detect and minimize publication bias? an empirical evaluation. Eval Health Prof 2001; 24: 109-25. 5. Kaplan B. Evaluating informatics applications--clinical decision support systems literature review. Int J Med Inform 2001; 64:15-37. 6. The Cochrane Collaboration. Cochrane Effective Practice and Organisation of Care Group, 2010. 7. Shepperd S, Lewin S, Straus S, Clarke M, Eccles MP, Fitzpatrick R, et al. Can we systematically review studies that evaluate complex interventions. PLoS Medicine 2009; 6: e1000086. 8. Lilford RJ, Mohammed MA, Braunholtz D, Hofer TP. The measurement of active errors: methodological issues. Qual Saf Health Care 2003; 12: 8ii-12. 9. Battles JB, Lilford RJ. Organizing patient safety research to identify risks and hazards. Qual Saf Health Care 2003; 12: 2ii-7. 10. Pagliari C, Sloan D, Gregor P, Sullivan F, Kahan JP, Detmer D et al. Literature review and conceptual map of the field of eHealth. Final report to the SDO programme. 2004. 11. Higgins JPT, Green S (editors). Cochrane Handbook for Systematic Reviews of Interventions Version 5.0.2 [updated September 2009]. The Cochrane Collaboration, 2009. Available from: www.cochrane-handbook.org (last accessed 4/1/11). 72 12. National Institute for Clinical Excellence. Guide to the Technology Appraisal Process. London: National Institute for Clinical Excellence, 2004. 13. Chaudhry B, Wang J, Wu S, Maglione M, Mojica W, Roth E et al. Systematic review: impact of health information technology on quality, efficiency, and costs of medical care. Ann Intern Med 2006; 144: 742-52. 14. Goldzweig CL, Towfigh A, Maglione M, Shekelle PG. Costs and benefits of health information technology: new trends from the literature. Health Aff (Millwood) 2009; 28: w282-93. 15. World Health Organization. More than words - International Classification for Patient Safety. Geneva: WHO, 2007. 16. Pawson R, Tilley N. Realistic Evaluation. London: Sage, 1997. 17. Kaltenthaler E, Brazier J, De NE, Tumur I, Ferriter M, Beverley C et al. Computerised cognitive behaviour therapy for depression and anxiety update: a systematic review and economic evaluation. Health Technol Assess 2006; 10: iii,xi-iii,168. 18. Pagliari C, et al. EH1 E-Health Scoping Exercise. Review of the Traditional Research Literature. 19. University of Bristol. eLearning: Using technology to support learning. http://www.bris.ac.uk/medical-school/staff/elearning/ (last accessed 4/1/11). 20. Hersh WR, Hickam DH. How well do physicians use electronic information retrieval systems? A framework for investigation and systematic review. JAMA 1998; 280: 1347-52. 21. Pluye P, Grad RM, Dunikowski LG, Stephenson R. Impact of clinical information-retrieval technology on physicians: a literature review of quantitative, qualitative and mixed methods studies. Int J Med Inform 2005; 74: 745-68. 22. Saranto K, Hovenga EJ. Information literacy-what it is about? Literature review of the concept and the context. Int J Med Inform 2004; 73: 503-13. 23. Fermann GJ, Suyama J. Point of care testing in the emergency department. J Emerg Med 2002; 22: 393-04. 24. German RR. Sensitivity and predictive value positive measurements for public health surveillance systems. Epidemiol 2000; 11: 720-27. 25. Kripalani S, LeFevre F, Phillips CO, Williams MV, Basaviah P, Baker DW. Deficits in communication and information transfer between hospital-based and primary care physicians: implications for patient safety and continuity of care. JAMA 2007; 297: 831-41. 26. Shojania KG, Duncan BW, McDonald KM, Wachter RM, Markowitz AJ. Making health care safer: a critical analysis of patient safety practices. Evid Rep Technol Assess (Summ) 2001; i-668. 27. Ammenwerth E, de KN. An inventory of evaluation studies of information technology in health care trends in evaluation research 1982-2002. Methods Inf Med 2005; 44: 44-56. 28. Mendis K. Health informatics research in Australia: retrospective analysis using PubMed. Inform Prim Care 2007; 15: 17-23. 29. Dinakarpandian D, Lee Y, Dinakar C. Applications of medical informatics in allergy/immunology. Ann Allergy Asthma Immunol 2007; 99: 2-9. 30. Duvvuri VR, Jianhong W. Information and communication technology developments in asthma management: a systematic review. Ind J Med Sci 2007; 61: 221-41. 31. de Keizer N, Ammenwerth E. The effects and quality of IT evaluation studies: Trends in 1982 - 2002. AMIA Annu Symp Proc 2005; 186-90. 32. de Keizer NF, Ammenwerth E. The quality of evidence in health informatics: How did the quality of healthcare IT evaluation publications develop from 1982 to 2005? Int J Med Inform 2008; 77: 41-9. 73 33. Friedman CP, Abbas UL. Is medical informatics a mature science? A review of measurement practice in outcome studies of clinical systems. Int J Med Inform 2003; 69: 261-72. 34. Eisenstein EL, Ortiz M, Anstrom KJ, Crosslin DR, Lobach DF. Economic evaluation in medical information technology: why the numbers don't add up. AMIA Annu Symp Proc 2006; 914. 35. Eisenstein EL, Ortiz M, Anstrom KJ, Crosslin DR, Lobach DF. Assessing the quality of medical information technology economic evaluations: room for improvement. AMIA Annu Symp Proc 2006; 234-38. 36. Chuang JH, Hripcsak G, Jenders RA. Considering clustering: a methodological review of clinical decision support system studies. Proc AMIA Symp 2000; 146150. 37. Berlin A, Sorani M, Sim I. A taxonomic description of computer-based clinical decision support systems. J Biomed Inform 2006; 39: 656-67. 38. Sim I, Berlin A. A framework for classifying decision support systems. AMIA Annu Symp Proc 2003; 599-603. 39. Cruz-Correia RJ, Vieira-Marques PM, Ferreira AM, Almeida FC, Wyatt JC, Costa-Pereira AM. Reviewing the integration of patient data: how systems are evolving in practice to meet patient needs. BMC Med Inform Decis Mak 2007; 7: 14. 40. Garritty C, El EK. Who's using PDAs? Estimates of PDA use by health care providers: a systematic review of surveys. J Med Internet Res 2006; 8: e7. 41. Oh H, Rizo C, Enkin M, Jadad A. What is eHealth (3): a systematic review of published definitions. J Med Internet Res 2005; 7: e1. 42. Gurses AP, Xiao Y. A systematic review of the literature on multidisciplinary rounds to design information technology. J Am Med Inform Assoc 2006; 13: 267-76. 43. Sargent DJ. Comparison of artificial neural networks with other statistical approaches: results from medical data sets. Cancer 2001;91:1636-1642. 44. Coiera E, Westbrook J, Wyatt J. The safety and quality of decision support systems. Methods Inf Med 2006; 45(Suppl 1): 20-25. 45. Corley ST. Electronic prescribing: a review of costs and benefits. Top Health Inf Manage 2003; 24: 29-38. 46. de Lusignan S. The barriers to clinical coding in general practice: a literature review. Med Inform Internet Med 2005; 30: 89-97. 47. Hanson CW, III, Marshall BE. Artificial intelligence applications in the intensive care unit. Crit Care Med 2001; 29: 427-35. 48. Kaushal R, Barker KN, Bates DW. How can information technology improve patient safety and reduce medication errors in children's health care? Arch Pediatr Adolesc Med 2001; 155: 1002-07. 49. Kuperman GJ, Gibson RF. Computer physician order entry: benefits, costs, and issues. Ann Intern Med 2003; 139: 31-39. 50. Kuperman GJ, Bobb AM, Payne T, Avery AJ, Gandhi TK, Burns G et al. Medication-related clinical decision support in computerized provider order entry systems: a review. J Am Med Inform Assoc 2007; 14: 29-40 51. Papshev D, Peterson AM. Electronic prescribing in ambulatory practice: promises, pitfalls, and potential solutions. Am J Manag Care 2001; 7: 725-36. 52. Stroetmann VN, Thierry J-P, Dobrev A. eHealth for Safety: Impact of ICT on Patient Safety and Risk Management. 2007. 53. Thursky K. Use of computerized decision support systems to improve antibiotic prescribing Expert Rev Anti Infect Ther 2006; 4: 491-507. 54. Zhu Y, Williams S, Zwiggelaar R. Computer technology in detection and staging of prostate carcinoma: a review. Med Image Anal 2006; 10: 178-99. 55. Scottish Intercollegiate Guidelines Network (SIGN) methodology filters. Available from: www.sign.ac.uk/methodology/filters.html (last accessed 4/1/11). 74 56. Oxman AD. Systematic reviews: Checklists for review articles. BMJ 1994; 309: 648-51. 57. Lorenzi NM, Riley RT. Organizational issues = change. Int J Med Inform 2003; 69:197-203. 58. Fernando B, Savelyich BSP, Avery AJ, Sheikh A, Bainbridge M, Horsfield P et al. Prescribing safety features of general practice computer systems: evaluation using simulated test cases. BMJ 2004; 328: 1171-72. 59. Panniers TL, Feuerbach RD, Soeken KL. Methods in informatics: using data derived from a systematic review of health care texts to develop a concept map for use in the neonatal intensive care setting. J Biomed Inform 2003; 36: 23239. 60. Avery AJ, Sheikh A, Hurwitz B, Smeaton L, Chen YF, Howard R et al. Safer medicines management in primary care. Br J Gen Pract 2002; 52(Suppl): S17S22. 61. Sim I, Gorman P, Greenes RA, Haynes RB, Kaplan B, Lehmann H et al. Clinical decision support systems for the practice of evidence-based medicine. J Am Med Inform Assoc 2001; 8: 527-34. 62. Wyatt J, Spiegelhalter D. Evaluating medical expert systems: what to test and how? Med Inform (Lond) 1990; 15: 205-17. 63. Sheikh A, Hurwitz B. Reducing error, improving safety. Log of errors is needed. BMJ 2000; 321: 505. 64. Anderson JG, Jay SJ, Anderson M, Hunt TJ. Evaluating the capability of information technology to prevent adverse drug events: a computer simulation approach. J Am Med Inform Assoc 2002; 9: 479-90. 65. Bell DS, Cretin S, Marken RS, Landman AB. A conceptual framework for evaluating outpatient electronic prescribing systems based on their functional capabilities. J Am Med Inform Assoc 2004; 11: 60-70. 66. Martin P, Haefeli WE, Martin-Facklam M. A drug database model as a central element for computer-supported dose adjustment within a CPOE system. J Am Med Inform Assoc 2004; 11: 427-32. 67. Arai L, Roen K, Roberts H, Popay J. It might work in Oklahoma but will it work in Oakhampton? Context and implementation in the effectiveness literature on domestic smoke detectors. Inj Prev 2005; 11: 148-51. 68. Small SD. Medical device-associated safety and risk: surveillance and stratagems. JAMA 2004; 291: 367-70. 69. Grol R, Grimshaw J. From best evidence to best practice: effective implementation of change in patients' care. Lancet 2003; 362: 1225-30. 70. Kawamoto K, Houlihan CA, Balas EA, Lobach DF. Improving clinical practice using clinical decision support systems: a systematic review of trials to identify features critical to success. BMJ 2005; 330: 765-68. 71. Butler KA, Esposito C, Hebron R. Connecting the design of software to the design of work. Commun ACM 1999; 42: 38-46. 72. Durieux P. Electronic medical alerts -- so simple, so complex. N Engl J Med 2005; 352: 1034-36. 73. Handler JA, Feied CF, Coonan K, Vozenilek J, Gillam M, Peacock PR, Jr. et al. Computerized physician order entry and online decision support. Acad Emerg Med 2004; 11: 1135-41. 74. Holbrook A, Xu S, Banting J. What factors determine the success of clinical decision support systems? AMIA Annu Symp Proc 2003; 862. 75. Zheng K, Padman R, Johnson MP, Diamond HS. Understanding technology adoption in clinical care: Clinician adoption behavior of a point-of-care reminder system. Int J Med Inform 2005; 74: 535-43. 76. Purcell GP. What makes a good clinical decision support system. BMJ 2005; 330: 740-41. 75 77. Casalino L, Gillies RR, Shortell SM, Schmittdiel JA, Bodenheimer T, Robinson JC et al. External incentives, information technology, and organized processes to improve health care quality for patients with chronic diseases. JAMA 2003; 289: 434-441. 78. PLoS Medicine Editors. Many reviews are systematic but some are more transparent and completely reported than others. PLoS Med 2007; 4: e147. 79. World Health Organization Regional Office for Europe. Health Evidence Network (HEN). 2007. 80. Greenhalgh T, Robert G, Macfarlane F, Bate P, Kyriakidou O. Diffusion of innovations in service organizations: systematic review and recommendations. Milbank Q 2004; 82: 581-629. 76 Chapter 3 The National Programme for Information Technology, NHS Connecting for Health and the Information Revolution Summary • The National Programme for Information Technology was one of the most comprehensive, ambitious and expensive eHealth-based overhaul of healthcare delivery ever undertaken. • As a transformation programme for the NHS that underpinned system reform, it was conceived as being far more than just the implementation of information technology. • This Programme had its origins in the 1998 Department of Health strategy Information for Health, which committed the NHS to lifelong electronic health records for everyone with round-the-clock, on-line access to patient records and information about best clinical practice for all NHS clinicians. • Officially launched in 2002, the Programme aimed over a 10-year period to create the infrastructure, tools and environment through which it is possible to deliver: o a longitudinal electronic patient record (from “cradle to grave”) accessible to multiple users throughout the NHS; this (i.e. NHS Care Records Service) together with the dedicated National Network for the NHS (N3) and the national database on which these records will be held (the Spine); represents the backbone to the Programme o a service through which prescriptions can be transferred electronically from general practitioners and other prescribers to pharmacists (Electronic Prescriptions Service) and which will in time integrate with the NHS Care Records Service (Electronic Transmission of Prescriptions) o An electronic appointment booking service enabling patients to electronically book hospital appointments (Choose & Book). 77 Summary continued… • The Programme was subsequently expanded to include, amongst other things: o Picture Archiving and Communication System o GP2GP, which is a system that enables electronic transfer of patient records between GP practices when patients change their practice o Quality Management Analysis System, which automates assessment of GP practice performance against criteria in the new GP contract. • Whilst these were the headline deliverables of the Programme, our scoping of the field identified other related eHealth projects or applications, which, although officially falling within the remit of the now defunct National Knowledge Service (such as ePrescribing and computerised decision support tools), were also within the broad remit of NHS Connecting for Health and were therefore also closely interconnected with the delivery of the Programme. More recently, the Programme expanded its remit yet further to include telehealthcare initiatives within its fold. • Originally managed directly by the Department of Health, oversight of the Programme transferred in 2005 to NHS Connecting for Health–a newly created arm’s length body. More recently, both the Programme and NHS Connecting for Health have been absorbed into the Department of Health’s Informatics Directorate. • Foremost amongst the roles of NHS Connecting for Health is responsibility for the national procurement of systems and services that were needed to ensure delivery of the national Programme. • Given the extremely high level of public expenditure, the Programme has attracted considerable public, professional, legal, financial, political and international scrutiny. • The election of the new coalition government has in effect marked the end of the original Programme; it’s recent strategy document, Liberating the NHS, and White Paper, An Information Revolution, signify a clear move 78 away from a centralised “replace all” to a more locally owned and delivered “connect all” model for health information technology. 3.1 Introduction The National Programme for Information Technology (NPfIT) remains one of the most ambitious and expensive IT-based overhaul of a health service ever undertaken. In this chapter, we aim to give an overview of the Programme being delivered by NHS Connecting for Health (NHS CFH) and the Department of Health (DH). To this end, we describe the key applications that NHS CFH is implementing and the structures that have been created to deliver these. Evidence of the impact of those on quality and safety of care and how they can be best implemented is discussed in later chapters in this report (Section 3). We also reflect on recent changes in the health care IT strategy that have come into force following the formation of the coalition government in May 2010. 3.2 NHS Connecting for Health There are presently many different computer systems in the NHS in England of variable quality and age; there is furthermore no national means to efficiently, securely and confidentially transfer healthcare information between different NHS locations. In July 2004, following the Review of its Arms Length Bodies,1 the Department of Health (DH) announced the creation of a new organisation, combining responsibility for the delivery of the NPfIT with the management of the IT-related functions of the NHS Information Authority (NHSIA), which had previously been earmarked for closure. This newly created organisation, NHS CFH, thus had within its remit the national procurement of critical IT healthcare systems and then ensuring the maintenance, development and effective delivery of these IT products and services. This remit continues following the recent reintegration of the NPfIT (and NHS CFH) into the Department of Health’s Informatics Directorate. Thus, whilst the core aim of NHS CFH is supporting the delivery of 79 NPfIT, it is also responsible for delivering a number of other programmes of work that were previously the responsibility of the NHSIA (see Box 3.1). Established in April 2005 as the national IT procurer for the health service, NHS CFH was charged with providing integrated IT infrastructure and systems for the NHS in England.2 It is, for example, responsible for electronically connecting 1.3 million NHS staff (of whom around 100,000 are doctors, 390,000 nurses and 120,000 other healthcare professionals) and giving patients access to their personal healthcare information. This body has thus, in a very real sense, been laying the technical infrastructure for the NHS. NHS CFH has also provided the policy focus for the DH on IT considerations for the NHS. This has included shaping the strategic infrastructure to ensure integration where necessary and, where choices are available, setting the standards required for local IT applications. This is made clear on NHS CFH’s website, which states that: ‘NHS Connecting for Health operates through a mixed economy of staff drawn from across the NHS, civil service, academia and the private sector. It utilises this rich diversity of experience—encompassing management, IT, clinical and medical skills—to deliver the national Programme and thus improve services to patients.’ Box 3.1 Services and applications being delivered by NHS CFH (for more information on each of those see NHS CFH’s website) • Addressing • Automatic identification and data capture (AIDC) • Blood safety tracking pilot • Choose and Book • Clinical Dashboards • Data Standards • Demographics • Deployment support • Education, Training and Development (ETD) 80 • Electronic Prescription Service (EPS) • ePrescribing • GP Support • Health and Social Care Integration Programme • HealthSpace • Implementation • Informatics Capability Development (ICD) • Information Governance (IG) • Map of Medicine • N3 - The National Network • NHS Business Partners • NHS Care Records Service (NHS CRS) • NHS Infrastructure Maturity Model (NIMM) • NHS Interoperability Toolkit • NHSmail • NHS Pathways • NHS Strategic Tracing Service (NSTS) • Pathology Messaging • Picture Archiving and Communications System (PACS) • Professionalising Health Informatics (PHI) • Prison Health IT • Registration Authorities and Smartcards • Research Capability Programme • Secondary Uses Service (SUS) • Specialty Systems Project • Spine • Summary Care Records (SCR) • Systems & Service Delivery Following the election in May 2010, the new coalition government integrated NHS CFH into the DH’s Informatics Directorate. The publication of its NHS strategy, Equity and Excellence: Liberating the NHS,3 and the subsequent launch of its White Paper, Liberating the NHS: An Information Revolution,4 has 81 resulted in a shift in strategic focus for NHS CFH from a centralised to a more locally led, pluralist, approach to implementing IT. 3.3 Background to the National Programme for Information Technology The Programme had its origins in the 1998 DH strategy Information for Health,5 which committed the NHS to lifelong electronic health records (EHRs), ensuring round-the-clock on-line access to patient records, and information to support best clinical practice for all healthcare professionals working within the NHS. Following the development and publication of the NHS Plan,6 a more detailed supporting document, Building the Information Core: Implementing the NHS Plan,7 published in January 2001, outlined the information and IT systems needed to deliver the NHS Plan and support patient-centred care and services. In March 2001, Derek Wanless, a commissioner with the Statistics Commission was asked to examine future trends affecting the health service in the UK over The Wanless Report,8 published in April 2002, had the next two decades. several key recommendations for IT in the NHS. These included: a doubling and ring-fencing of IT spending; stringent, centrally-managed national standards for data and IT; and better management of IT implementation in the NHS, including a national programme. The Wanless Report coincided with the publication of Delivering the NHS Plan,9 which developed the vision of ‘…a service designed around the patient’, offering more choice of where and when they accessed treatment. In June 2002, following the Wanless Report and Delivering the NHS Plan, the Department of Health published its new strategy for developing IT in the NHS Delivering 21st Century IT Support for the NHS–A National Strategic Programme.10 This strategy laid out the first steps for the Programme, including the creation of a ministerial taskforce and the recruitment of a Director General for the NPfIT. control Formally established in October 2002, it had a mandate to specification, procurement, resource management, performance 82 management and delivery of the IT agenda centrally and implement modern, integrated IT infrastructure and systems for all 8500 general practices and their community services, and the approximately 300 NHS hospitals in England by 2010. The original scope is presented in Figure 3.1. The significantly expanded scope is in contrast presented in Figure 3.2. Figure 3.1: Original scope for the National Programme for Information Technology in 2002 Source: Connecting For Health (2007) Reprinted with permission from NHS Connecting for Health. Crown copyright Figure 3.2: NHS Connecting for Health and National Programme for Information Technology overview in February 2007 83 Source: NHS Connecting For Health (2007) Reprinted with permission from NHS Connecting for Health. Crown copyright. 3.4 Programme scope, function and key activities The NPfIT’s (and NHS CFH and the DH’s) scope continues to evolve, as does the terminology associated with it, such that it is difficult for an outside observer (or indeed those “inside” the organisation) to have a full picture of the Programme’s activities. Broadly, it comprises clinical change, clinical systems, and the underlying infrastructure for the NHS in England, which now serves over 50 million people. A key aim of the NPfIT, as originally conceived, was to support clinicians in providing care by giving them better access to patient information and supporting them in decision-making whenever and wherever this support is needed. The Programme was seen as an essential element in delivering New Labour’s NHS Plan6 to create the infrastructure for facilitating improved patient care by enabling clinicians and other NHS staff to increase the efficiency, effectiveness and safety of healthcare provision. As presented in Figure 3.1, the core original deliverables of the Programme are: • An electronic NHS Care Records Service (NHS CRS), which consists of a nationally transferable Summary Care Record (SCR) and a more local Detailed Care Record (DCR). An essential component of NHS CRS is the Spine—a central storage and communication service for records that also supports other core systems and services. • An electronic prescription service (EPS), this being implemented through the Electronic Transmission of Prescriptions (ETP) programme, which allows for the digital transmission of prescriptions from GPs and other prescribers to pharmacies. Details of dispensed prescriptions are also sent electronically from pharmacies to the Prescription Pricing Authority, the organisation that reimburses dispensers, i.e. community pharmacies for the medication they have supplied to patients. 84 • An electronic booking service, Choose and Book, which aims to make it easier and faster to book convenient and accessible hospital appointments for patients. • A National Network for the NHS (N3)—the largest virtual private network (VPN) in Europe—which provides the IT infrastructure to meet NHS needs with fast, reliable, high-bandwidth connectivity. As noted above, this original list of deliverables, however, expanded to include the following: • GP2GP, which is a system designed to allow direct secure transfer of patient records between GPs when patients change practice. • The Quality Management and Analysis System (QMAS), which collects data on quality of care provided by general practices, measured against national Quality and Outcomes Framework (QOF) targets described in the Revisions to the General Medical Services Contract.11 • HealthSpace has been developed with a wide scope including support of Choose and Book. This is a secure online personal health organiser, which is accessible to anyone aged over 16 and living in England. • Picture Archiving and Communications Systems (PACS), which enables images such as x-rays and scans to be stored electronically and then viewed through any computer or screen linked to the PACS system. • NHSmail provides an email service for NHS staff accessible using web browsers or commonly used email clients. • A linked Directory Service contains details of all NHS staff, including a user interface to facilitate searching for people and browsing of the NHS organisational structure. Directory and NHSmail services also have an integrated outbound fax service and outbound short message service (SMS). Below we consider in more detail some of the above described core deliverables of the Programme. 85 3.4.1 NHS Care Records Service English GPs and primary care providers already routinely use EHRs (see Chapter 6). Whilst useful, these records are however currently only available from local primary care providers (i.e. the practice with which the patient is registered). The NHS CRS in contrast aims to provide a live, interactive patient record service, which is accessible to healthcare professionals at all times, irrespective of wherever they work in the NHS.12 It consists of a locally held Detailed Care Record (DCR) and a nationally transferable SCR. The SCR was first introduced into selected ‘Early Adopter’ practices in 2007 and had been rolled-out to an estimated 400+ GP practices (out of 8,000) by September 2010. Through the use of an electronic SCR service, clinicians can access key details from a patient’s medical record, such as allergies, current prescriptions, date of birth, and address. The SCR is however just the first step toward what is seen as a service to provide safe, secure access to up-to-date detailed healthcare records for every patient in England. The plan is to enable clinicians to access patients’ records securely using a Smartcard, when and where needed, via a nationally maintained information repository. Despite widely acknowledged problems (including concerns about the opt-out consent model, delays in implementation and limited patient interest) with the implementation of the SCR and HealthSpace, the importance of patients having direct access to and the ability to contribute to their own medical records has also been emphasised by the coalition government.13 While the online care record should eventually be—depending on individual access rights—available in full to NHS staff who care for the patient, patients will also for the first time have access to a summary of their healthcare records through HealthSpace. NHS CFH developed an extensive public awareness campaign to introduce this radical new way of sharing and storing patient information to the public, this involves, amongst other things, information leaflets,14 15 and a patient video.16 In an attempt to allay concerns regarding 86 possible breaches of patient confidentiality, the NHS published its commitments to maintain security of data in its NHS Care Records Guarantee (see Chapter 18 for a fuller discussion of these issues).17 An initial independent evaluation of HealthSpace has recently reported.18 The Spine The Spine is the national database of key information about patients' health and care. It forms the quintessential core of the NHS CRS, but will also support other key elements of the Programme such as Choose and Book and ETP, each of these using the Spine's messaging capabilities as part of their own services. Key components making up the Spine include: • Personal Demographics Service (PDS), which facilitates unique linkages of patient data from different sources. As the name implies, this contains patient’s demographic information, including a unique NHS Number, name, address and date of birth.19 The PDS does not however contain any clinical data or information that may be considered particularly sensitive, such as on ethnicity or religion. • Secondary Uses Service (SUS), which provides timely, anonymous patient data for non-clinical care purposes. This includes commissioning, public health, policy and planning and research (e.g. monitoring trends of use of service).20 SUS aims to support national initiatives, such as Payment by Results (PbR). Where possible, SUS captures data automatically through the Programme’s NHS operations systems. The initial content for SUS will be person-specific, building on existing flows such as commissioning datasets, cancer waiting times, clinical audit, central returns and supporting demographic data. It has scope in the future however to include non-person-specific data such as finance, workforce and estates. • Access Control Framework, which controls access to clinical records of patients and registers and authenticates all users storing type of relationship between healthcare professionals and patients, as well as 87 patient preferences on information sharing (for example, whether sensitive information is restricted from sharing).21 National Network for the NHS The National Network for the NHS, better known as N3, connects all NHS organisations and provides the NHS’s IT infrastructure, network services and broadband connectivity requirements.22 The networking solutions provided by N3 has now delivered the systems and services that enable the fast, secure sharing of information, files and data between NHS sites (see Chapter 5 for further details on health information exchange and interoperability). N3 is one of the largest virtual private networks (VPN) in the world. Fast access to up-to-date patient records, the streamlining of clinical practice and a reduction in administrative tasks are all expected from the combination of the latest networking technology and the bespoke applications of the Programme. N3 provides support for NHS organisations in implementing new services, such as the use of video-conferencing for appointments with consultants. It is intended that it will also offer substantial savings on the cost of telephony by enabling NHS organisations to converge their voice and data networks. 3.4.2 Choose and Book Driven by the Government’s agenda to give public and patients more choice, Choose and Book has been introduced as a national electronic referral service to allow patients a choice of place, date and time for their first out-patient appointment. Patients can, if appropriate, choose a hospital or clinic, and book their appointment to see a specialist either in conjunction with a member of the practice team or, if more convenient, from home either by telephone or over the Internet.23 When a GP refers a patient to a specialist, Choose and Book helps identify relevant clinics together with details about availability. After considering 88 individual preferences and clinical requirements, members of the primary care team can assist patients in making their appointment at the time of referral. Patients then receive confirmation of the time, date and location of their appointment. Some patients may want or need more time to consider their choices and, if so, they can take the appointment request letter away with them and book the appointment later, either online or over the phone. Choose and Book thus allows them to discuss options with family members and make any necessary arrangements before scheduling the appointment. The rollout of Choose and Book began with early adopter sites in the summer of 2004, before being rolled-out more widely in 2005 and 2006. Its implementation has now been completed and the majority of patients are now referred for their initial consultant assessment via Choose and Book. Some independent evaluations of Choose and Book have now been published.24 25 3.4.3 ePrescribing In 2003, NPfIT commenced a stakeholder engagement and planning process in relation to perhaps one of the most important eHealth applications, namely implementation of electronic prescribing (hereafter referred to as ePrescribing) in hospitals. ePrescribing is defined by NHS CFH as ‘…the utilisation of electronic systems to facilitate and enhance the communication of a prescription, aiding the choice, administration or supply of a medicine through decision support and providing a robust audit trail for the entire medicines use process’.26 As defined, NHS CFH’s vision of ePrescribing clearly incorporates computerised decision support system functionality (CDSS) (see Chapters 8 and 10 for detailed discussions on CDSS and electronic prescribing). While UK GPs have been using ePrescribing for nearly two decades (with varying degree of functionality, but constantly increasing in sophistication), this has not been the case within the hospital sector in the UK.27 Diversity of 89 different clinical specialty requirements and complexity of prescribing in hospitals have been seen as major barriers to implementation. ePrescribing should, once implemented, facilitate wider improvements in prescribing and administration processes, including reductions in paperwork, improved audit trails for medication and enhanced communication (for example, between hospital departments and pharmacies). ePrescribing, with its built-in functionalities—such as decision and knowledge support, alerts for contra-indications, allergic reactions and drug interactions, formulary guidance or management, training and guidance for prescribers, reminders, and audit trails of medication administration—represents one of NHS CFH’s key activities designed to reduce risk of iatrogenic harm and thereby improve patient safety. Ultimately, the goal is to integrate it seamlessly with EHRs and other computer systems. 3.4.4 Electronic Prescription Service As described above, virtually all UK GPs are already using ePrescribing and use this technology to issue approximately 3.4 million prescriptions on each working day. Given that the annual number of prescriptions issued in England is expected to grow at about five per cent per year, it is clear why the ability to automate repeat prescribing has proven so popular with GPs. Building on these existing primary care computer capabilities, EPS introduces important new add-on functionality. Instead of printing ‘electronically prescribed’ prescriptions (i.e. prescribed with the support of ePrescribing system), GPs will be in a position to electronically send the prescription to the dispensing pharmacist. The new system should lead to significant time savings, especially insofar as it enables the more efficient processing of repeat prescriptions, which now make up about 70 per cent of all prescriptions issued. The sending of repeat prescriptions using EPS will in due course happen (semi) automatically. 90 These electronic prescriptions are received by the EPS, a secure information hub, from which they can then be downloaded by participating dispensers. True paperless prescribing will occur in a later release, when patients will be able to choose (or “nominate”) a particular dispensing contractor, i.e. pharmacy to automatically receive their electronic prescriptions. In time, the electronic submission of reimbursement claims by these dispensers will also be supported by the EPS system. The EPS is being introduced in primary care settings, such as GP practices and community pharmacies, across England, and will continue to expand to other settings in which primary care prescribing takes place. The first iteration of the EPS, known as Release 1, which is mostly invisible to patients and end users, went live in February 2005.28 Progress is behind schedule, however; Release 2, which will be in two main phases, has now begun: Stage one will be the 'transition' phase, and stage two will be the 'full EPS' phase.29 30 In the ‘transition’ phase, prescribers and dispensers will gradually begin replacing most paper prescriptions with electronic prescriptions. This is when intended EPS-related benefits should really begin to be demonstrated as it becomes possible for patients to nominate dispensers to receive their electronic prescriptions and for dispensaries to electronically submit reimbursement claims. Some hand-signed FP10 prescription forms will continue to be required in this phase as only those dispensers working with nominated prescriptions will be able to accept electronically-signed prescriptions. By the ’full EPS’ phase, most sites will be using Release 2. At this point, all within-scope prescriptions (e.g. controlled drugs). will be able to be sent, signed and received electronically— whether nominated or not. The plan is that in time, the EPS will be integrated with NHS CRS. Information on what has been prescribed and dispensed will be able to be automatically recorded in the patient’s care record. Access to more accurate (or entirely new) information about what has been prescribed and dispensed will be of great 91 assistance to the healthcare professionals who need it (for example, a doctor in an accident and emergency department or a GP a patient visits while away from home), thus helping to improve standards of care. NHS Connecting for Health Evaluation Programme has commissioned an evaluation project to evaluate the impact of the new service on quality and safety of care, which will report later in 2011.31 3.4.5 Picture Archiving and Communications Systems Picture Archiving and Communications Systems (PACS) capture, store, distribute and display static or moving digital images such as x-rays or scans, thereby potentially enabling more efficient diagnosis and treatment. The ability to store digital images will form an essential part of every patient’s electronic record, thereby potentially removing the need to print on film and to file or distribute images manually. These images can then be sent and viewed at one, or across several, NHS locations. This should in turn result in increased capacity of diagnostic services and should furthermore speed up the time to diagnosis. For patients, this should also mean fewer delays, fewer wasted appointments due to lost or low quality images, and less retesting, which in turn should reduce the total lifetime radiation dose. Patient care also has the potential to benefit as clinicians and care teams have the ability to work together viewing common information across one or more locations. PACS has been procured to provide full access to digital images in NHS organisations throughout England and its implementation is now complete. An initial evaluation of PACS has reported.32 3.4.6 The NHSmail Email and Directory Service Historically, there have been many different local email systems operating in the NHS. The quality and reliability of these services varied substantially and they incurred substantial costs. In addition, none of the services were secure enough to allow the transmission of patient information resulting in information 92 being sent frequently via mail or fax incurring further costs for paper, printing and postage as well as slowing down the process. The NHSmail service was launched in October 2004 to replace these disparate email services with a standardised platform that was capable of being used for the transmission of patient information. NHSmail provides a central, secure email service, thus reducing the overall email costs to the NHS and providing a swift and secure means of exchanging information across the NHS.33 The NHSmail platform is a centrally-funded service that is free at the point of use, and offers a helpdesk that is available to support users 24 hours a day, seven days a week. It preserves patient confidentiality and is protected by both anti-virus and anti-spam software. NHSmail is the only system approved by the DH, British Medical Association (BMA) and the Royal College of Nursing (RCN) for NHS users to securely exchange patient data within the NHS. NHSmail aims to improve workflow by allowing authorised users to log-on from anywhere, at any time, giving them access to the NHS Directory, with contact details for over one million NHS staff. NHSmail also includes features that allow users to send free SMS and fax messages directly from the system, and to share calendars and mailboxes with other NHSmail users.33 It has been upgraded and is now based on Microsoft’s Exchange 2007, but still only provides limited storage space (20Mb). 3.4.7 Support for primary care GP Systems of Choice This is the NHS CFH scheme through which the NHS funds the provision of GP clinical IT systems in England. GP Systems of Choice (GPSoC) allows practices and PCTs to benefit from a range of quality assured GP clinical IT systems purchased from existing suppliers who are contracted to work within the NPfIT. Practices can choose between systems provided by their local service provider or by suppliers that are contracted to offer systems on the GPSoC Framework. 93 GPSoC introduces standards which will improve the quality of service that practices receive from their GPSoC framework supplier. The Quality Management and Analysis System (QMAS) The new GMS contract was introduced in April 2004. A key component was the Quality and Outcomes Framework (QOF) of national achievement targets describing how GP practices would be rewarded financially, based on their achievement in up to four domains: clinical; organisational; patient experience; and additional services. To support the QOF and the GMS contract, the Programme developed a single, national IT system, known as Quality Management and Analysis System (QMAS). QMAS tracks the performance of each practice, comparing it against QOF goals allowing for the payment of financial incentives based on quality of care.34 The system is also used to compute national disease prevalence rates for a range of conditions.35 QMAS last underwent a major upgrade in August 2006, in support of the 2006–07 GMS contract.34 GP2GP When patients transfer from the care of one GP to another, it is clearly important that their medical records be transferred as swiftly as possible. Ensuring that medical records arrive in time for the patient’s first consultation may have significant clinical benefits, as it has been estimated that not having this information could threaten the quality of care in over 50 per cent of consultations.36 Using the traditional paper-based system, the transfer of patient records may take weeks or months leaving clinicians without the information they need to provide the best possible care for patients.36 GP2GP is designed to address this problem. It enables the transfer of the electronic component of a GP patient health record. When a patient registers with a new practice, GP2GP will move the patient’s current EHR from their previous general practice to their new one. On receipt of the EHR, the new 94 practice will undertake certain housekeeping activities, e.g. authorise current repeat prescriptions listed in the EHR, and will have the clinically important medical history in hand at the time of the patient’s first consultation (see Chapter 5).36 The GP2GP system includes a feature called “autosend”, which allows the EHR to be sent without any action on the part of the sending GP. Not only does this cut down on the administrative cost to the sender, but it also ensures that the patient’s new GP will receive the records quickly, within minutes or hours.37 The GP2GP system met its March 2007 implementation goal of achieving rollout in 500 practices and is currently engaged in a nationwide rollout among practices meeting certain 'entry and readiness criteria’.36 Current estimates suggest that 60 per cent of practices are now using GP2GP with an average of 10,000 transfers taking place each week.38 Apart from the aforementioned NPfIT deliverables, NHS CFH is involved in a number of other work-streams and deliverables with regards to IT implementation. 3.5 An information revolution? The formation of the new coalition government in May 2010–with its mandate to usher in a new austerity phase for public services–has resulted in a change in direction for NPfIT (see Box 3.2). Whilst the details of the new direction have yet to be agreed, it seems clear from the overall strategy document, Equity and Excellence: Liberating the NHS, and the consultation White Paper, Liberating the NHS: An Information Revolution, that central IT expenditure will be substantially scaled back, but there will remain a commitment to maintaining the centralised infrastructure of the NPfIT, in particular N3, the SCR, and the Spine.3 4 There is also a clear commitment to expanding the role of telehealthcare initiatives allowing healthcare to be delivered remotely, and enabling self-care. The key underpinning thrust of the consultation paper however is for more localised IT decision-making and responsibility, and the far 95 greater release of data to support all aspects of healthcare delivery and research. 3.6 Conclusions NHS CFH and its national Programme are complex ambitious endeavours that have impacted on many aspects of NHS care provision. Whilst many of the infrastructure-related aspirations have been delivered, the deliverables that have the potential to impact more substantially on day-to-day patient care such as the SCR, DCR, HealthSpace and EPS have proved more challenging to deliver. Whether the change in philosophy from a “replace all” to a “connect all” approach ushered in by the change of government proves more successful remains to be seen. Initially, the core aspirations of this modernisation agenda relate to improving data storing and management and supporting professional decision-making; more recently, this focus has expanded to also include the delivery of care from a distance. We focus on these three issues in the next chapter and explore the potential these have to impact on the quality and safety of healthcare provision. 96 Box 3.2 DH statement on the future of the NPfIT, 9 September 201039 ‘A Department of Health review of the National Programme for IT has concluded that a centralised, national approach is no longer required, and that a more locallyled plural system of procurement should operate, whilst continuing with national applications already procured. A new approach to implementation will take a modular approach, allowing NHS organisations to introduce smaller, more manageable change, in line with their business requirements and capacity. NHS services will be the customers of a more plural system of IT embodying the core assumption of ‘connect all’, rather than ‘replace all’ systems. This reflects the coalition government’s commitment to ending top-down government and enabling localised decision-making. The review of the National Programme for IT has also concluded that retaining a national infrastructure will deliver best value for taxpayers. Applications such as Choose and Book, Electronic Prescription Service and PACS have been delivered and are now integrated with the running of current health services. Now there is a level of maturity in these applications they no longer need to be managed as projects but as IT services under the control of the NHS. Consequently, in line with the broader NHS reforms, the National Programme for IT will no longer be run as a centralised national programme and decision making and responsibility will be localised.’ References 1. Department of Health. Reconfiguring the Department of Health's Arm's Length Bodies. London: HMSO, 2004. 2. Department of Health. Arm's length bodies review. Available from: http://www.dh.gov.uk/en/Aboutus/Deliveringhealthandsocialcare/Armslengthbodie s/DH_4105578 (last accessed 4/1/11). 3. Department of Health. Equity and Excellence: Liberating the NHS. London: HMSO, 2010. 4. Department of Health. Liberating the NHS: An Information Revolution. London: HMSO, 2010. 5. Department of Health. Information for health an information strategy for the modern NHS. Available from: 97 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. http://www.dh.gov.uk/en/PublicationsAndStatistics/LettersAndCirculars/HealthSer viceCirculars/DH_4005016 (last accessed 4/1/11). Department of Health. The NHS Plan: A Plan for Investment, a Plan for Reform. London: Department of Health, 2000. Department of Health. Building the Information Core: Implementing the NHS Plan. London: Department of Health, 2001. Wanless D. The Wanless Report. London: HMSO, 2002. Department of Health. Delivering the NHS Plan: next steps on investment, next steps on reform. London: Department of Health, 2002. Department of Health. Delivering 21st century IT support for the NHS: national strategic programme. London: Department of Health, 2002. Revisions to the GMS contract 2006/07 Delivering investment in general practice. Available from: http://www.bma.org.uk/employmentandcontracts/independent_contractors/genera l_medical_services_contract/NewDEs0706.jsp (last accessed 4/1/11). Robertson A, Cresswell K, Takian A, Petrakaki D, Crowe S, Cornford T, et al. Implementation and adoption of nationwide electronic health records in secondary care in England: qualitative analysis of interim results from a prospective national evaluation. BMJ 2010; 341: c4564. NHS Connecting for Health. Review agrees value of a core Summary Care Record http://www.connectingforhealth.nhs.uk/newsroom/news/valueofscr (last accessed 3/1/11). NHS Connecting for Health. Patient Information leaflet: The NHS Care Records Service Better information for better, safer care. Available from: http://www.stokepct.nhs.uk/pdfs/scr/easy_read_summary_leaflet.pdf (last accessed 4/1/11). NHS Connecting for Health. Available from: http://www.nhscarerecords.nhs.uk/faqs/ (last accessed 4/1/11). NHS Connecting for Health. NHS Care Records Service Patient Video. Available from: http://www.connectingforhealth.nhs.uk/v/?v=nhscfhcrsvideo.mp4&w=320&h=255 (last accessed 4/1/11). National Information Governance Board for Health and Social Care. The NHS Care Record Guarantee. Available from: http://www.nigb.nhs.uk/guarantee_guarantee.pdf (last accessed 4/1/11). Greenhalgh T, Hinder S, Stramer K, Bratan T, Russell J. Adoption, non-adoption, and abandonment of a personal electronic health record: case study of HealthSpace. BMJ 2010; 341: c5814. NHS Connecting for Health. The Personal Demographics Service. Available from: http://www.connectingforhealth.nhs.uk/systemsandservices/demographics/pds (last accessed 4/1/11). NHS Connecting for Health. Secondary Uses Service. Available from: http://www.connectingforhealth.nhs.uk/systemsandservices/sus/ (last accessed 4/1/11). NHS Connecting for Health. Spine Factsheet. Available from: http://www.connectingforhealth.nhs.uk/resources/systserv/spine-factsheet (last accessed 4/1/11). NHS Connecting for Health. N3. Available from: http://www.n3.nhs.uk/ (last accessed 4/1/11). Choose and Book Website. Available from: http://www.chooseandbook.nhs.uk/staff/started/getting-started-overview (last accessed 4/1/11). 98 24. Rashid M, Abeysundra L, Mohd-Isa A, Khan Y, Sismeiro C. Two years and 196 million pounds later: where is Choose and Book? Inform Prim Care 2007; 15: 111-19. 25. Rabiei R, Bath PA, Hutchinson A, Burke D. The National Programme for IT in England: clinicians' views on the impact of the Choose and Book service. Health Informatics J 2009; 15:167-78. 26. NHS Connecting for Health. ePrescribing FAQs. Available from: http://www.connectingforhealth.nhs.uk/systemsandservices/eprescribing/faqs (last accessed 4/1/11). 27. Crowe S, Cresswell K, Avery AJ, Slee A, Coleman JJ, Sheikh A. Planned implementations of ePrescribing systems in NHS hospitals in England: a questionnaire study. JRSM Short Rep 2010; 1: 33. 28. NHS Connecting for Health. Electronic Prescription Service (EPS). Available (last from: http://www.connectingforhealth.nhs.uk/systemsandservices/eps accessed 4/1/11). 29. NHS Connecting for Health. Introducing release 2. Available from: http://www.connectingforhealth.nhs.uk/systemsandservices/eps/introduced/releas e2 (last accessed 4/1/11). 30. NHS Connecting for Health. EPS Release 2 national rollout. Available from: http://www.connectingforhealth.nhs.uk/systemsandservices/eps/staff/implementati on/release2/rollout (last accessed 4/1/11). 31. Lichtner V, Petrakaki D, Hibberd R, Venters W, Cornford A, Barber N. Mapping stakeholders for system evaluation - the case of the Electronic Prescription Service in England. Stud Health Technol Inform. 2010;160(Pt 2):1221-5. 32. Collin S, Reeves BC, Hendy J, Fulop N, Hutchings A, Priedane E. Implementation of computerised physician order entry (CPOE) and picture archiving and communication systems (PACS) in the NHS: quantitative before and after study. BMJ 2008; 337: a939. 33. NHS Connecting for Health. About NHSmail. Available from: http://www.connectingforhealth.nhs.uk/systemsandservices/nhsmail/about (last accessed 4/1/11) 34. NHS Connecting for Health. What is QMAS? Available from: http://www.connectingforhealth.nhs.uk/systemsandservices/gpsupport/qmas (last accessed 4/1/11). 35. Connecting for Health. QMAS IT - QMAS getting ready for the end of the year. Available from: http://www.connectingforhealth.nhs.uk/systemsandservices/gpsupport/qmas/qma sendofyear.pdf (last accessed 4/1/11). 36. NHS Connecting for Health. GP2GP - Electronic patient record transfer. Available from: http://www.connectingforhealth.nhs.uk/systemsandservices/gpsupport/gp2gp/impl ementation/3914.pdf (last accessed 4/1/11). 37. NHS Connecting for Health. Explaining autosend. Available from: http://www.connectingforhealth.nhs.uk/systemsandservices/gpsupport/gp2gp/how doesitwork/autosend (last accessed 4/1/11). 38. NHS Connecting for Health. GP2GP. Available from: http://www.connectingforhealth.nhs.uk/systemsandservices/gpsupport/gp2gp/pro gress (last accessed 4/1/11). 39. Department of Health. The future of the National Programme for IT Available from: http://www.dh.gov.uk/en/MediaCentre/Pressreleases/DH_119293 (last accessed 4/1/11). 99 Chapter 4 Exploring, describing and integrating the fields of quality, safety and eHealth Summary • eHealth is a relatively new and rapidly evolving field and so many of the concepts, terms and applications are still evolving. • There is furthermore no agreed definition of eHealth, with some researchers using this to relate primarily to the area of consumer informatics, whereas others use it more generically to refer to any of the ways in which information technology can be employed to improve delivery of healthcare. For the purposes of this report, we considered it important to use an inclusive definition and chose to use Eysenbach’s definition, as adapted by Pagliari, as the basis for our work: ‘eHealth is an emerging field of medical informatics, referring to the organisation and delivery of health services and information using the Internet and related technologies. In a broader sense, the term characterises not only a technical development, but also a new way of working, an attitude, and a commitment for networked, global thinking, to improve healthcare locally, regionally and worldwide by using information and communication technology.’ • Whilst the number of eHealth applications is very large, these can conceptually be divided into three broad domains relating to key activities they support: o storing, managing and sharing data o informing and supporting clinical decision-making o delivering expert professional and/or consumer care remotely. • Most frameworks of quality currently in use incorporate the following key dimensions of care: effectiveness of treatments; appropriateness of the means of delivery; acceptability; efficiency; and equity. 100 Summary continued… • Whilst there are no internationally agreed definitions of patient safety, adaptations of the National Patient Safety Agency’s definition of Patient Safety Incidents is increasingly being used. This, in its original definition, states that a ’…patient safety incident is any unintended or unexpected incident which could have harmed or did lead to harm for one or more patients being cared for by the NHS.’ • There are a number of patient safety taxonomies; however, our scoping of this literature found that the Joint Commission on Accreditation of Healthcare Organizations Patient Safety Event Taxonomy and the related World Health Organization’s Classification are the most comprehensive and clinically relevant in that they incorporate five key areas: o impact of medical error o type of processes that failed o domain, i.e. the setting in which an incident occurred o cause or factors leading to the safety incident o prevention and mitigation factors to reduce risk of recurrence and/or improve outcomes in the case of a further incident. • Integrating the fields of eHealth, quality and safety clearly demonstrates the numerous ways in which information technology has the potential to improve the effectiveness and efficiency of many facets of healthcare delivery. This includes, for example, helping clinicians to access comprehensive information on their patients, aiding monitoring of their conditions and the treatments being issued, reducing inappropriate variability in healthcare delivery, and proactively identifying and alerting clinicians to threats to patient safety. • The integration of these domains however also highlights ways in which introduction of new eHealth applications can result in unintended consequences and inadvertently introduce new risks. 101 4.1 Introduction In order to help identify the focus of our work, during the formative phase of work we undertook mapping of the domains of eHealth, quality and safety, looking particularly for areas of intersection between these fields of enquiry. This proved challenging for several reasons, not least that eHealth, quality and safety are relatively novel and at times contested fields of enquiry. This chapter summarises our attempts at this underpinning conceptual work and provides the framework within which we interpreted the findings chapters reported in Section 2 of this volume. The analytic framework that we have developed thus draws on the following main strands: • Exploration of notions and constructs of quality and its assessment • A classification of patient safety risks, derived from the World Health Organization’s (WHO) International Classification for Patient Safety1 and a systematic overview of safety taxonomies by the Joint Commission on Accreditation of Healthcare Organizations2 • The core applications of eHealth, represented within the conceptual model developed by Pagliari et al.3 • The headline deliverables of NHS Connecting for Health (NHS CFH) and other key elements of the National Programme for Information Technology (NPfIT) (See Chapter 3). These strands encompass a number of cross-cutting themes around behavioural and organisational change, usability and interoperability of technology and its evaluation. 4.2 Quality of healthcare Modern medicine faces important challenges. Never before has scientific development impacted so profoundly on human health and our knowledge on how to impact on disease and health grown so rapidly. Several hundred thousand scientific articles are now published each year that in one way or 102 another propose improvements to ways of improving health and/or the delivery of healthcare. Since the first randomised controlled trial (RCT) conducted over 60 years ago, the number has now grown to over 10,000 annually.4 5 This new evidence in turn creates a constant need for constant questioning of the evidence underpinning medicine and, in many cases, change. Despite these advances in our ability to use new medicines, procedures and approaches to improve the health of individuals and populations, there is a growing body of knowledge that demonstrates that translating these insights into practice frequently fails.6-10 Over the last decade, hundreds of authoritative studies have clearly revealed the very considerable variations in the performance of healthcare professionals and indeed health systems;6-10 the net effect of this variation is that, even in countries with mature health systems, typically only about 50 per cent of the population receive optimal standards of care.11 Moreover, there is the continuing paradox of the inverse care law, describing how those most in need are typically least likely to receive care.12 13 There is a long history of interest in the quality of care.14 Over forty years ago, Avedis Donabedian,15 widely considered the father of the quality movement,16 17 invigorated focus on quality of healthcare as a measure of the pursuit of best care. In a broader sense, the focus on quality is a result of a complex convergence of scientific, political, economic and social issues. But despite this long history of focusing on quality considerations, the response by the medical profession has tended to be mixed.18 Maximising patient safety is an essential component of the wider quality improvement agenda. Although these topics are often considered separately, reports such as the Institute of Medicine’s Crossing the Quality Chasm8 and To Err is Human7 closely aligned safety and quality, clearly emphasising the potential of information technology (IT) to improve both the quality and safety of care. Of note is that analyses of safety incidents have in a sense served as a catalyst to the integration of the fields of informatics, human factors engineering, 103 and cognitive and social psychology, bringing these within the fold of the quality agenda.19 4.2.1 Defining and measuring the quality of healthcare Experts have struggled to formulate a concise, meaningful, and generally applicable definition of quality as it pertains to healthcare.18 As with safety (discussed below), there remains no single agreed definition of quality, although many have attempted to elucidate the concept and many broad frameworks for understanding it have been proposed.15 20-31 While trying to capture the essence of this elusive construct, all definitions agree that quality care is that which helps an individual and/or population to maximise their welfare, duration of life, and which leads to desired health outcomes. For example, Donabedian defines it in terms of the structures, processes and outcomes of care;15 Maxwell20 in terms of accessibility, effectiveness, efficiency, acceptability, equity and relevance; and Schuster in terms of the underuse, misuse and overuse of services.23 Others have presented a detailed list of condition-specific quality of care indicators.30 The simple and widely cited framework developed by Campbell et al. draws on several of these characterisations and considers quality in terms of its relevance to the individual—access and effectiveness (of clinical care and inter-personal care) and the public or populations—access, effectiveness, equity, efficiency and cost.25 We have integrated these within the conceptual map shown as Figure 4.1, which highlights the core domains of effectiveness, efficiency, equity, acceptability and access, illustrating their inter-dependencies (shown by the arrows and repetition of sub-terms). This integration also recognises the central role of effectiveness in both the quality and safety debate, echoing the assertion that ‘…quality-improvement efforts based on evidence of effectiveness are likely to be more readily embraced and may save more lives than will efforts to improve safety that cleave to the concept of accidental death and lack a solid evidence-base.’19 Cost is subsumed within efficiency, although the effectiveness, equity, accessibility and acceptability of care also have downstream cost implications. One of the most widely adopted definitions of quality is the Institute of Medicine’s, which defines quality as the 104 ‘…degree to which health services for individuals and populations increase the likelihood of desired health outcomes and are consistent with current professional knowledge.’8 This is the definition we have adopted in our report. Figure 4.1 Map of quality concepts Efficiency Streaming processes Increasing access (to data) Reducing costs Acceptability Enhanced patient satisfaction communication, empowerment Effectiveness Clinical appropriateness (evidence-base + tailoring) Clinical impact Equity Enhanced access Reduced variation Access To services, knowledge, data, people, support 105 Measures and indicators of quality of care Lilford et al. highlight how:32 ‘The distinction between a measure of quality and an indicator of quality is important. Generally speaking we have very few real measures of quality. For example, post-operative length of stay is a measure of the patient's hospital stay, but only an indicator of quality, e.g. a patient's long stay might represent postoperative complications or poor discharge arrangements. Thus, the term indicator is preferable.’ Quality indicators assess aspects of care associated with patient outcomes in a variety of domains. Broadly speaking, indicators can be summarised into those that assess the technical provision of care and the interactions between patients and providers. Concepts such as effectiveness, access, capacity, safety, patient-centredness, equity and disparities are commonly used to describe the technical or inter-personal aspects of care. Measuring the quality of care or the impact of interventions aiming to improve quality is thus far from simple33 and needs adjusting for, amongst other things, treatment refusal.34 A further complication comes from the observation that different methods of measurement of the quality of care do not necessarily provide the same answers.32 35 This complexity (and in some case vagueness) of definitions of quality means that researchers and practitioners often find it difficult to understand and relate to it. The very language and terminology of quality of care leaves many healthcare professionals somewhat bewildered. Terms such as desired patient outcomes, patient-reported outcome measures, criteria, standards, profiles, observed and expected mortality, case-mix and case-severity adjustments, charts, tables, leagues, quality control, continuous quality management, quality improvement, quality assurance are often poorly understood. Moreover, what constitutes quality often changes depending on the goal of the measurement.36 106 One overarching benchmark for quality however, as defined by the Institute of Medicine, is provision of care according to current best knowledge. Broadly speaking, measures of quality can be classified into those that measure processes and those that measure outcomes. As a process, the quality of care can be measured according to whether patients were offered recommended services. For example, a study of processes found that quality of care increased as the number of long-term conditions experienced by patients increased.37 Process measures are really only valid when there is a proven relationship with health outcomes; such process measures are often termed “intermediate clinical outcomes” because they mediate the association between the process and outcomes of care. For example, the relationship between glycaemic control and risk of diabetes-related complications is well established and so HbA1C is in people with diabetes a valid process measure of quality. Another commonly measured process measure of quality of care is rate of admissions to hospital.38 For example, a high rate of emergency or unplanned admissions to hospital is generally considered to be a reflection of suboptimal community-based patient care; and admission rates for so-called “ambulatory care sensitive conditions” have been proposed as a marker of the quality care received by patients with these conditions in the community. However, in high income countries such as the UK, which have well-developed health systems offering universal coverage, the impact of socio-economic factors (such as deprivation) and health behaviours (such as smoking) are likely to outweigh the influence of healthcare factors on unplanned hospital admissions for longterm illnesses.39 Outcome measures of quality are on the other hand only appropriate to study if they are modifiable as a result of a process that aims to improve this outcome. So deaths from causes that are not modifiable cannot be considered measures of quality of care. Alternative measures, such as premature deaths from long-term illnesses (for example, deaths from 107 coronary heart disease or stroke in people under 65 years of age) have been proposed as markers of the quality of health systems. In-hospital death rates in various forms, (e.g. standardised hospital mortality ratios) have also been proposed as markers of quality or safety. However, their use has been criticised because in-hospital deaths will be strongly influenced by variations in hospital case-mix. Furthermore, the methods used to standardise inhospital mortality are strongly influenced by the coding of secondary diagnoses in hospital records and this can introduce artefactual errors and bias in the estimates of standardised hospital mortality rates.40 There is also now a clear recognition of the importance of patients’ or consumers’ perspectives into what constitutes quality healthcare. Language has adapted to reflect patient-centredness and researchers now routinely talk about “desired health outcomes”, (i.e. those desired by the patient and not necessarily doctor) and “patient expectations”. Variability in measured quality of care poses a great challenge for policymakers, healthcare professionals and patients alike with outcomes ’…influenced by definitions, data quality, patient case-mix, clinical quality of care and chance.’ 32 For example, an association between a higher volume of activity and improved outcomes in hospital care is now supported by evidence from more than 300 studies across a wide range of procedures and conditions.41 Nevertheless, evidence is far from uniform.42-44 The clinical and policy significance of these findings is complicated by the methodological shortcomings of many studies. Differences in case mix and processes of care between high- and low-volume providers may explain part of the observed relationship between volume and outcome. That said, it is far from clear how policymakers should act on the finding that high volume of activity is associated with better outcomes. This research has also been largely hospital-based, and the association between provider size or volume of activity and outcomes in primary care is less well understood. 108 While measurable differences do not necessarily translate into clinically meaningful differences in patients' lives,45 46 measuring quality of care is nonetheless an essential prerequisite to quality improvement. Only by measuring outcomes of quality improvement initiatives and interventions can we know whether they actually result in an improvement. The UK Quality and Outcomes Framework (QOF) is an example of a national policy initiative aimed at measuring the quality of primary care.47 In 2006, in an attempt to improve the quality of care, the US Congress signed a pact with the American Medical Association which pledged to develop over 100 standard measures of doctors performance that will be reported to the federal government.48 4.2.2 Improving quality of care Problems with quality have traditionally been classified into three areas: shortage of technical and/or interpersonal competence; overuse or unnecessary or inappropriate use of services; and underuse of or lack of access to needed and appropriate services. Appropriateness is thus at the heart of defining quality of care (see Figure 4.1).49 Yet defining what is appropriate is far from easy50;51 and indicators of quality of care cannot simply be transferred between countries or indeed clinical settings as variation in professional culture or clinical practice may be important confounders.52 Improving quality of care is complex and includes approaches such as: implementation of evidence-based practice (where possible catalysed by the development of clinical practice guidelines for example);53 leadership commitment;54 continuous quality improvement;55 continuing medical education and professional development;56 regulation; assessment and accountability; competition, continuing professional development;57 and feedback; 59 58 audit and patient empowerment through, for example, creating public pressure by publishing performance data.60 61 The UK is now seen as an international leader in investigating the impact of financial incentives to drive quality improvement initiatives, which overall seem to have relatively modest short-term effects.47;62 An example of a national initiative is The 109 100 000 Lives Campaign and building on this, the 5 Million Lives Campaign.63;64 Although attempts to improve quality are numerous, the strength of the evidence in support of most of these initiatives is either weak or absent.6;65 We lack an understanding of which approaches are most appropriate for what types of improvement in what settings and of the determinants of successful performance change and none of the popular models for improving clinical performance appear to be superior.6 Furthermore, it has only recently been recognised that quality and quality improvement initiatives, as well as schemes to improve patient choice, need to include mechanisms to also measure and monitor potential harms.46 4.3. Safety Approximately 850,000 medical errors occur in NHS hospitals every year resulting in an estimated 40,000 deaths.66;67 These are, by any standard, shocking statistics and reflect an unacceptable state of healthcare.68 Safety is an intricate feature of high quality care although it has only begun to receive attention as a quality parameter in recent years. The Institute of Medicine’s landmark reports on healthcare safety and quality, To Err Is Human7 and Crossing the Quality Chasm8 and equivalent reports in the UK66 69 have articulated a broad agenda for quality and safety improvement in healthcare that needs to be based on evidence,70 recognising the current unacceptable state of identifying, analysing and learning from medical mishaps.71 4.3.1 Defining the safety of healthcare One of the most fundamental challenges of patient safety is the variety of different concepts underlying this broad construct and the lack of a single agreed definition; the range and number of classification schemes72-81 characterising incident reporting systems and patient safety research reflects 110 these underlying tensions. Diverse understanding of patient safety related terminology82-84 impacts importantly on how we are able to learn from potentially dangerous events85 when attempting to interpret or synthesise the results of published studies, which address somewhat different theoretical propositions, (e.g. avoidance of error, risk prevention, reporting or investigation of incidents) and use different outcome measures. Patient safety encompasses aspects such as harm to patients, incidents that lead to harm, the antecedents or processes that increase the likelihood of incidents, and the attributes of organisations that help guard against harm and enable rapid recovery when risk escalates.86 Most definitions of patient safety and medical error recognise that organisational factors interact with human factors to facilitate and mitigate errors.87 There is, however, a tension between focusing on individual practices and seeing safety primarily as a systems’ problem.88 The latter approach focuses on developing systems that prevent errors and, equally importantly, ensures that clinicians provide effective care.89;90 Shojania et al. have defined a patient safety practice as ‘…a type of process or structure whose application reduces the probability of adverse events resulting from exposure to the healthcare system across a range of diseases and procedures.’91 In conceptualising patient safety we found it helpful to focus on its goal. Battles and Lilford92 specify how the ‘…goal of patient safety is to reduce the risk of injury or harm to patients from the structure and process of care. This can be accomplished by eliminating or minimising unintended risks and hazards associated with the structure and process of care’ and ‘…a vision for patient safety would be zero health care associated injuries or harm.’ In an attempt systematically to organise and represent patient safety theory we considered various other definitions and constructs of safety related 111 terminology.72-81;93 What became evident is that in the past efforts to define and classify “errors” or “mistakes” were somewhat theoretically and methodologically flawed.74 Contributions to learning from risk-averse industries such as aviation94-96 have been seminal in recognising key safety principles, such as: that errors can never be completely eradicated and generally derive from faulty system design not from negligence; that major safety incidents are only the ”tip of the iceberg” of procedures and processes that indicate possibilities for organisational learning; and most importantly that prevention of safety incidents should be an ongoing process based on open and full disclosure and reporting. The key objective of designing safe systems is to make it difficult for the individual to err and to design systems that absorb errors, mistakes, slips and lapses that inevitably occur due to a characteristic of human nature—fallibility.96 Among many safety models Reason's Swiss Cheese Model has become the dominant paradigm for analysing medical errors and patient safety incidents (See Figure 4.2).97 98 The holes in the defences arise for two reasons: active failures (e.g. shortage in communication or improper ventilation technique); and latent conditions (e.g. inadequate patient monitoring, or inadequate staffing skills mix). Nearly all adverse events involve a combination of these two sets of factors.99 Vincent et al. have adapted and further developed Reason’s model and describe the people who are directly involved as the inheritors, rather than the instigators of an accident sequence (see Figure 4.3).80 These frameworks facilitated systematic and conceptually driven approaches to organisational risk assessment. 112 Fig 4.2 The Swiss Cheese Model of accident causation—model of how defences, barriers, and safeguards may be penetrated by an accident trajectory with permission from British Medical Journal. Figure 4.3 The anatomy of an organisational accident Reprinted from Vincent et al. (1998) with permission from British Medical Journal. When we approach quality and safety as intricately related we then see how safety incidents can lie either in the structure or the process of care. Figure 4.4 illustrates the process as occurring within the structure of healthcare.92 113 Figure 4.4 A structure and process model for patient safety based on Donabedian Reprinted with permission from J Battles 92 Quality and Safety in Health Care Recognising the problem of multiple definitions of patient safety related concepts, a number of organisations have advocated the use of a standard taxonomy and terminology.101 The National Patient Safety Agency’s (NPSA) definition of Patient Safety Incidents states that a ‘…patient safety incident is any unintended or unexpected incident which could have harmed or did lead to harm for one or more patients being cared for by the NHS.’ More recently the WHO’s World Alliance for Patient Safety has produced a conceptual map and classification scheme in collaboration with a number of high profile international organisations, including the NPSA.1 This builds on the taxonomy developed by the Joint Commission on Accreditation of Healthcare Organisations (JCAHO), on behalf of the WHO, which was based on systematic review and expert consensus.74 75 Homogeneous elements of models—which comprise terms and the relationships between terms that make up the building blocks of a classification scheme—were categorised into five complementary root nodes or primary classifications (See Box 4.1). For our review we found the combination of the original JCAHO taxonomy2 and the WHO conceptual map and classification scheme1 102 to be most useful for understanding the potential role of eHealth solutions in facilitating 114 safer patient care, as well as their unintended consequences (See Box 4.1). These differentiate types of problem (communication, patient management, clinical practice), cause (organisational, technical, human), impact (medical and non-medical), issues for prevention (accuracy, communication, alarms) and domain (setting, target, people). In so doing, they recognise that adverse events may result from multiple systemic features operating at different levels, such as the task, the team, the work environment and the organisation.80 They also recognise the importance of psychological and human factors in the nature, mechanisms and causes of error and the fact that liability to error is strongly affected by the context and conditions of work, as has been argued by Leape and others.79 These are further discussed below when combined and integrated with eHealth map in Section 4.5. Box 4.1 Five complementary primary classifications of patient safety developed by the JCAHO 1. Impact: The outcome or effects of medical error and systems failure, commonly referred to as harm to the patient. 2. Type: The implied or visible processes that were faulty or failed. 3. Domain: The characteristics of the setting in which an incident occurred and the type of individuals involved. 4. Cause: The factors and agents that led to an incident. 5. Prevention and mitigation: The measures taken or proposed to reduce incidence and effects of adverse occurrences. 2 Chang et al. (2005) Reproduced with permission from International Journal Quality Health Care 4.3.2 Improving safety of healthcare There are two main approaches to tackling the problem of human fallibility: the person and the system approaches.98 Trying to find the causes for patient safety incidents is often messy and multi-faceted, resisting a clean, simple fix. In essence, approaches to safety incidents centre on examining the chain of events that led to an accident or near miss and considering all acts of those who were in any way involved, and then, crucially, by looking 115 further back at the working conditions of staff and the organisational context of the incident trying to understand why it occurred.103 A number of authors have emphasised the limitations of voluntary reporting by healthcare providers as the principal means for detection of safety incidents.104 For example, among all types of medical errors, cases in which the wrong patient undergoes an invasive procedure warrant special attention. However, such major errors tend to be under-reported.105 As a response to high profile policy reports66;69 that urged the health service to improve patient safety, and in an attempt to facilitate the process of learning from mistakes, the NPSA’s National Reporting and Learning System was created. This, for the first time nationally, encouraged the systematic reporting by healthcare workers, patients and carers of patient safety incidents, including “near misses”, which did not result in actual harm.106 It is recognised that a system, which does not criminalise mistakes107 is needed and the focus on learning from safety incidents.108 There is also a need to learn how to accurately submit information to better prioritise, organise and streamline event analysis.109 Experience from the aviation industry shows that as reporting rises, the number of serious events begins to decline. Paradoxically, in the short term at least, an increase in reporting of patient safety incidents has been a clear sign that the NPSA has been successful in beginning to promote an open and fair culture in which we can all learn from the mistakes of others– i.e. the first steps in the process of creating an ”organisation with a memory”.106 We consider how data from the NRLS have been useful in helping to think about how to reduce drug allergy-related errors in Chapter 12. There have furthermore been a number of other important reports demonstrating clear opportunities from this database, which now contains well over five million entries. Yet reporting systems are far from fully effective in monitoring more serious violations, usually only providing information after a violation has caused some harm.110 New approaches and methods are needed to study violations 116 in healthcare and IT will play a key role in these; for example by monitoring, in real-time, variability in quality—this can identify threats to patient safety111 or through tools, such as event monitoring and natural language processing that can inexpensively detect adverse events in clinical databases.112 Before we describe in more detail how eHealth, quality and safety interact we briefly consider key conceptual considerations in the field of eHealth. 4.4. Defining the field of eHealth Drivers for increased use of IT in healthcare are multiple,113 but can be summed up in one overarching message: the increased pervasiveness and use of IT in all areas of personal, business and public life. eHealth for healthcare delivery is now one of the main priority areas for most developed and indeed developing countries and the stakes for eHealth are high. During the 58th World Health Assembly in 2005, the ministers of health of the 192 member states of the United Nations (UN), approved the so-called eHealth Resolution WHA58.28eHealth (a resolution is the highest legal entity possible in the United Nations system).114 For an extract from the resolution see Box 4.2. The WHO now also runs a Global Observatory for eHealth (http://www.who.int/goe/en/), this underscoring the importance now being attached to technology to help realise the much needed health improvement agenda internationally. Such strong support for eHealth, in spite of scepticism of some agencies and member states, shows global political commitment that has translated into unprecedented investment in IT in healthcare. eHealth is a new term that in a very simple way aims to describe, that something now happens or is assisted by IT, that it happens “electronically”. For example, like “e” before mail to indicate that mail is now sent electronically, rather than by post, eHealth refers to the increasing use of IT to help in the delivery of healthcare. 117 This “e” has however assumed other connotations–for example, notions of modernity and progress, and representing a drive institute a paradigm shift in healthcare that translates into increased efficiency, enhanced quality and safety, and public empowerment and involvement in the pursuit of health.115 The term eHealth has over 50 different definitions.116 It is now widely used in different domains of society from academic117;118 to policy.113;119 Questions remain about how the differing concepts and understandings of the term “eHealth” are used by and impact on different stakeholders.116 Nevertheless, there is a trend to use eHealth in an encompassing way for any use of IT related to health. A recent analysis of the literature suggests the concept is best understood as the application of predominantly networked digital information technologies to support the organisation and delivery of care.3 It thus sits within the broader field known as health informatics, which in addition encompasses the fields of hardware, software development and other technological issues. For this review, we considered it important to use an inclusive definition of eHealth (but which nonetheless framed this as a subset of health informatics as the broader more technological issues were beyond the scope of this enquiry) and thus chose to use Eysenbach’s definition, as adapted by Pagliari, as the basis for our work:115;117 ‘eHealth is an emerging field of medical informatics, referring to the organisation and delivery of health services and information using the Internet and related technologies. In a broader sense, the term characterises not only a technical development, but also a new way of working, an attitude, and a commitment for networked, global thinking, to improve health care locally, regionally, and worldwide by using information and communication technology.’ 118 The conceptual map shown in Figure 4.5 is derived from an extensive review of eHealth research and commentary, conducted for the NHS Service Delivery and Organisation (SDO) Programme by Pagliari et al.3 It represents a synthesised overview of the important application domains within eHealth, set within the context of the broader field of medical (or health) informatics applications. This illustrates the overlap between the two topics (which are often treated as being equivalent), but also emphasises the pervasive role of networked information and communication technologies to eHealth, as distinct from more independent applications of computers and digital devices. In addition, it reveals the overlap with related subfields such as biomedical informatics, eBusiness, eLearning and public health informatics, with the generic areas of computing and telecommunications, and medical equipment. Development of this map drew on the Medical Informatics Scientific Content Map developed by the International Medical Informatics Association120 and structural elements of Medline’s MeSH tree for Medical Informatics. Box 4.2 An excerpt from the United Nations eHealth Resolution ‘The Fifty-eighth World Health Assembly…Noting the potential impact that advances in information and communication technologies could have on health-care delivery, public health, research and health-related activities for the benefit of both low- and high-income countries; Aware that advances in information and communication technologies have raised expectations for health; … Stressing that eHealth is the cost-effective and secure use of information and communications technologies in support of health and health-related fields, including health-care services, health surveillance, health literature, and health education, knowledge and research, 1. URGES Member States: (1) to consider drawing up a long-term strategic plan for developing and implementing eHealth services in the various areas of the health sector, including health administration… (2) to develop the infrastructure for information and communication technologies for health as deemed appropriate to promote equitable, affordable, and universal access 119 to their benefits, and to continue to work with information and telecommunication agencies and other partners in order to reduce costs and make eHealth successful; … (6) to establish national centres and networks of excellence for eHealth best practice, policy coordination, and technical support for health-care delivery, service improvement, information to citizens, capacity building, and surveillance; (7) to consider establishing and implementing national electronic public-health information systems and to improve, by means of information, the capacity for surveillance of, and rapid response to, disease and public-health emergencies’ The core applications of eHealth, according to this map, are summarised as: • Storing and managing data: this refers to electronic health records (EHRs) and records systems, including clinical and administrative data (and stored images), at both the patient and population levels. • Informing and supporting decisions: this includes applications to aid professionals, namely clinical decision support tools and systems, with on-line access to research evidence, guidelines and professional development tools. It also includes consumer health informatics applications such as on-line patient information, decision aids, targeted educational interventions or therapy, personal health records and online peer support. • Providing expertise or care at a distance: this includes, firstly, telemedicine applications, which chiefly concern professional-toprofessional interaction for advice or case conferencing and also remote access to health records and decision support. It also includes telecare applications, which represent an increasingly important part of the eHealth landscape and concern the provision of remote professional-patient communication and support (telehealthcare), as well as the use of technology to promote supported self-care in the home (including self-monitoring, education etc.). Each of these three areas overlaps with the others, to a greater or lesser extent, and there is considerable interplay between the individual 120 applications, reflecting both the complexity of the area and the pervasive theme of networked data and technologies. For example, telecare applications may integrate decision aids and online information, in addition to telemonitoring and remote clinician advice, while the effectiveness of both computerised decision support systems (CDSSs) and telecare is highly dependent on valid electronic health records. Likewise, computerised provider (or physician) order entry (CPOE) is conceptually related to the EHR, since it concerns electronic documentation and data transfer, but it also often integrates CDSS, which may in turn accommodate both prompts for guideline-compliant prescribing, error alerting and or linkage to incident reporting systems. While digital devices, per se, are not within the scope of eHealth, the application of networked devices linked to an EHR is for example. This might include radio-frequency identification (RFID) bracelets linked to patient records and CPOE. Figure 4.5 has been elongated to fit the page and the upper, bold, arrow is designed to illustrate the overlap between the outer two circles. Nevertheless, the display is useful in that it emphasises the central importance of the EHR in maximising the benefits of eHealth applications for the effective organisation and delivery of healthcare. Figure 4.5 also recognises the value of the EHR and related applications for research, disease surveillance and health service planning, as well as the overlap with electronic knowledge repositories, which are critical to the development of effective decision support. 121 Public health information systems Mobile access to records; evidence & CDS Medical conferencing; clinical email Systems biology; genomics Bioinformatics Genetic patient data [Adapted from Pagliari, 2005] Medical tools, Equipment Home & ambulatory disease monitoring & self-management support Patient-provider email; internet consultations Remote interventions (e.g. telepsychiatry, telesurgery) Care Administrative systems e-Business Monitoring/planning (e.g. bed occupancy, stock, workflow, costs,). Transacting (e.g. ordering, billing) Administrative (e.g. for audit, purchasing, billing, tracking service utilisation etc.) Diagnostic or treatment advice from subject experts (e.g. telepathology) Expertise & Knowledge Delivering Expertise & Care at a Distance Pervasive eHealth Theme: Networked Digital Information and Communication Education & Library Sciences eLearning Knowledge repositories e.g. NELH Integrated records, supporting multiple stakeholders Medical images Records Systems Clinical (e.g. for capturing, displaying, sharing, linking or exchanging patient-specific data; or populating decision support) Storing & Managing Data eHealth Medical Informatics Computerised medical equipment e.g. scanners, Use of digital chemical assay tools equipment for patient care vital signs sensors e.g. robotics, bionics, imaging, & monitors, remote disease monitoring, drug dispDigital patient tracking ensers devices linked to (e.g. RFID) clinical networks, EHR, decision support [Described in terms of its applications] Data Patient-specific records, supporting care of individuals. Population-based data, aiding research, policy & planning. Administrative data, aiding organisational & business processes. e.g. e-Management of clinical trials. Use of EHR databases for epidemiological research. GIS for disease surveillance. Health promotion via WWW Supporting Professionals Case-specific diagnostic or treatment advice based on patient data & expert knowledge/evidence Automated prompts & reminders for guideline compliant prescribing, screening or reporting + safety alerts Electronic guidelines, research reports & CME tools Public Health Supporting Patients/Citizens On-line information on health/lifestyle or illness/treatments Tools for balancing risks/benefits/preferences to aid choices Targeted educational interventions Personal Health Records & selfmanagement aids On-line support networks Informing & Supporting Decisions Emerging ICT (egeg WWW, wireless, satellite, DTV, GRID, broadband, GIS, robotics Healthcare IT virtual reality, mobile/ wearable/ systems, hardware, software, implanted devices, pervasive computing, computational networks, data integration tools, PACS, information architectures, specific devices imaging, nanoICT technology, (eg smartcards, mobile phones applications for GPRS PDAs), bar-codes, RFID, NHS-Net etc. health providers, patients & citizens Computing & Telecommunications Figure 4.5 A conceptual map of eHealth 122 4.4.1 Cross-cutting issues for eHealth In addition to defining and elucidating the applications of eHealth, Pagliari et al.’s review for NHS SDO identified a number of cross-cutting issues that reflect common themes of published research and commentary on the topic. These include: • managing change (individual & organisational behaviour, public perception) • integrating appropriate evaluation (tailoring to technology stage & research questions) • optimising human-computer interaction (usability & accessibility) • integrating data and systems (standards & interoperability; data quality & clinical coding) • ethico-legal issues (e.g. data privacy and governance, digital health inequalities) • quality improvement (clinical benefits, patient enablement, efficiency, cost, safety) and • patient safety (reducing error and risks, preventing harm). This review addressed all of these themes, with the exception of ethico-legal issues, for which the focus on systematic reviews of effectiveness is less suitable. Particular attention was paid to quality and safety, human and organisational issues affecting the technology adoption, socio-technical factors affecting the usability of technology, challenges for evaluation and maximising the communicative and integrative power of technologies and data. 4.4.2 NHS Connecting for Health and eHealth Our tracing of NHS Connecting for Health (NHS CFH; see Chapter 3) reveals that it is a highly complex group of sub-programmes, only a fraction of which are represented within the headline deliverables typically cited. It has common strands with major quasi-independent initiatives, such as the National Knowledge Service and NHS Direct Online, while there is considerable internal complexity in the form of multiple design and 123 specification programmes around IT architectures, standards and individual applications; supplier procurement and control mechanisms, and education and user engagement. Chapter 3 maps the key deliverables of NHS CFH and areas of synergy and overlap with other aspects of the National Programme for Information Technology (NPfIT). These elements are now distributed throughout all three main eHealth applications domains, although the primary focus has been on delivering electronic applications that support NHS staff and organisations. A core component of NHS CFH is the NHS Care Records Service (NHS CRS), the primary objective of which was to deliver an integrated EHR, which will in due course support other applications such as appointment booking (Choose and Book) and the electronic transfer of prescriptions (ETP). The EHR is also central to the effective use of CDSSs, which NHS CFH and the National Knowledge Service have been collaborating to develop. The National Knowledge Service and NHS Direct are engaged in knowledge delivery to professionals and patients, NHS Direct is using CDSS algorithms for triage and to support patient self-care support. The transfer of clinical email and images supports the exchange of expertise and clinical care over geographic boundaries. The importance of internetbased data transfer and communications, as facilitated by New National Network (N3), also identifies these programmes with common definitions of eHealth. While the primary focus of NHS CFH has been professional and organisational interventions, a number of elements are consumer-oriented (most obviously HealthSpace). In addition, while telehealthcare was originally outside the remit of the programme, areas such as remote disease monitoring and patient access to GP records are also being considered by the Programme. Patient access to GP records has already been made available independently of NHS CFH by one system supplier (EMIS) through the EMIS Access system (https://www.emisaccess.co.uk/). Figure 4.6 shows the application domains at the centre of the larger eHealth map and is intended to illustrate the areas addressed by the commissioning brief and this report. 124 Figure 4.6 Areas of eHealth prioritised in the research commissioning brief Informing & Supporting Decisions Supporting Patients/Citizens On-line information on health/lifestyle or illness/treatments Tools for balancing risks/benefits/preferences to aid choices Targeted educational interventions Personal Health Records & selfmanagement aids On-line support networks Supporting Professionals Case-specific diagnostic or treatment advice based on patient data & expert knowledge/evidence Automated prompts & reminders for guideline compliant prescribing, screening or reporting + safety alerts Electronic guidelines, research reports & CME tools Storing & Managing Data Data Patient-specific records, supporting care of individuals. Population-based data, aiding research, policy & planning. Administrative data, aiding organisational & business processes. Integrated records, supporting multiple stakeholders Medical images Records Systems Clinical (e.g. for capturing, displaying, sharing, linking or exchanging patient-specific data; or populating decision support) Administrative (e.g. for audit, purchasing, billing, tracking service utilisation etc.) Delivering Expertise & Care at a Distance Expertise & Knowledge Diagnostic or treatment advice from subject experts (e.g. telepathology) Medical conferencing; clinical email Mobile access to records; evidence; CDS Care Patient-provider email; internet consultations Remote interventions (e.g. telepsychiatry, telesurgery) Home & ambulatory disease monitoring & self-management support 125 4.5 Integrating the concepts of quality, safety and eHealth As described previously, the concepts of quality and safety are closely related, and arguments for the power of IT to improve healthcare quality and safety have been well made (although research evidence has yet to catch up with the vision). There is also potential for technology to introduce new risks, which can be rooted in different aspects of the hardware, software or networks (e.g. reliability and usability), the users (including cognitive and psychological factors) or the teams and organisations in which they are used (e.g. safety culture).121 The links between technology, quality and safety are also reflected in the conclusions of the recent SDO eHealth review,3 which summarised the headline benefits as in Box 4.3. Box 4.3 Headline benefits of eHealth, as discussed by SDO Improving healthcare access by • helping to alleviate barriers to effective healthcare introduced by physical location or disability • facilitating consumer empowerment for self-care and health decision-making. Improving healthcare quality by: • aiding evidence-based practice • tailoring care to individuals, where IT enables more informed decision-making based on evidence and patient-specific data • improving transparency and accountability of care processes & facilitating integrated and shared care • reducing errors and increasing safety. Improving healthcare cost-efficiency by • streamlining healthcare processes, reducing waiting times and waste • improving diagnostic accuracy & treatment appropriateness. 3 Source: Pagliari et al. (2005) 126 eHealth applications have the potential to improve the safety of patient care in many ways; for example, by ensuring that records are legible, less likely to be lost and more likely to reflect an accurate patient identity; by improving diagnostic accuracy, supporting evidence-based prescribing and flagging dangers through decision support; by aiding organisational learning through incident reporting systems, or by improving the capacity of patients to mitigate risks through identifying errors and maximising self-care.122 However, introducing new technology also comes with risks attributable to the hardware or software itself or its misuse by healthcare personnel.123;124 Hence, it is appropriate for this project to define the possible risks and safety benefits of eHealth applications and this requires an understanding of the frameworks by which patient safety has been traditionally understood and classified. Our efforts have been aided by multiple sources, including the comprehensive 2001 review of the safety and quality literature by the US Agency for Healthcare Research and Quality (AHRQ).91 Key elements of the safety and quality frameworks, discussed in this chapter, are triangulated with the core application domains from the eHealth map in Figure 4.7, illustrating the relevance of safety considerations to all aspects of technology design, development and deployment. 127 Figure 4.7 Integrated maps of quality, safety and 128 eHealth 4.5.1 Integrated Joint Commission on Accreditation of Healthcare Organizations safety classification scheme and eHealth applications action Below we further integrate earlier introduced (See section 4.4) conceptual maps and classification schemes of safety from JCAHO2 and WHO1;102 with a more detailed description of integration with those of eHealth. Figure 4.8 presents the core concepts within the JCAHO taxonomy, which have been annotated to illustrate issues for eHealth. (More elaborate organisational charts exist, but this level is most easily triangulated with issues for eHealth). 129 Management IT skills. Process redesign. NKS Protocols/Processes e-protocols (e.g. referral), coding & QoF, IT standards e.g. training & supervision around safety Knowledge Transfer eg documentation, incentives, standards e.g. around information sharing or event reporting Culture e.g. staffing/training, safety financing Facilities IT usability, accessibility, dependability, + QI loops e.g. design, malfunction obsolescence SYSTEMS EHR/ e-forms Remote advice for professional; remote patient consultation EHR/Coding Patient websites Promote change & innovation IT planning & procurement Organisational Documentation Consent Process Advice or Interpretation Information quality Communication Negligence Recklessness Intentional rule violations CAUSE Lack of tools/procedures for tracking, auditing, alerting Technical IT skills, ability to recognise good & poor quality info. Use of unsecure ecomms for sensitive info. Knowledge-based (interpretation) Rule-based (retrieval) Skills based (execution) Interpretation of data (images, reports, records) information, evidence, decision support 130 Data extraction Data entry, software use External factors Clinical Decision Support; CPOE, e-prescribing, operating room systems, bar-coding & RFID tagging (e.g. blood, patients), EHR, eLearning, audit Practitioner HUMAN (Error) Post-intervention Inaccurate prognosis Intervention Incorrect procedure enacted Correct procedure omitted Clinical Performance (Practitioner level) Pre-Intervention Questionable diagnosis Patient factors Patient e-comms. e.g. lab, booking, referral. EHR eg e-referral, EHR systems Referral or consultation Tracking or follow-up Patient Management (Organisation level) TYPE OF PROBLEM Figure 4.8: Core features of the JCAHO Patient Safety Taxonomy— annotated to reflect sample eHealth solutions and issues CDS (alerts, alarms, reminders) linked to EHR, supporting by high quality coding. Integrated EHR; better end comms (e.g. lab, referral, 2 opinion) Bar Coding, EHR, NHS Number, Record Linkage, RFID Due to faulty infor-mation, interventions or confidentiality breaches Figure 4.8 continues… Morbidity and mortality due to poor CDSS, ecomms or EHR integrity Confidentiality & negligence suits Poor adoption of technology Cost to modify or replace IT & re-train staff Legal settlements. Lost bed days & waiting times 131 4.5.2 Detailed map of the impact on quality and safety of healthcare by ePrescribing According to Barber et al., errors related to medicines’ management are probably the most common type of medical error in both primary and secondary care in the UK.125 Of all types of medicines’ management errors— prescribing, dispensing, administration, monitoring, repeat prescribing126— prescribing errors are typically the most serious.125 Thus from a combined conceptual map of quality, safety and eHealth here we present in detail how a specific application, namely ePrescribing can proactively and reactively improve quality and safety of prescribing. The International Classification for Patient Safety (ICPS) presents a comprehensive conceptual framework developed to be used in conjunction with other processes, systems and maps (see Figure 4.9). Based on the patient safety and eHealth conceptual frameworks discussed above, we present a list of problems related to medication incidents, contributing factors (staff and environment) and a detection process that may mitigate the process of prescribing, prevent or reduce the likelihood of problems happening, and assist in the recognition of these errors (see Table 4.1). 132 Figure 4.9 should be read together with Table 4.1. Legend: The solid lines enclose the 10 major classes of the ICPS and represent the semantic relationships between them. The dotted lines represent the flow of information. Incident type is medication incident and action taken to reduce risk is use of ePrescribing. National Patient Safety Agency International classification of Patient Safety – permission applied for. 133 Table 4.1 This should be read together with Figure 4.9. It presents problems leading to medication safety incidents, contributing prescriber, communication and environment factors, detection processes and finally theoretical modes of action of ePrescribing application that influence all the former. 134 Medication safety incidents* Wrong Patient Wrong Drug Wrong Dose/Strength of Frequency Wrong Formulation or Presentation Wrong Route Wrong Quantity Contraindication Omitted Medicine or Dose Adverse Drug Reaction Contributing prescriber factors Cognitive factors Perception/understanding Knowledge-based/problem solving Rule-based Slip/lapse error/absentmindedness/ forgetfulness Technical error in execution (physical) Failure to synthesise/act on available information Performance factors Distraction/inattention Fatigue/exhaustion Behavior/violation Non-compliance Routine violation Risky behavior Reckless behavior Problem with substance abuse/Use Sabotage/criminal act Communication factors Paper-based Verbal Contributing environment factors Remote/long distance from service Detection process Error recognition By change in patient’s status By system/environment Change/alarm By a count/audit/review Proactive risk assessment Theoretical modes of action of ePrescribing as a pro-active and reactive application to improve quality and safety of prescribing 1. Improved identification 2. Improved legibility 3. Automatic checks on previous response to the drug 4. Visual or audible alerts and warnings related to drug–drug interactions 5. Automatic checks for individual’s characteristics (e.g. drug-allergies, comorbidities, weight, gender or ethnicity) that may influence the choice of drug, dose, strength of frequency, formulation or presentation, route or quantity 6. Automatic checks for results of clinical investigations (e.g. laboratory values of kidney function tests or blood pressure) that may influence the choice of drug, dose, strength of frequency, formulation or presentation, route or quantity 7. Automatic checks on previous response to the drug 8. Automatic checks on duplicate therapies 9. Automatic check on the amount of medication prescribed at last visit notifying about under and overuse of medication and reducing likelihood of under and overprescribing 10. Automatic barring of too high (dangerous) doses according to patient’s characteristics (single, daily or life dose limits) 11. Advice regarding evidence-based alternative first-choice drug(s) from the same or similar class, drug dose, strength of frequency, formulation or presentation, route or quantity 12. Linking to algorithms emphasising (offering as a first choice when a drug is selected) costeffective drug, dose, strength of frequency, formulation or presentation, route or quantity 13. Advice regarding the cost of medication and cheaper equivalent drug(s) from the same or similar class, dose, strength of frequency, formulation or presentation, route or quantity 14. Reminders about corollary orders 15. Access to information about previously prescribed medication (anytime & anywhere) 16. Reminders on drug guidelines 17. Faster response to clinicians’ prescriptions 18. Reduced patient data re-keying 19. Link to formulary—full information and knowledge-base about the medication prescribed 20. Instant provision of information about formulary-based drug coverage including onformulary alternatives and co-pay information 21. Improved communication amongst prescribers and dispensers (e.g. call back queries, instant reporting that item is out of stock, alerts for unfilled, unrenewed prescriptions) 22. Automatic monitoring of orders and audit 23. Shorter process turn-around time—transit time to dispensing site, time till first dose, prescription renewal or refill 24. Reduction in use of paper 25. Data are available for immediate analysis including post-marketing reporting, drug utilisation review, etc. 135 26. Possibility of remote or long-distance prescribing 27. Electronic access to checklists, protocols and or policies 136 4.5.3 An example of a scenario to illustrate the potential of eHealth to improve the quality and safety of laboratory results reporting As a further example, a detailed analysis of potential theoretical impact of eHealth on patient safety is presented in Table 4.2. This provides a scenario of eHealth’s impact on the quality and safety of clinical processes using the example of laboratory results reporting. 137 Table 4.2 Potential hazards in laboratory result reporting Action Potential hazard Doctor orders • Unnecessary testing (e.g. previous test (either by recent tests not obvious on computerised computer screen). Recent hospital provider order out-patient investigation results not entry (CPOE) or available to GP, no available online paper form, or protocol to advise on appropriate verbally tells testing) patient to get x • May not enter correct patients checked) details, e.g. wrong patient called up on computer scheme ( “similar name”) • Electronic ordering, incorrect test ordered or test left out, patient not informed to go for test, no follow up of uncompleted tests. • Patient misunderstands instruction • Appropriate test does not get done Nurse calls • Wrong patient presents (e.g., patient for test mishears name, nurse does not check name) • “Similar name” problem as above • Misidentified specimen sent to lab Form completed • Wrong labels picked up and applied sample bottles by accident labelled • Wrong labels printed because of similar names • Misidentified specimen sent to lab Blood drawn • Wrong sample bottle used • Sample arrives late at lab and is old resulting in inaccurate analysis • Unsuitable specimen Samples and forms arrive at laboratory • • • • Possible solution Shared patient records (primary and secondary care). Regular integration of results into GP system Clinician education to use patient date of birth for calling up records. Test request routinely recorded on computer and regularly audited against completion Clinician education, patient asked to check details on specimen label Clinician education, patient asked to check details on specimen label Easily available online advice. Computer alarm warning nurse that time for last blood test is passed request routinely Lost samples before reaching lab, Test recorded on computer and clinician unaware that sample has regularly audited against not arrived. completion Lost specimen Error at data entry level if form not Bar-coding of all specimens bar-coded Misidentified sample analysed 138 Table 4.2 Potential hazards in laboratory result reporting continued… Action Analysis performed. Result verified Posting of laboratory results—printed, emailed or webbased. Results pulled into practice computer system, scanned and entered manually, or entered automatically from web by software Results checked by clinician, on paper, scanned document or electronic entry Potential hazard • Lab-based calibration, poor reagents, etc. Possible solution errors—e.g. storage of • • • • Test request routinely recorded on computer and regularly audited against completion. Clinician education in mail box hygiene Patient asked to check for result • Results entered into wrong Administration education. Use of date of birth/NHS patient’s records • Result unavailable for number for data entry. Test request routinely intended patient • Patient mistreated because of recorded on computer and regularly audited erroneous result against completion. Printed lab result goes missing email not noticed Mailbox full Wrong clinician/surgery gets results (delay) • Confidentiality risk • Clinician does not read mail (e.g. on holiday no alternative provision made) • Abnormal result not noticed, due to lack of knowledge, poor highlighting of abnormal results, fatigue, information overload, multiple presentation of different lab results and set overlooked • Abnormal result missed Clinician/admin education. Regular check on mail boxes (especially holiday/illness) to ensure completion. Double checking of all “normal” results Presentation of tests individually on screen rather than layered 139 Table 4.2 Potential hazards in laboratory result reporting continued… Action Potential hazard Possible solution Patient informed Test request routinely • Patient does not call for result • Patient is told some results recorded on computer normal, not aware that further and regularly audited completion. potentially abnormal results are against Patient informed is part of to follow trail to be • Patient is told wrong result audit (misidentification, wrong record completed Clinician education on called up) of date of • Confidentiality risk, by phone , use birth/postcode/password post or in person • Patient does not get message before giving information Follow up of abnormal result • • • • Integration with computer alerting systems • • (phone/email/letter about abnormal result) Patient does not understand significance of result and does not follow-up as intended Patient does not attend followup of abnormal result, system not in place to detect this Patient attends follow-up. One abnormal result noted by clinician but does not notice second abnormality Important result not followed up Result scanned or added as free text, not in searchable form detectable by computer alerting system Inappropriate drug prescribed despite biochemical, abnormality. Patient action required (e.g. send for) logged in computer. Regular audit to check attendance when sent for. All abnormal results highlighted in journal text Clinician/administration education 140 4.6 Conclusions There are clear parallels in the concepts of quality, safety and eHealth. Each of these areas has suffered from variability in definitions, epistemologies and terminologies, which represent obstacles to the synthesis of existing research. Effectiveness is a core construct of the quality concept, which also accommodates efficiency (and cost), access, equity and accessibility. Prevention of harm and identification of error and risks lie at the centre of the safety concept, and human and organisational factors (including safety culture) are seen as important contributory factors. Networked data and communications are a pervasive theme within the concept of eHealth and EHRs a core application area. Tools for informing and supporting decisions, and aiding clinical practice and patient care at a distance are also central eHealth domains. The multiple objectives of NHS CFH overlap with those of several other NHS technology programmes; for example, NHS Direct. The primary focus of NHS CFH is the delivery of networks and tools for supporting healthcare professionals and organisations; however, elements of the Programme are consumer-oriented and there have been recent moves towards integrating remote healthcare within the scope of the programme. Nevertheless, the scope of this project’s commissioning brief is firmly located within the former domain. Evidence of effectiveness is crucial for motivating the adoption of new eHealth applications by clinicians and a primary objective of the review is to synthesise available evidence on the outcomes of eHealth interventions and identify best practice implementation and adoption strategies. References 1. World Health Organization. International Classification for Patient Safety. Version 1.0 for Use in Field Testing 2007 - 2008. Geneva: WHO. 2. Chang A, Schyve PM, Croteau RJ, O'Leary DS, Loeb JM. The JCAHO patient safety event taxonomy: a standardized terminology and classification schema for near misses and adverse events. Int J Qual Health Care 2005; 17: 95-105. 3. Pagliari C. Literature review and conceptual map of the field of eHealth. Final report to the SDO programme, 2004. 4. Editor. Streptomycin in pulmonary tuberculosis (Reprinted from BMJ 1948:ii:790-1). BMJ 1998; 317: 1248. 5. Editor. Fifty years of randomised controlled trials. BMJ 1998; 317. 6. Grol R. Improving the quality of medical care: Building bridges among professional pride, payer profit, and patient satisfaction. JAMA 2001; 286: 2578-85. 141 7. Institute of Medicine. To Err Is Human: Building a Safer Health System. Washington, DC, National Academy Press, 2000. 8. Institute of Medicine. Crossing the Quality Chasm: A New Health System for the 21st Century. Washington, D.C., National Academy Press, 2001. 9. Institute of Medicine. Patient Safety: Achieving a New Standard for Care. Washington, D.C., National Academy Press, 2003. 10. Institute of Medicine. Preventing Medication Errors. Washington, D.C., National Academy Press, 2007. 11. McGlynn EA, Asch SM, Adams J, Keesey J, Hicks J, DeCristofaro A et al. The Quality of Health Care Delivered to Adults in the United States. N Engl J Med 2003; 348: 2635-45. 12. Tudor Hart J. The inverse care law. Lancet 1971;297:405-412. 13. Watt G. The inverse care law today. Lancet 2002;360:252-254. 14. Spiegelhalter DJ. Surgical Audit: Statistical lessons from Nightingale and Codman. J R Stat Soc Ser A Stat Soc 1999;162: 45-58. 15. Donabedian A. Evaluating the quality of medical care. Milbank Mem Fund Q 1966; 44: 166–203. 16. Best M, Neuhauser D. Avedis Donabedian: father of quality assurance and poet. Qual Saf Health Care 2004; 13: 472-73. 17. Donabedian A. The quality of medical care. Science 1978; 200: 856-64. 18. Blumenthal D. Quality of care -- What is it?- Part one of six. N Engl J Med 1996; 335: 891-94. 19. Brennan TA, Gawande A, Thomas E, Studdert D. Accidental deaths, saved lives, and improved quality. N Engl J Med 2005; 353: 1405-09. 20. Maxwell RJ. Quality assessment in health. BMJ 1984; 288: 1470-72. 21. Howie JGR, Heaney D, Maxwell M. Quality, core values and the general practice consultation: issues of definition, measurement and delivery. Fam Pract 2004; 21: 458-68. 22. Berwick DM. Quality comes home. Ann Intern Med 1996; 125: 839-43. 23. Schuster MA, McGlynn EA, Brook RH. How good is the quality of health care in the United States? Milbank Q 1998; 76: 517-63. 24. Engels Y, Campbell S, Dautzenberg M, van den Hombergh P, Brinkmann H, Szecsenyi J et al. Developing a framework of, and quality indicators for, general practice management in Europe. Fam Pract 2005; 22: 215-22. 25. Campbell SM, Roland MO, Buetow SA. Defining quality of care. Soc Sci Med 2000; 51: 1611-25. 26. Berwick DM. Errors today and errors tomorrow. N Engl J Med 2003; 348: 2570-72. 27. Starfield B. New paradigms for quality in primary care. Br J Gen Pract 2001; 51: 30309. 28. Grol R, Baker R, Moss F. Quality improvement research: understanding the science of change in health care. Qual Saf Health Care 2002; 11: 110-11. 29. Brook RH, McGlynn EA, Shekelle PG. Defining and measuring quality of care: a perspective from US researchers. Int J Qual Health Care 2000; 12: 281-95. 30. Reeves D, Campbell SM, Adams J, Shekelle PG, Kontopantelis E, Roland MO. Combining multiple indicators of clinical quality: an evaluation of different analytic approaches. Med Care 2007; 45: 489-96. 31. Pronovost PJ, Berenholtz SM, Needham DM. A framework for health care organizations to develop and evaluate a safety scorecard. JAMA 2007; 298: 2063-65. 32. Lilford R, Mohammed MA, Spiegelhalter D, Thomson R. Use and misuse of process and outcome data in managing performance of acute medical care: avoiding institutional stigma. Lancet 2004; 363:1147-54. 33. Kassirer JP. The quality of care and the quality of measuring it. N Engl J Med 1993; 329: 1263-65. 34. Lilford R, Girling A, Stevens A, Almasri A, Mohammed MA, Braunholtz D. Adjusting for treatment refusal in rationing decisions. BMJ 2006; 332: 542-44. 142 35. Kerr EA, Hofer TP, Hayward RA, Adams JL, Hogan MM, McGlynn EA et al. Quality by any other name?: A comparison of three profiling systems for assessing health care quality. Health Serv Res 2007; 42: 2070-87. 36. Pugh MJV, Berlowitz DR, Montouris G, Bokhour B, Cramer JA, Bohm V et al. What constitutes high quality of care for adults with epilepsy? Neurology 2007; 69: 202027. 37. Higashi T, Wenger NS, Adams JL, Fung C, Roland M, McGlynn EA et al. Relationship between Number of Medical Conditions and Quality of Care. N Engl J Med 2007; 356: 2496-04. 38. Bottle A, Gnani S, Saxena S, Aylin P, Mainous AG, III, Majeed A. Association Between quality of primary care and hospitalization for coronary heart disease in England: national cross-sectional study. J Gen Intern Med 2008; 23: 135-41. 39. Calderón-Larrañaga A, Carney L, Soljak M, Bottle A, Partridge M, Bell D, et al. Association of population and primary healthcare factors with hospital admission rates for chronic obstructive pulmonary disease in England: national cross-sectional study. Thorax 2010 ePub ahead of print 40. Lilford R, Pronovost P. Using hospital mortality rates to judge hospital performance: a bad idea that just won't go away. BMJ 2010; 340: c2016. 41. Halm EA, Lee C, Chassin MR. Is volume related to outcome in health care? A systematic review and methodologic critique of the literature. Ann Intern Med 2002; 137: 511-20. 42. Pitches D, Mohammed M, Lilford R. What is the empirical evidence that hospitals with higher-risk adjusted mortality rates provide poorer quality care? A systematic review of the literature. BMC Health Serv Res 2007; 7: 91. 43. Millett C, Car J, Eldred D, Khunti K, Mainous AG, III, Majeed A. Diabetes prevalence, process of care and outcomes in relation to practice size, caseload and deprivation: national cross-sectional study in primary care. J R Soc Med 2007; 100: 275-83. 44. Saxena S, Car J, Eldred D, Soljak M, Majeed A. Practice size, caseload, deprivation and quality of care of patients with coronary heart disease, hypertension and stroke in primary care: national cross-sectional study. BMC Health Serv Res 2007; 7: 96. 45. Heath I, Hippisley-Cox J, Smeeth L. Measuring performance and missing the point? BMJ 2007; 335: 1075-76. 46. McDonald R, Roland M. Pay for performance in primary care in England and California: comparison of unintended consequences. Ann Fam Med 2009; 7: 121-27. 47. Alshamsan R, Majeed A, Ashworth M, Car J, Millett C. Impact of pay for performance on inequalities in health care: systematic review. J Health Serv Res Policy 2010; 15: 178-84. 48. Pear R. A.M.A. to Develop Measure of Quality of Medical Care. New York Times 2006 Feb 21. 49. Hodgson DC, Brierley JD, Cernat G, Bondy S, Slaughter PM, Pinfold SP et al. The consistency of panelists' appropriateness ratings: do experts produce clinically logical scores for rectal cancer treatment? Health Policy 2005; 71: 57-65. 50. Naylor CD. What is Appropriate Care? N Engl J Med 1998; 338: 1918-20. 51. Anderson GM, Brown AD. Appropriateness ratings: overuse, underuse, or misuse? Lancet 2001; 358: 1475-76. 52. Marshall MN, Shekelle PG, McGlynn EA, Campbell S, Brook RH, Roland MO. Can health care quality indicators be transferred between countries? Qual Saf Health Care 2003; 12: 8-12. 53. Sheldon TA, Cullum N, Dawson D, Lankshear A, Lowson K, Watt I et al. What's the evidence that NICE guidance has been implemented? Results from a national evaluation using time series analysis, audit of patients' notes, and interviews. BMJ 2004; 329: 999. 54. Soumerai SB, McLaughlin TJ, Gurwitz JH, Guadagnoli E, Hauptman PJ, Borbas C et al. Effect of local medical opinion leaders on quality of care for acute myocardial infarction: a randomized controlled trial. JAMA 1998; 279: 1358-63. 143 55. Ferguson TB, Jr., Peterson ED, Coombs LP, Eiken MC, Carey ML, Grover FL et al. Use of continuous quality improvement to increase use of process measures in patients undergoing coronary artery bypass graft surgery: a randomized controlled trial. JAMA 2003; 290: 49-56. 56. Peck C, McCall M, McLaren B, Rotem T. Continuing medical education and continuing professional development: international comparisons. BMJ 2000; 320: 432-35. 57. Colton D. Quality Improvement in health care: conceptual and historical foundations. Eval Health Prof 2000; 23: 7-42. 58. Chassin MR. Assessing strategies for quality improvement. Health Aff (Millwood) 1997; 16: 151-61. 59. Jamtvedt G, Young JM, Kristoffersen DT, O'Brien MA, Oxman AD. Does telling people what they have been doing change what they do? A systematic review of the effects of audit and feedback. Qual Saf Health Care 2006; 15: 433-36. 60. Marshall MN, Shekelle PG, Leatherman S, Brook RH. The public release of performance data: what do we expect to gain? a review of the evidence. JAMA 2000; 283: 1866-74. 61. Werner RM, Asch DA. The unintended consequences of publicly reporting quality information. JAMA 2005; 293: 1239-44. 62. Campbell SM, Reeves D, Kontopantelis E, Sibbald B, Roland M. Effects of pay for performance on the quality of primary care in England. N Engl J Med 2009; 361: 36878. 63. Berwick DM, Calkins DR, McCannon CJ, Hackbarth AD. The 100 000 Lives Campaign: Setting a goal and a deadline for improving health care quality. JAMA 2006; 295: 324-27. 64. McCannon CJ, Hackbarth AD, Griffin FA. Miles to go: an introduction to the 5 Million Lives Campaign. Jt Comm J Qual Patient Saf 2007; 33: 477-84. 65. Brown CA, Belfield CR, Field SJ. Cost effectiveness of continuing professional development in health care: a critical review of the evidence. BMJ 2002; 324: 652-55. 66. Department of Health. An Organisation with a Memory: Report of an expert group of learning from adverse events in the NHS. London: HMSO, 2000. 67. Aylin P, Tanna S, Bottle A, Jarman B. How often are adverse events reported in English hospital statistics? BMJ 2004; 329: 369. 68. Blendon RJ, DesRoches CM, Brodie M, Benson JM, Rosen AB, Schneider E et al. Views of practicing physicians and the public on medical errors. N Engl J Med 2002; 347: 1933-40. 69. Department of Health, Chief Pharmaceutical Officer. Building a Safer NHS for Patients: improving medication safety. London: HMSO, 2004. 70. Shojania KG, Duncan BW, McDonald KM, Wachter RM, Markowitz AJ. Making Health Care Safer: a critical analysis of patient safety practices. AHRQ Publication No. 01-E058. 2001. Rockville, MD, Agency for Healthcare Research and Quality (AHRQ). 71. Leape LL. Reporting of medical errors: time for a reality check. Qual Saf Health Care 2000; 9: 144-45. 72. Elder NC. Classification of medical errors and preventable adverse events in primary care: a synthesis of the literature. J Fam Pract 2002;51:927-932. 73. Dovey SM, Meyers DS, Phillips RL, Jr., Green LA, Fryer GE, Galliher JM et al. A preliminary taxonomy of medical errors in family practice. Qual Saf Health Care 2002; 11: 233-38. 74. Chang A, Schyve PM, Croteau RJ, O'Leary DS, Loeb JM. The JCAHO patient safety event taxonomy: a standardized terminology and classification schema for near misses and adverse events. Int J Qual Health Care 2005; 17: 95-105. 75. Runciman WB, Baker GR, Michel P, Dovey S, Lilford RJ, Jensen N, et al. Tracing the foundations of a conceptual framework for a patient safety ontology. Qual Saf Health Care 2010; 19: e56. 144 76. Runciman WB, Helps SC, Sexton EJ, Malpass A. A classification for incidents and accidents in the health-care system. J Qual Clin Pract 1998; 18: 199-211. 77. Hurwitz B, Sheikh A (eds.). Health Care Errors and Patient Safety. London: WileyBMJ, 2009. 78. Vincent C, Taylor-Adams S, Chapman EJ, Hewett D, Prior S, Strange P et al. How to investigate and analyse clinical incidents: Clinical Risk Unit and Association of Litigation and Risk Management protocol. BMJ 2000; 320: 777-81. 79. Leape LL. Error in medicine. JAMA 1994; 272:1851-57. 80. Vincent C, Taylor-Adams S, Stanhope N. Framework for analysing risk and safety in clinical medicine. BMJ 1998;316:1154-1157. 81. Nebeker JR, Barach P, Samore MH. Clarifying Adverse Drug Events: A clinician's guide to terminology, documentation, and reporting. Ann Intern Med 2004; 140: 795801. 82. Tamuz M, Thomas EJ, Franchois KE. Defining and classifying medical error: lessons for patient safety reporting systems. Qual Saf Health Care 2004; 13: 13-20. 83. Elder N, Pallerla H, Regan S. What do family physicians consider an error? A comparison of definitions and physician perception. BMC Fam Pract 2006; 7: 73. 84. Perneger T. The Swiss cheese model of safety incidents: are there holes in the metaphor? BMC Health Serv Res 2005; 5: 71. 85. Sutcliffe KM. Defining and classifying medical error: lessons for learning. Qual Saf Health Care 2004; 13: 8-9. 86. McLoughlin V, Millar J, Mattke S, Franca M, Jonsson PM, Somekh D et al. Selecting indicators for patient safety at the health system level in OECD countries. Int J Qual Health Care 2006; 18: 14-20. 87. Veazie PJ. An individual-based framework for the study of medical error. Int J Qual Health Care 2006; 18: 314-19. 88. Leape LL, Berwick DM, Bates DW. What practices will most improve safety?: evidence-based medicine meets patient safety. JAMA 2002; 288: 501-07. 89. Shojania KG, Duncan BW, McDonald KM, Wachter RM. Safe but sound: patient safety meets evidence-based medicine. JAMA 2002; 288: 508-13. 90. Forster AJ, Shojania KG, van Walraven C. Improving patient safety: moving beyond the "hype" of medical errors. Can Med Assoc J 2005; 173: 893-94. 91. Shojania KG, Duncan BW, McDonald KM, Wachter RM, Markowitz AJ. Making health care safer: a critical analysis of patient safety practices. Evid Rep Technol Assess (Summ ) 2001; i-668. 92. Battles JB, Lilford RJ. Organizing patient safety research to identify risks and hazards. Qual Saf Health Care 2003; 12: ii2-ii7. 93. Boxwala AA, Dierks M, Keenan M, Jackson S, Hanscom R, Bates DW et al. Organization and representation of patient safety data: current status and issues around generalizability and scalability. J Am Med Inform Assoc 2004; 11: 468-78. 94. Wiegmann DA. Human error analysis of commercial aviation accidents: application of the Human Factors Analysis and Classification system (HFACS). Aviat Space Environ Med 2001; 72: 1006-16. 95. Helmreich RL. On error management: lessons from aviation. BMJ 2000; 320: 781-85. 96. Wilf-Miron R, Lewenhoff I, Benyamini Z, Aviram A. From aviation to medicine: applying concepts of aviation safety to risk management in ambulatory care. Qual Saf Health Care 2003; 12: 35-39. 97. Reason J. Human Error. Cambridge: Cambridge University Press, 1990. 98. Reason J. Human error: models and management. BMJ 2000; 320: 768-70. 99. Panesar SS, Cleary K, Sheikh A. Reflections on the National Patient Safety Agency's database of medical errors. J R Soc Med 2009; 102: 256-58. 100. Vincent C, Taylor-Adams S, Stanhope N. Framework for analysing risk and safety in clinical medicine. BMJ 1998; 316: 1154-57. 101. World Health Organization. Towards a Common International Understanding of Patient Safety Concepts and Terms: Taxonomy and Terminology Related to Medical 145 102. 103. 104. 105. 106. 107. 108. 109. 110. 111. 112. 113. 114. 115. 116. 117. 118. 119. 120. 121. 122. 123. 124. 125. 126. Errors and System Failures. Geneva: World Health Organization, Department of Health Service Provision, 2003. World Health Organization. More than words - International Classification for Patient Safety. Geneva: WHO, 2007. Pringle M. Significant event auditing and root cause analysis. In: Hurwitz B, Sheikh A (eds.). Health Care Errors and Patient Safety. London: Wiley-BMJ, 2009: 193-206. Vincent C. Incident reporting and patient safety. BMJ 2007; 334: 51. Chassin MR, Becher EC. The Wrong Patient. Ann Intern Med 2002; 136: 826-33. Stephenson T. The National Patient Safety Agency. Arch Dis Child 2005; 90: 226-28. Holbrook J. The criminalisation of fatal medical mistakes. BMJ 2003; 327: 1118-19. Pierson S, Hansen R, Greene S, Williams C, Akers R, Jonsson M et al. Preventing medication errors in long-term care: results and evaluation of a large scale webbased error reporting system. Qual Saf Health Care 2007; 16: 297-302. Tuttle D, Holloway R, Baird T, Sheehan B, Skelton WK. Electronic reporting to improve patient safety. Qual Saf Health Care 2004; 13: 281-86. Amalberti R, Vincent C, Auroy Y, de Saint Maurice G. Violations and migrations in health care: a framework for understanding and management. Qual Saf Health Care 2006; 15: i66-i71. Harley M, Mohammed MA, Hussain S, Yates J, Almasri A. Was Rodney Ledward a statistical outlier? Retrospective analysis using routine hospital data to identify gynaecologists' performance. BMJ 2005; 330: 929. Bates DW, Evans RS, Murff H, Stetson PD, Pizziferri L, Hripcsak G. Detecting adverse events using information technology. J Am Med Inform Assoc 2003; 10: 11528. Wyatt JC, Sullivan F. eHealth and the future: promise or peril? BMJ 2005; 331: 139193. Healy JC. The WHO eHealth Resolution - eHealth for all by 2015? Methods Inf Med 2007; 46: 2-4. Eysenbach G. What is e-health? J Med Internet Res 2001; 3: E20. Oh H, Rizo C, Enkin M, Jadad A. What is eHealth (3): a systematic review of published definitions. J Med Internet Res 2005; 7: e1. Pagliari C, Sloan D, Gregor P, Sullivan F, Detmer D, Kahan JP et al. What is eHealth (4): a scoping exercise to map the field. J Med Internet Res 2005; 7: e9. Jones R, Rogers R, Roberts J, Callaghan L, Lindsey L, Campbell J et al. What is eHealth (5): a research agenda for eHealth through stakeholder consultation and policy context review. J Med Internet Res 2005; 7: e54. Watson R. EU wants every member to develop a "roadmap" for ehealth. BMJ 2004; 328: 1155. Lorenzi N. Medical informatics scientific content map. Available from: http://www.imia.org/endorsed/2002_scientific_map.lasso (last accessed 3/1/11) Pronovost P, Sexton B. Assessing safety culture: guidelines and recommendations. Qual Saf Health Care 2005; 14: 231-33. Bates DW, Gawande AA. Improving safety with information technology. N Engl J Med 2003; 348: 2526-34. Ash JS, Berg M, Coiera E. Some unintended consequences of information technology in health care: the nature of patient care information system-related errors. J Am Med Inform Assoc 2004; 11: 104-12. Ash JS, Sittig DF, Campbell E, Guappone K, Dykstra RH. An unintended consequence of cpoe implementation: shifts in power, control, and autonomy. AMIA Annu Symp Proc 2006; 11-15. Barber N, Rawlins M, Dean FB. Reducing prescribing error: competence, control, and culture. Qual Saf Health Care 2003; 12: i29-i32. Avery AJ, Sheikh A, Hurwitz B, Smeaton L, Chen YF, Howard R et al. Safer medicines management in primary care. Br J Gen Pract 2002; 52: S17-S22. 146 SECTION 3 FINDINGS 147 Descriptive overview of results Our initial searches revealed a total of 46,349 potentially eligible articles of which 67 were relevant systematic reviews. During the second phase of work, which included an expansion into the field of telehealthcare, we identified a further 95 relevant systematic reviews, bringing the total number to 162 systematic reviews (see Appendix 3). These are listed in Appendix 4 and a critique of each individual review is provided in Appendix 5. The chapters in this section (Chapters 5-17) draw in the main on evidence from the systematic reviews, which have been assessed using an adapted version of the Critical Appraisal Skills Programme criteria (see Executive summary and Appendix 2), but we also, where necessary, draw on evidence from a broader body of trial, technical, descriptive, qualitative and policy relevant work, all of which is referenced in individual chapters, to help contextualise our findings. 148 Chapter 5 Health information exchange and interoperability Summary • Effective and efficient sharing of clinical information is essential to the future coherent development of health systems, which increasingly deliver care to individual patients through an array of primary and secondary care providers working in a range of locations. • The ideal in this respect is for professionals–and also patients if they so wish–to have the ability simultaneously to access and seamlessly transfer, contribute to and integrate clinical data from disparate sources. • Health information exchange and interoperability considerations aim to provide a shared platform and syntax by which different types of systems can safely and effectively communicate healthcare data. • Whilst there is as yet only limited research on the benefits and risks of health information exchange and interoperability, investment in this infrastructure is probably the key to realising economic returns associated with other more directly patient relevant eHealth applications such as electronic health record systems and ePrescribing. • Additional benefits will be gained through enabling reuse of coded data for health planning, public health, research and audit. • Whilst increasing levels of interoperability are clearly desirable for many reasons, greater centralised storage of sensitive data also inevitably increases the risk of threats to data security and inadvertent breaches of confidentiality. • Currently, most UK healthcare settings are characterised by relatively low levels of interoperability capability, this being particularly true of the hospital sector, where paper-based records are still widely used. • There is increasing recognition for the need to minimise the risks and inefficiencies to care provision associated with the introduction of technology that fails to integrate with existing IT systems. 149 Summary continued… • The NHS has already begun to increase the potential for health information exchange and interoperability through, for example, systems such as LabLinks that send laboratory results electronically to general practices, electronic transfer of discharge summaries and clinic letters, the National Network for the NHS, the central Spine and associated NHS Care Records Service; other developments include encouragement of common operational standards such as Health Level Seven and co-operating with the Continua Health Alliance. • Key outstanding issues that face healthcare systems in realising the potential for seamless exchange of information include the need to develop and deploy standard coding structures across all care settings (e.g. using Systematized Nomenclature of Medicine-Clinical Terms), facilitate integration of the increasing amounts of patient-generated data (through HealthSpace, home sensors and telemetry devices, for example), ensure that minimum standards of interoperability are implemented across the increasingly decentralised NHS and improve secure audited access to electronic health records to minimise the risks of threats to privacy. • Making progress in optimising the ability to share and integrate data seems particularly relevant given the expressed government objective of creating an “information revolution” within the NHS. 5.1 Introduction Patients in most economically developed countries now commonly receive their healthcare from a range of practitioners, many of whom work in different healthcare settings. For most people, this will include health centres, pharmacy practices, dentists, and opticians, a variety of hospitals and laboratory services, and occasionally ambulance services. Guided self-monitoring through teleheathcare may in the future also play an increasingly important role in care provision (see Chapter 13). 150 All these services and systems hold electronic records for patients (and with electronic self-monitoring the patients hold electronic information themselves). However, such records are largely held in silos with limited or no ability to interact with one another and little prospect for direct access by the patients and their clinicians to the bulk of these records. This situation can lead to wasted resources through, for example, double entry of demographic data and, duplication of test requests, and increases the risk of errors through delayed diagnosis, lost data and inaccurate data entry. Effective and efficient sharing of clinical information is therefore essential to the future development of modern health systems. This chapter aims to provide an overview of health systems characterised by high levels of health information exchange and interoperability (HIEI) capability, review the benefits that are likely to accrue from such developments and the possible risks associated with prompting system change along these lines. The chapter is not intended to be a detailed discussion of the technical infrastructures and ontologies which underpin HIEI (for which interested readers are referred to Coiera’s introduction1 or the review by Tanebaum2), but rather to provide a basis for considering the key applications being introduced into the NHS, which are considered in detail in subsequent chapters in this Section. 5.2 Definition, description and scope 5.2.1 Definition Health information exchange and interoperability refers to the ability to access, contribute to and integrate data from disparate sources. As defined by the Health Information Security and Privacy Collaboration, it refers to:3 ‘The ability of systems or components to exchange health information and to use the information that has been exchanged accurately, securely, and verifiably, when and where needed.’ 151 5.2.2 Description Health information exchange and interoperability allows clinicians and patients to access necessary parts of the patient record. It does this by providing a shared platform and syntax by which different types of systems can exchange data seamlessly. However, the degree to which different systems can interact varies. Walker et al. provide a helpful conceptually driven analytic framework and taxonomy for HIEI, which is illustrated in Box 5.1.4 This four level taxonomy reflects the amount of human involvement required, the sophistication of information technology (IT) and the level of standardisation needed for different degrees of information sharing capability between healthcare organisations.4 Box 5.1 A conceptual analytic framework and taxonomy for health information exchange and interoperability • Level 1: Non-electronic data—no use of IT to share information (for example, ‘snail’ mail and meetings). • Level 2: Machine-transportable data—transmission of non-standardised information via basic IT; information within the document cannot be electronically manipulated (for example, fax or personal computer [PC]–based exchange of scanned documents, pictures, or portable document format [PDF] files). • Level 3: Machine-organisable data—transmission of structured messages containing non-standardised data; requires interfaces that can translate incoming data from the sending organisation’s vocabulary to the receiving organisation’s vocabulary; usually results in imperfect translations because of vocabularies’ incompatible levels of detail (for example, email of free text, or PC-based exchange of files in incompatible/proprietary file formats, HL-7 messages (see Glossary). • Level 4: Machine-interpretable data—transmission of structured messages containing standardised and coded data; idealised state in which all systems exchange information using the same formats and vocabularies (for example, automated exchange of coded results from an external lab into a provider’s EHR automated exchange of a patient’s ‘problem list’). 4 Adapted from: Walker et al. (2005) Permission Requested 5.2.3 Scope These considerations are clearly cross-cutting and of relevance to almost any eHealth application, irrespective of whether this is used for storage, decision support 152 or support of remote delivery of care. Given the historic low-level of interoperability between existing information systems in the NHS, there is the risk that this situation will be compounded if due attention is not given to in the context of introducing new applications. Using this framework, it is clear that most systems in England will, despite the work of NHS Connecting for Health (NHS CFH) and the introduction of the key National Programme for Information Technology (NPfIT) deliverables (see Chapter 3), for the foreseeable future be operating at Levels 1, 2 or 3. For example: • Transfer of records between GPs: Even though UK primary care has a long history of using the electronic health record (EHR), changing general practitioners (GPs) has typically necessitated the printing out of the patients’ record and the manual entry of a summary of this record into the new GP’s system, this cumbersome process resulting in the loss of considerable amounts of data (Level 1). Currently, an average general practice in England deals with around 500 patient record transfers each year; inner-city and university practices will deal with far more. Re-entering data creates a huge administrative burden that raises the possibility of omitting important information and transcription errors. More recently, for most practices certain types of attachment (scanned files) can now be transferred, but these cannot be easily integrated with different GP software (Level 2). As discussed in Chapter 3, GP2GP, which permits communication between GP systems, is now being rolled-out across England. The exchange takes place remotely and automatically on completion of registration. This means that patients’ records, including, for example, full diabetes and blood pressure records, are available within hours of registration and as a result many hours, and in some cases, days of repeat data entry are avoided (Level 3). There are, however, some problems that still need to be addressed, these including the fact that the differences between the systems necessitates data “cleaning”; there is furthermore the need to reorganise data, particularly when this involves transfer of data between computer systems, but even when transfers occur between practices that use the same computers systems if they do not organise their data in exactly the same way. 153 • Choose and Book: This eBooking system permits GPs and their patients to access hospital appointment systems and make appointments at times to suit them with the hospital of their choice (Level 3). • NHS Care Record Service (NHS CRS): The aim is that this will eventually replace the current mix of electronic and paper records held throughout the NHS allowing immediate access, with permission, to some aspects of the patient’s record (the Summary Care Record (SCR)) from any part of the country and more detailed clinical, laboratory and radiology records (Detailed Care Record (DCR)) to local providers. This will in time allow the integration of decision support functionality (Levels 3-4; see Chapters 3 and 6 for further details). • Picture Archiving and Communications System (PACS): Until recently, radiological images were printed on film and held in hospital records departments. If patients needed to be seen at other locations (e.g. for a specialist assessment), images had to be physically transported with many not arriving in time, or being lost in transit, and resulting in repeat investigations or delay in diagnosis. The PACS, which has been widely implemented by NHS CFH, now allows access to digital centrally stored images within and between hospitals and eventually should permit full interoperability and compatibility with other NPfIT services. It permits, for example, a doctor in an accident and emergency department to access specialist opinion in another trust and images will, it is planned, also be accessible in some primary care settings. The aim is that in due course, PACS will integrate with the NHS CRS, thereby removing the cumbersome barrier between images and other aspects of the patient record (Level 3-4). • ePrescribing and the Electronic Prescription Service (EPS): For most patients, prescriptions are still generated by computer in their practice, printed on paper and then taken to pharmacies where the same information is entered into the pharmacy computing system to print labels and maintain a database of their dispensing activity. The paper prescriptions are then sent on to the pricing bureau to be again computer-entered for payment (Level 1). The Electronic Prescription Service being implemented by NHS CFH should allow for the transmission of prescriptions from GPs and other prescribers 154 directly to pharmacies, thus automatically populating their databases for dispensing. Prescriptions are then also sent electronically by pharmacists to the Prescription Pricing Authority (PPA), the organisation that reimburses dispensers, i.e. community pharmacies for the medication they have supplied to patients. The system provides a robust audit trail of prescribing activity, reduces administrative effort and may reduce errors of transcription. Such ePrescribing systems have the potential to decrease delay in the dispensing of orders, reduce errors related to handwriting or transcription, allow order entry on or off-site, error checking for duplicate or incorrect doses, and facilitate inventory and posting of charges (Level 3; see Chapter 10). • Laboratory results: While laboratory data are digitally recorded and stored, results are often printed out on paper and sent to those ordering the tests. In many cases, these results are then scanned and entered into an electronic record; the most relevant data are still in many cases still then manually entered into, for example, the GP patient records (Level 1). Recent developments have provided access to all patients’ laboratory results across several hospitals and practices (SCI store in Scotland)5 (Level 2) and results for many practices in other parts of the UK may now be downloaded, for example via LabLinks,6 directly into practice computer systems with occasional need for modification (Level 3). 5.3 Theoretical benefits and risks 5.3.1 Benefits The examples in the preceding section illustrate the types of benefits that more integrated computer systems can yield. Walker et al., in a review of the potential of HIEI to improve services in the United States (US),4 have summarised the broad range of benefits that may be realised (see Box 5.2). 155 Box 5.2 The potential benefits of health information exchange and interoperability • Interoperability between both freestanding and hospital-based outpatient clinicians: This would enable computer-assisted reduction of redundant tests, and it would reduce delays and costs associated with paper-based ordering and reporting of results. In addition, provider-laboratory connectivity would give clinicians better access to patients’ longitudinal test results, eliminate errors associated with reporting results orally, optimise ordering patterns by making information on test costs readily available to clinicians, and make testing more convenient for patient. • Connectivity between external radiology centres: This would reduce redundant tests and would save time and costs associated with paper- and filmbased processes. Interoperability here could also improve ordering by giving radiologists access to relevant clinical information, thereby enabling them to recommend optimal testing; improve patient safety by alerting both the provider and the radiologist to test contraindications; facilitate coordination of care and help prevent errors of omission by enabling automated reminders when followup studies are indicated; and lessen adverse environmental impacts by reducing the use of chemicals and paper in film processing. • Out-patient providers and pharmacies: Interoperability between out-patient providers and pharmacies would reduce the number of medication-related phone calls for both clinicians and pharmacists. It would also improve clinical care by facilitating the formation of complete medication lists, thereby reducing duplicate therapy, drug interactions and other adverse drug events, and medication abuse. It could also enable automated refill alerts, offer clinicians easy access to information about whether patients fill prescriptions, and complete insurance forms required for some medications. In addition, it could help identify affected patients in the event of drug recalls, uncover new side effects, and improve formulary management. • Provider-provider connectivity: Has the potential to save time associated with handling chart requests and referrals. Connectivity would reduce fragmentation of care from scattered records and improve referral processes. • Provider connectivity to the US public health system: This would make reporting of vital statistics and cases of certain diseases more efficient and complete. However, the most important impact of public health interoperability would almost certainly derive from earlier recognition of emerging disease 156 outbreaks and bio-surveillance, as it becomes easier to identify warning signs and trends by aggregating data from many sources. Since robust quantitative evidence about the value of HIEI in earlier recognition of disease and biosurveillance does not yet exist, we did not project value from these sources6. • Provider-payer transactions: These enjoy a relatively high degree of standardisation, largely because of Health Insurance Portability and Accountability Act. Some transactions are highly automated, but others are not, particularly in smaller organisations. Most of the benefits they describe are also likely to apply to the NHS. In summary, HIEI should save resources through the reduction of administration by ending repeat data entry, preventing duplication of investigation, facilitating more targeted investigation and improving access to records for front-line staff. This improved access to more comprehensive records should also help obviate the problems associated with access to the current fragmented records, which can, for example, reduce the risks of errors that commonly occur when people move across healthcare transition boundaries; e.g. being discharged from hospital back to primary care. Walker et al. estimated the projected financial benefits in the US based on the impact on laboratory services alone at $31.8 billion. In their analysis of aspects of interoperability for which they were able to allocate:4 ‘…dollar values, net savings from national implementation of fully standardised interoperability between providers and five other types of organisations [they] estimate that this could yield $77.8 billion annually, or approximately 5% of the projected $1.661 trillion spent on US healthcare in 2003.’ Although a detailed analysis to quantify the patient safety benefits expected from the NPfIT–which has had at its core the need to overcome historic problems with data sharing–has not been conducted by NHS CFH, it believes, based on a limited preliminary economic analysis that this could be worth several billions of pounds of savings to the NHS. This estimate includes: 157 ‘£2.5 billion as the human value of preventable fatalities from medication errors arising from inadequate information about patients and medicines; a large proportion of the £500 million spent each year on treating patients who are harmed by medication errors and adverse reactions; a reduction in the payments by NHS Trusts each year (approximately £430 million each year) for settlements made on clinical negligence claims.'7 In discussing the potential impact on patient safety, Kaelber and Bates have estimated that developments in HIEI should through optimising the availability of information substantially reduce the incidence of prescribing errors and resulting medication-related adverse drug events.8 Improved potential for continuity of care is another important potential benefit of such developments. The two detailed case studies by Blick (1997 & 2001) are also important in that they provide insights into the potential gains associated with improved information management in the fields of pathology services.9;10 In the first, he discusses the need to focus on the broader non-analysis related quality considerations such as specimen logging, tracking and the issuing of reports, which, based on simulations with his advanced information management system, can greatly be improved.9 The second paper highlights the difficulties posed by stand-alone point-of-care (POC) systems in critical care medicine, these including the lack of recording of results in patients’ electronic health records (EHRs) and the inability easily to bill for such procedures.10 Such failings he hypothesises could again be reduced through use of better designed and connected information management systems. 5.3.2 Risks Movement across geographical boundaries The ability to transmit data relies heavily on adequate identification of patients. In England, patients now have a unique NHS Number.11 A unique Community Health Index (CHI) number also exists in Scotland,12 therefore, even within the UK, translation algorithms may be required when patients move between different countries. This ability to track patients as they move has also been identified as a major issue in countries that do not have unique identifiers. 158 Transcription errors While HIEI should reduce the frequency with which data are keyed in and is hence likely to reduce error, the potential to key in the wrong information into the wrong file still exists. When transcription errors occur in systems interacting at Level 1 or Level 2, the necessary human interaction means that at least some check on data is occurring. Obvious errors that a machine may not detect (for example, a pregnancy test result wrongly attributed to a 65-year-old woman, or the contraceptive pill being prescribed to a man) may go unnoticed unless plausibility checks are incorporated into higher level systems. Privacy and security Security is a major concern. Any system which makes it easier for clinicians and patients to access their data inevitably makes it easier for unauthorised individuals to gain access. Similarly, the more people that have access to a patient’s record the more risk there is that, whether by accident or design, confidential information will leak.13 Casual access by clinical staff to unauthorised material is, for example, a long recognised challenge.14 Audit trails and personal logons which can identify unauthorised entry are another option, but in the fast-paced environment of hospital wards, GP surgeries and pharmacies it is easy for machines to remain logged-on and unauthorised access to occur. While such breaches were always possible, HIEI potentially increases the harm of such breaches. System failure The risk that failure in one component could have a major effect on the functioning of the whole system increases exponentially as systems become more integrated. A practice which keeps a local record will not, for example, be affected by a failure in communication with a central server. Therefore, the HIEI system as a whole, and all its components, must be reliable and be able to assure a uniform, satisfactory level of service quality in addition to running being securely and regularly backed up, so that organisations can rely on the overall system availability. It must provide for realtime access to information, particularly for urgent care specialties such as emergency medicine and intensive care. 159 Considerations relating to Implementation As previously mentioned, most UK healthcare settings are characterised by relatively low levels of HIEI; this is particularly true of the hospital sector. Similar problems exist in the US and solutions have tended to evolve locally, but this can create subsequent problems with reluctance to yield ownership of a well-known local system to become part of a larger more integrated system with thus no clear ways for HIEI capabilities to easily expand beyond these initial projects.15 Likewise some vendors who rely on interoperability of their products as a means of generating revenue are likely to see the use of a common platform as a threat. There are parallels here with general practice computing in the UK. Improving HIEI to the desired level is likely to generate long-term revenue savings; however, this will require considerable up-front investment in hardware and software capabilities with early adopters paying most. There is considerable uncertainty in some quarters about the extent to which the development of EHR should proceed to before agreed systems of HIEI are developed with early adopters of such records fearful that any subsequent standards agreement will leave them stranded with obsolete systems. However, experts in the US strongly advocate the integrated approach, expressing serious reservations about the potentially wasted energy in developing local systems.15 5.4 Empirically demonstrated benefits and risks 5.4.1. Benefits Given that this remains a relatively new field of empirical enquiry, we unsurprisingly found only limited empirical evidence summarised in systematic reviews (see Appendix 5) in well-circumscribed areas to draw upon. Below, we discuss the main findings contained within these three systematic reviews (SR) and additional selected original investigations. In the SR by Georgiou et al, they considered the impact of computerised physician order entry (CPOE) systems embedded in hospital pathology information management systems.16 This review identified 19 controlled studies employing a variety of designs, including randomised controlled trials (RCTs), quasi-experimental 160 designs and before-after studies. This review helpfully identified a suite of indicators to assess the impact of these computerised systems on various aspects of pathology service function. Overall, however, they concluded that there are some data to support the impact of these integrated systems on process measures such as a reduction in the number of tests, the ordering of redundant tests and the costs of these investigations, there is as yet little evidence of a beneficial impact on patient outcomes such as length of stay, adverse events and mortality. In a SR focusing on communication between hospitals and family practices Kripalani et al. found that deficit in communication and information transfer at hospital discharge are common, these needlessly jeopardising patient care.17 From their critique of a large body of evidence (73 studies in total), they were able to identify 18 controlled investigations (which included three RCTs) focusing in the main on comparisons between computer- and hand-generated discharge summaries. Comparison between studies was challenging because of the range of outcome measures used, but nonetheless overall they found that computer-generated summaries were generated by hospital clinicians in a more timely fashion; there was also some evidence of improved quality of information in these summaries and the timing of the receipt of these summaries by family practitioners. There was however no clear evidence that these improvements in the processes of care translated into improved clinical outcomes for patients. More recently, Fontaine et al have reported a SR on HIEI amongst primary care providers.18 Their investigations identified a total of over 60 eligible reports, of which 39 were peer-reviewed publications. Twenty of these peer-reviewed publications presented original data; however, only two of these studies employed RCT designs. These original studies focused on a range of relevant outcome measures, but overall the only robust finding was for evidence of increased efficiency in relation to accessing relevant patient data generated from outside the practice (such as pathology results) and staff time involved in processing claims. There was relatively little in the way of clear evidence of positive impacts on either the quality or safety of care. Importantly, they concluded that it was not as yet possible to demonstrate a positive return on investment. 161 There have been early reports of regional attempts at developing interoperability networks in the US, which provide helpful pointers to some of the challenges that have needed to be overcome and structures and processes that can prove helpful in making progress locally. Halamka et al. have, for example, reported on their experiences of establishing an interoperability network in Massachusetts, which focused on the sharing of clinical and financial data.15 This has been a time- consuming process, but one which the authors believe has been successful as judged by the feasibility work, and the development of relevant structures and policies that have helped to place this network on a stable footing. It is also claimed that this has resulted in substantial cost savings in the processing of claims (from $5 to 25 cents per transaction). Early successes in the feasibility and claims about the sustainability of creating data exchange consortia were however also made by the Santa Barbara County Care Exchange (SBCCE) scheme, and they too reported the ability to generate cost savings was a key ingredient to the early success of this venture.19 However, this scheme has recently disbanded for reasons that included a lack of true multistakeholder involvement, a pre-occupation in high-tech solutions at the expense of meeting the everyday needs of healthcare providers, over-optimistic assumptions on financial returns and the difficult question of how to balance the sharing of information with privacy concerns.20;21;22 Taking a mainly European perspective, Dobrev et al have conducted an interesting series of economic case studies of implementations of EHRs and ePrescribing systems in 11 countries.23 These detailed investigations involved a combination of literature reviews, qualitative interviews, analysis of cost and benefit data, and economic modelling, from which they concluded that investment in eHealth requires up-front investment from healthcare organisations, but that this in most cases results in net economic gains to society (but not necessarily healthcare organisations). In contrast with many of the claims made about eHealth, they conclude that this does not therefore necessarily equate with a return on investment, i.e. the capital outlay translates into improvements in the quality of care, but at a cost. Another important conclusion and one that is of considerable relevance to the discussions in this chapter is that these benefits are primarily mediated through enhanced inter162 operability, which results in the ability to reuse data for multiple ends such as audit, research, planning and surveillance. Finally, the authors underscored the point that it typically takes years for these benefits to accrue: investment in eHealth must not therefore be seen as an “easy win”. 5.4.2 Risks Johnson has in an unpublished study identified vulnerabilities in highly integrated information systems in both the US (Veteran Affairs Administration system) and England (Lorenzo), which led to system downtimes necessitating a return to paperbased models of care for which healthcare organisations and professionals were illprepared.24 He urges the need to take steps to minimise such risks and also for contingency planning for scenarios in which downtimes occur. The issue of consent and privacy is also important and has in England at least in particular centred on consent models for the Summary Care Record. The response of moving from an opt-out to opt-in model has only partially allayed concerns in this respect.13;25;26 Also of relevance to HIEI is the ongoing issue of concerns relating to data quality in EHRs, which typically serve as the lynchpin for integration between different information systems.27 Inaccuracies and ambiguities in the core clinical record can, as a result of information sharing, be magnified and result in additional work to reconcile records. Finally, as the discussion on the Santa Barbara21 and European Commission23 above highlight, there is likely to be a need for substantial upfront and net expenditure to achieve effective HIEI with no guaranteed direct returns to healthcare organisations or indeed society on such investments. 5.5 Implications for policy, practice and research 5.5.1 Policy There are several features of UK practice which provide a positive platform for increased HIEI, although important challenges remain. This includes the single- payer nature of the NHS, which in contrast to other health systems where there may 163 be a multitude of healthcare purchasers, both public and private. There is also, for example, strong central government support for increasing information sharing.28 The central procurement of major IT by NHS CFH now allows for much greater enforcement of standards nationally in relation to interoperability compared to previously haphazard IT commissioning with numerous systems, although concerns are being raised about a return to the previous picture resulting from the increasing move to localisation in the post-New Labour era. The certification model now being pursued in the US through the Certification Commission for Health Information Technology is in this respect potentially of considerable relevance to the UK situation.29 The adoption of Health Level Seven (HL7)–which is a volunteer organisation that “…provides a framework (and standards) for the exchange, integration, sharing and retrieval of electronic health information through defining standards, guidelines and methodologies”32–by NHS CFH is a particularly important development.30 The HL7 version 3 messages and clinical document architecture are used by the NHS to facilitate the transfer of information. Importantly, this standard is being adopted internationally, for example, in Australia by the National EHealth Transition Authority.31 Alternative systems such as OpenEHR32 also have strong support and are gaining ground in many European countries. While ensuring computable semantic interoperability between complex datasets will continue to pose significant problems to healthcare IT,33 security and confidentiality also remain a challenge and further developments need to be encouraged to reassure both clinical staff and the public that information, widely available to many different groups within the NHS, will remain secure. 5.5.2 Practice One of the key challenges that face healthcare systems in realising the potential for seamless exchange of information is the development and deployment of standard coding structures across all care settings. Currently, in the UK, different codes are used by hospitals and general practices. However, NHS CFH, with strong central backing, has adopted the Systematized Nomenclature of Medicine-Clinical Terms (SNOMED-CT) system.34;35 This is a common computerised language that will in due 164 course be used by all computers in the NHS to facilitate communications between different healthcare professionals. It is a joint development between the NHS and the College of American Pathologists to create an agreed terminology and is likely to be widely adopted internationally. It has greater depth and coverage of healthcare than other versions of clinical terms such as Read codes and will facilitate the exchange of healthcare and clinical knowledge by clinicians, researchers and patients worldwide. One of the real opportunities for the future of healthcare, but an important challenge to integrated care, is the growth of patient-generated data from telemetric devices. The Continua Health Alliance is attempting to standardise protocols and improve interoperability of such devices and NHS CFH is an important member of this consortium.36 The current policy of encouraging HIEI through the rigorous setting of standards, through contractual arrangements, for software and hardware intended to link with the SCR has been shown to be effective, particularly in primary care computing systems (for example, GP2GP) and such arrangements should continue. Part of the reasons underpinning the success of GP2GP has been the close liaison with general practice representative bodies such as the British Medical Association’s (BMA) General Practitioners Committee (GPC) and the Royal College of General Practitioners (RCGP) on the GP2GP Project Board, this dating back to the outset of the project. Such liaisons in general practice and other parts of the NHS are important in facilitating the introduction of innovations such as HIEI (see Chapters 17 and 18). 5.5.3 Research Further research to identify both the benefits and costs of HIEI and the barriers (for example, concerns regarding confidentiality and commercial interests) and facilitators to its implementation are required. Research on how patient-generated data can easily be incorporated into the EHR and how these are then synthesised and made use of is also necessary. 165 More generally, the UK occupies a strong international position in terms of access to health and other related datasets and the expertise in analysing these data, whether for public health surveillance, epidemiological work or health services research. The relatively few linkages between the many stand-alone datasets however prevents innovation in this area, necessitating complex and less than perfect algorithms to undertake linkages; more widespread use of unique identifiers could help greatly in helping to unlock the potential within these datasets and help to further the information revolution that is now being called for.28 References 1. Coiera E. Guide to Health Informatics. 2nd ed. London: Hoder Arnold; 2003. 2. Tanebaum AS. Computer Networks. 4th ed. Englewood Cliffs, New Jersey: Prentice Hall; 2002. 3. Health Information and Security Privacy Collaboration. Available from: http://healthit.hhs.gov/portal/server.pt?CommunityID=1206&spaceID=399&parentname= &control=SetCommunity&parentid=&PageID=0&space=CommunityPage&in_hi_totalgrou ps=1&in_hi_req_ddfolder=6652&in_ra_topoperator=or&in_hi_depth_1=0&in_hi_req_pag e=20&control=advancedstart&in_hi_req_objtype=18&in_hi_req_objtype=512&in_hi_req_ objtype=514&in_hi_req_apps=1&in_hi_revealed_1=0&in_hi_userid=8969&in_hi_groupo perator_1=or&in_hi_model_mode=browse&cached=false&in_ra_groupoperator_1=or&in _tx_fulltext=health+information+exchange (last accessed 28/12/10). 4. Walker J, Pan E, Johnston D, Adler-Milstein J, Bates DW, Middleton B. The value of health care information exchange and interoperability. Health Aff (Millwood) 2005; W510-W5-18. 5. The Scottish Government website. eHealth Strategy 2008-11.Available from: http://www.scotland.gov.uk/Publications/2008/08/27103130/4 (last accessed 28/12/10). 6. Department of Health. Report of the Review of NHS Pathology Services in England. An Independent Review for the Department of Health. Available from: http://www.dh.gov.uk/en/Publicationsandstatistics/Publications/PublicationsPolicyAndGui dance/DH_4137606. 7. National Audit Office. The National Programme for IT in the NHS. http://www.nao.org.uk/publications/nao_reports/05-06/05061173.pdf (last accessed 28/12/10). 8. Kaelber DC, Bates DW. Health information exchange and patient safety. J Biomed Inform 2007; 40: S40-5. 9. Blick KE. Decision-making laboratory computer systems as essential tools for achievement of total quality. Clin Chem 1997; 43: 908-12. 10. Blick KE. The essential role of information management in point-of-care/critical care testing. Clin Chim Acta 2001; 307: 159-68. 11. NHS Choices. The NHS Number. Available from: http://www.nhs.uk/NHSEngland/thenhs/records/Pages/thenhsnumber.aspx (last accessed 28/12/10). 12. The Scottish Government. Community Health Index number http://www.scotland.gov.uk/Topics/Health/NHS-Scotland/DeliveryImprovement/1835/1865/1852 (last accessed 28/12/10). 13. Anderson R, Brown I, Dowty T, Inglesant P, Health W, Sasse A. Database State. London: Jospeh Rowntree Reform Trust, 2009. 166 14. Silvester N. Doctor who hacked into Prime Minister's health records escapes http://www.dailyrecord.co.uk/news/healthprosecution. Available from: news/2010/01/10/doctor-who-hacked-into-prime-minister-s-health-records-escapesprosecution-86908-21955907/ (last accessed 27/12/10). 15. Halamka J, Overhage JM, Ricciardi L, Rishel W, Shirky C, Diamond C. Exchanging health information: local distribution, national coordination. Health Aff (Millwood) 2005; 24: 1170-79. 16. Georgiou A, Williamson M, Westbrook JI, Ray S. The impact of computerised physician order entry systems on pathology services: A systematic review. Int J Med Inform 2006; 76: 514-29. 17. Kripalani S, LeFevre F, Phillips CO, Williams MV, Basaviah P, Baker DW. Deficits in communication and information transfer between hospital-based and primary care physicians: implications for patient safety and continuity of care. JAMA 2007; 297: 83141. 18. Fontaine P, Ross SE, Zink T, Schilling LM. Systematic review of health information exchange in primary care practices. J Am Board Fam Med 2010; 23: 655-70. 19. Brailer DJ, Augustinos N, Evans L, Karp S. Moving toward electronic information health exchange: Interim report on the Santa Barbara County Data Exchange. Prepared for California Healthcare Foundation. 2003. 20. Brailer DJ. From Santa Barbara to Washington: a person's and a nation's journey toward portable health information. Health Aff (Millwood). 2007; 26: w581-8. 21. Miller RH, Miller BS. The Santa Barbara County Care Data Exchange: what happened?" Health Affairs 2007; 26: w568–w580. 22. Vest JR, Gamm LD. Health information exchange: persistent challenges and new strategies. JAMIA 2010; 17: 288-94. 23. Dobrev A, Jones T, Stroetmann V, Stroetmann K, Vatter Y, Peng K. Interoperable eHealth is Worth it: Securing Benefits from Electronic Health Records and ePrescribing. Bonn/Brussels: European Commission, 2010. 24. Johnson CW. Case studies in the failure of healthcare information systems. Personal communication, 2009. 25. Greenhalgh T, Stramer K, Bratan T, Byrne E, Mohammad Y, Russell J. Introduction of shared electronic records: multi-site case study using diffusion of innovation theory. BMJ 2008; 337: a1786. 26. Greenhalgh T, Stramer K, Bratan T, Byrne E, Russell J, Potts HW. Adoption and nonadoption of a shared electronic summary record in England: a mixed-method case study. BMJ 2010; 340: c3111. 27. Thiru K, Hassey A, Sullivan F. Systematic review of scope and quality of electronic patient record data in primary care. BMJ 2003; 326: 1070. 28. Department of Health. Liberating the NHS: An Information Revolution. London, 2010 29. Certification Commission for Health Information Technology. Available from: http://www.cchit.org/ (last accessed 28/12/10). 30. Health Level 7. Available from: http://www.hl7.org.uk/ (last accessed 27/12/10) 31. National E-Health Transition Authority. http://www.nehta.gov.au/ (last accessed 28/12/10). 32. OpenEHR. Available from: http://www.openehr.org/home.html (last accessed 28/12/10). 33. Mead CN. Data interchange standards in healthcare IT--computable semantic interoperability: now possible but still difficult, do we really need a better mousetrap? J Healthc Inf Manag 2006; 20: 71-78. http://www.ihtsdo.org/nc/our-standards/snomed34. SNOMED-CT. Available from: ct/?sword_list%5B%5D=snomed (last accessed 28/12/10). 35. Simpson CR, Anandan C, Fischbacher C, Lefevre K, Sheikh A. Will Systematized Nomenclature of Medicine-Clinical Terms improve our understanding of the disease burden posed by allergic disorders? Clin Exp Allergy 2007; 37: 1586-93. 36. Continua Health Alliance. Available from: http://www.continuaalliance.org/index.html (last accessed 28/12/10). 167 Chapter 6 Electronic health records Summary • The electronic health record is a complex construct encompassing digitised patient and healthcare records and the information systems into which these are embedded. • A range of terms has been used to describe different types of electronic health record and there is, at yet, no one agreed taxonomy. Electronic health records systems vary in their complexity, with some being closely coupled to other eHealth functionalities such as decision support. These sources of variation are evident in published evaluations, presenting a challenge to evidence synthesis to inform policymaking. • Electronic health records are at the heart of eHealth implementation plans around the world. The long-term vision is for fully integrated, longitudinal, patient records, supported by universal data and interoperability standards, and high level communication and decision support technologies. This is some way from being realised. • Electronic health records have the potential to improve the accessibility, completeness and accuracy of healthcare documentation and support clinical information exchange across teams. Electronic health record systems that integrate decision support can aid clinical decision-making and guideline compliance, enhance health promotion activities and reduce medical errors and adverse drug events. Those that are accessible to patients may enable self-management and shared decision making. • Many of the practical benefits of electronic health records, such as improved care quality, patient safety and public health, may arise from the secondary uses of data for audit, surveillance, planning and research. • Despite the many theoretical benefits of electronic health record systems, previous studies have yielded mixed evidence of their effectiveness and there have been relatively few long-term economic appraisals. 168 Summary continued… • Theoretical benefits pertaining to savings in time and cost have yet to be substantiated by empirical evidence although there is some evidence of enhanced data accuracy, completeness, legibility and accessibility. • Achieving the successful adoption of electronic health records depends greatly on understanding the context in which they are to be implemented, including characteristics of the organisation and the work processes of potential users, as well as on the quality of technical support and training provided and the extent to which decision support is integrated. • The quality of data recorded in electronic health records varies widely, due to a range of socio-technical factors surrounding individual users, variations in the mechanisms and practice of data coding, and compatibility of the different systems used. Standardised and widely accepted measures of data quality in electronic health records are lacking, but the introduction of the Systematized Nomenclature of Medicine-Clinical Terms has potential to improve the consistency of clinical coding across the NHS. • The sensitive nature of the personal information held in electronic health records calls for clear technical and governance measures to ensure appropriate access and safeguard privacy. While patients generally find clinical uses of electronic health records acceptable, secondary uses of privileged health data by third parties arouse more concern. 6.1 Introduction The implementation of electronic health records (EHR) is a core objective of national and organisational eHealth strategies worldwide. This is driven by a recognition of the potential benefits of EHR for healthcare quality and efficiency, as well as their central importance for the delivery of eHealth systems and services, as illustrated in the conceptual map in Chapter 4.1 This chapter reviews core concepts surrounding EHR, considers the theoretical benefits and risks associated with their introduction, and reviews 169 the empirical evidence demonstrating their impacts on quality, safety and organisational efficiency. 6.2 Definition, description and scope 6.2.1 Definitions The EHR is a complex construct encompassing different types of digitised medical and healthcare records and the information systems into which these are embedded. Many terms have been used to describe different aspects of the EHR and it has been difficult to achieve universal consensus on a single taxonomy. The EHR has been defined in terms of: data, e.g.: “an individual patient's medical record in digital format”;2 systems, e.g. “the set of components that form the mechanism by which electronic health records are created, used, stored, and retrieved. It includes people, data, rules and procedures, processing and storage devices, and communication and support facilities”;3 and databases e.g. “a repository of information regarding the health of a subject of care, in computer processable form”.3 More comprehensive definitions recognise all of these components.a Examples of prominent European, United States (US) and international policy definitions are provided in Box 6.1. These are broadly compatible and tend to emphasise the role of the EHR as an integrator of patient information across time and between providers, although, to an extent, this objective still represents a vision rather than a reality. a EHR literature has also been categorised in terms of its philosophical underpinnings 19 170 Box 6.1: Definitions of electronic health records used by key UK, US and International Organisations “The concept of a longitudinal record of a patient’s health and healthcare to combine information from primary healthcare with periodic care from other institutions” [Source: UK Department of Health]4 “A longitudinal collection of patient-centric, healthcare information, available across providers, care settings, and time. It is a central component of an integrated health information system.” [Source: US Institute of Standards & Technology]5 “Repositories of electronically maintained information about individuals’ lifetime health status and healthcare, stored such that they can serve the multiple legitimate users of the record.” [Source: EC, Information Society and Media]6 “Contains all personal health information belonging to an individual; is entered and accessed electronically by healthcare providers over the person’s lifetime; and extends beyond acute inpatient situations including all ambulatory care settings at which the patient receives care.” [Source: WHO]7 “A repository of information regarding the health status of a subject of care in computer processable form, stored and transmitted securely, and accessible by multiple authorised users. It has a standardised or commonly agreed logical information model which is independent of EHR systems. Its primary purpose is the support of continuing, efficient and quality integrated health care and it contains information which is retrospective, concurrent, and prospective.” [Source: ISO]3 A longitudinal electronic record of patient health information generated by one or more encounters in any care delivery setting. Included in this information are patient demographics, progress notes, problems, medications, vital signs, past medical history, immunizations, laboratory data, and radiology reports. The EHR automates and streamlines the clinician's workflow. The EHR has the ability to generate a complete record of a clinical patient encounter, as well as supporting other carerelated activities directly or indirectly via interface—including evidence-based decision support, quality management, and outcomes reporting.” [Source: HIMSS]8 Box 6.2 illustrates some of the more specific terms that have historically been used to describe different types of electronic records in healthcare. These are not mutually exclusive and the landscape is complicated by widespread variation in the uses of particular terms to describe equivalent concepts.b There is nonetheless an important conceptual distinction to be made between b For example, electronic medical record in the US, electronic health record in Canada, electronic patient record in England and electronic health care record in Europe. 171 digitised patient-specific records gathered in one setting (whether episodic or longitudinal) and those that draw together information from multiple settings, as in the integrated, longitudinal model referred to above. Terms used to describe the former often include the words ‘patient’ or ‘medical’, whilst there is an emerging international consensus around the use of ‘electronic health record’ to reflect the broader concept. In light of the semantic complications, we have elected to use EHR as a generic term in this report, as others have also begun to do, but we remain mindful of the underlying complexity.9 It is worth noting the link between the concept of health information exchange (HIE), which concerns systems that enable the transfer of data between providers, and the concept of the EHR. While, on the one hand, the former involves communications and the latter data, the systems into which EHR are set are very often equivalent to those described under the banner of HIE. This conceptual link is illustrated in this extract from a recent commercial document: “… EHR relates to something much broader – namely, a system-wide health record that takes data through a Health Information Exchange. The EHR contains all patient data and follows the individual wherever he or she goes. The HIE pulls in data relating to that patient from the insurance company or government agency, from hospitals and physician offices, labs, pharmacies and other sources of clinical and administrative data. Then, using clinical best practices from organizations such as the American Medical Association, it analyzes the patient information and derives alerts and recommendations to turn the data into actionable information. As a result, a fragmented healthcare environment is transformed into a single portal where data are consolidated and recommendations are provided at the point of care.”10 172 Box 6.2 Examples of Terms Used to Describe Types of Electronic Health Record11-15 • Electronic Medical Record (EMR),16 (EPR), 21 22 23 17 18 19 20 Electronic Patient Record Computerised Medical Record (CMR),22 24 Computerised Patient Record (CPR),25 Computer-Based Medical Record (CBMR),26 Computer-Based Patient Record (CBPR),27 Digital Medical Record28 29 (sometimes reserved for web-based records), Automated Health Records (AHR),30 31 and Automated Medical Records (AMR).32 Largely equivalent terms used to describe digitally stored patient records, including those created on computer or transcribed or scanned from paper records (AHR, AMR). These usually refer to information from single providers (e.g. a GP practice, a diabetes clinic, a hospital), but are often used interchangeably with the broader term EHR. In the UK, the terms Summary Care Record (SCR) and Emergency Care Summary (ECS) have been used to refer to an abbreviated patient record held nationally for use in emergency or out-of-hours settings, which is typically derived from the electronic primary care record,33 34 35 whilst Detailed Care Record (DCR) is used to describe more detailed records held in one setting (i.e. EMR/EPR).36 The term Continuity of Care Record (CCR) describes a standard for a summarised EHR that can be accessed by and added to by different health professionals managing a patient, so as to support integrated and concurrent care. 37 38 • Electronic Health Record (EHR), Electronic Care Record (ECR),39 Integrated Care Record (ICR),40 Electronic Health Care Records (EHCR):12 A record that contains information from multiple providers of the patient’s care. It varies in how the information is integrated (e.g. centralised data storage versus linkage to federated data stores), how much information is integrated (detailed or summary), and the providers of the information (e.g. an integrated diabetes care record vs. a more generic shared care record). Longitudinal health record (LHR) is used to describe the EHR when available over time 41 42 • Interoperable Electronic Health Record (IEHR):43 44 This term has been used to describe an EHR that can interface with a range of records systems and databases, and with tools for decision-making and communication. 173 Box 6.2 continued… • Personal Health Record (PHR):45 Sometimes also referred to as Electronic Patient Carried Medical Records (EPCMR) or Patient Medical Records (PMR),46 Personal Electronic Health Record (PEHR),47 Electronic Personal Health Records (ePHR).48 These refer to records that are accessible by the patient themselves. These vary in locus of control (patient or provider as author of content or controller of access), medium (e.g. SmartCard, PC, web), functional complexity (e.g. inclusion of patient information, prescription renewal, booking) and extent of integration with provider-held EHR systems. In Australia, the term Person-Controlled Electronic Health Record (P-CEHR) describes an EHR that is accessible to the patient, who can choose which healthcare professionals may access it.49 50 Elsewhere the term Personally Controlled Health Record (PCHR) has been used to describe a class of electronic Personal Health Record (PHR) such as Google Health and Microsoft’s HealthVault in which the patient controls both record access and content. • Population health record:51 refers to “aggregated and usually de-identified data [which] may be obtained directly from EHRs or created de novo from other electronic repositories. It is used for public health and other epidemiological purposes, research, health statistics, policy development, and health service management.” • 3 Virtual EHR:52 This has no authoritative definition, but is related to the concept of an online record. NB. While this box focuses on records, rather than information systems, the two concepts are closely related and the term EHR has been used to refer to both. In England, previous Department of Health (DH) reports have attempted to differentiate the EHR from the EPR by defining the latter as a record of a patient's contacts with one healthcare provider, such as a general practice, and the former as a record of their overall health and healthcare, which combines information stored in EPRs held by different healthcare providers, such as general practices, community services, and hospitals. However, many reports and studies do not make this distinction.53 174 The term Personal Health Record (PHR) has been used to describe a range of patient-oriented EHR, ranging from those designed for the individual to record and manage their own health information for personal use, to those that that aid the sharing of relevant health information between patient and provider, through enabling the patient to access their provider-held EPR,c or to record data from personal health-monitoring devices. Different platforms and applications have been designed to support PHR, and the online ‘portal’ model, used by organisations such as Kaiser Permanente in the US, is gaining currency.48 Such portal models also enable patients to undertake a variety of administrative communications, such as appointment booking and prescription renewal and to access patient information. There remains wide variability between PHR models and approaches. For example, in a recent report on EHR impact for the European Commission’s Information Society and Media Directorate-General, Groen et al firmly place PHR at the patientcontrolled end of the spectrum, whilst acknowledging that a range of healthcare providers may contribute towards it. According to these authors, “The personal health record (PHR) is different from an electronic health record (EHR) system maintained by a healthcare provider organisation. The PHR is maintained by the individual patient. These individuals own and manage the information in the PHR, which comes from both multiple healthcare providers and the individuals themselves.” 6 54 Box 6.3 illustrates some of the personal health management functions that such a system may provide. The US National Committee on Vital and Health Statistics, amongst others, has noted on the difficulties presented by the variable conceptual and technical landscape of PHR existing today “This lack of consensus makes collaboration, coordination and policymaking difficult. It is quite possible now for people to talk about PHRs without realizing that their respective notions of them may be quite different”.55 c In the UK, this has generally been considered under the banner of ‘Record Access’, since more complex PHR technologies remain common. 175 Box 6.3: Different Features of Personal Health Records Potential Functions of electronic Personal Health Records • • • • • • • • Access to provider’s electronic clinical record Personal health organiser or diary Self management support Secure patient-provider communication Links to information about illness, treatments, or self care Links to sources of support Capture of symptom or health behaviour data Patient empowerment and education Information that may be contained in the Personal Health Record • • • • • • Fitness data, data from routine medical checkups Automatic interpretations and/or suggestions based on data entered Data imported from other health devices Details of medical appointments with automatic reminders Over the counter medication record Health and wellness information, including patient-oriented summaries and links to research Adapted from Pagliari et al (2007), Ross et al (2003) and Groen et al (2007), 46 56 6 52 with permission In the UK, patient awareness and use of PHRs remains low. While lack of exposure to such technologies is clearly a contributing factor, the explanation may also lie in aspects of the UK healthcare system and culture which differ from those in other parts of the world where PHR are available. For example, a key benefit of PHR in disaggregated health systems is that they can act as a point of data integration mediated by the patient. In the UK, general practitioners effectively have this integrating role and the benefits of PHR may be more around addressing the ‘Information Society’ agenda of enabling citizens to access their personal information and supporting accessible online services, as well as accommodating the growing use of personal heath monitoring devices, which provide relevant data streams outwith the NHS. This may help to explain why the fledgling NHS system HealthSpace (which provides patient access to the NHS Summary Care Record (SCR), booking functions and patient information) was not widely taken up in early pilots.45 In contrast, ‘home grown’ systems designed in general practice to enable patients to view their detailed GP record via kiosk or online portal appear have met with more favourable responses, particularly in patients with complex or long term health needs.57 176 The relationships and differences between the terms are illustrated in Figure 6.1 below. Although it is complex to capture the meanings these varied terms denote and the contexts in which they are used, an attempt is made in Figure 6.2 to provide an overview of this terminology. 177 Fig 6.1: Map of key terms used to describe electronic health records Population Health Record Electronic Health Record (Generic Term) Integrated & longitudinal Usually provider or setting-specific Aggregated & anonymised • Electronic Health Record (EHR) • Integrated Health Record (IHR) • Electronic Health Care Record (EHCR) • Interoperable Electronic Health Record (IEHR) • Integrated Care Record (ICR) • Longitudinal Health Record (LHR) Subset of data, for emergency or multispecialty care. • Summary Care Record (SCR) or • Electronic Patient Record (EPR) • Emergency Care Summary (ECS) • Electronic Medical Record (EMR) • Continuity of Care record (CCR) • Computerised Medical Record (CMR) • Computerised Patient Record (CPR) • Computer-Based Medical Record (CBMR) • Digital Medical Record (DMR, sometimes reserved for web-based records) • Electronic Client Record (ECR) • Automated Health Record (AHR) • Automated Medical Record (AMR) Older term, usually a digital version of a paper record • Personal Health Record (PHR) • Electronic Patient-Carried Medical Record (EPCMR) • Electronic Personal Health Record (ePHR) • Personally Controlled Patient Record (PCPR) Patient held, accessible, or controlled NB. Personal Health Record may also have elements of the integrated EHR 178 Data providers and users GP GP clinic EMR Time Clinician Hospital EMR Lab specialist PHR Radiology EMR (PACS / RIS) Pharmacy EMR LHCR Pharmacist Primary use of data Radiologist IHCR Laboratory EMR (LIS) Registries Data & Services Health Information Data Warehouse EMR EPR Figure 6.2 Infostructure of EHR, with index below Electronic Social Care Record Community Care EMR Multiple health care professionals Interoperability Middleware Multiple patients Non-Shareable EHR Confidential Data 179 Researchers, Policy makers etc Secondary use Shareable EHR Anonymised Data EHR - electronic medical record of multiple episodes of care of a population by several healthcare professionals over a long period of time the patient over the web 180 PHR - electronic medical record of multiple episodes of care of a patient by several healthcare professionals over a long period of time accessible by LHCR - electronic medical record of multiple episodes of care of a patient by several healthcare professionals over a long period of time IHCR - electronic medical record of multiple episodes of care of a patient by several healthcare professionals over a short or long period of time EPR - electronic medical record of multiple episodes of care of a patient by one healthcare professional over a period of time EMR - electronic medical record of a single episode of care of a patient by one healthcare professional Definitions used: Abbreviations used: Laboratory information system LIS, Picture Archiving and Communication Systems PACS, Radiology Information Systems RIS Health Informatics –Electronic Health Record –Definition, scope and context, ISO, 2005.3 EHRS Blueprint-an interoperable EHR framework, Canada Health Infoway, April 2006;58 EHR Impact study, European Commission Information Society and Media, April 2008;6 Source: In part based on: 6.2.2 Settings and usage of EHR EHR may be used in most healthcare settings, including primary care, secondary care, tertiary care and increasingly remote and home care. In practice, however, there are wide variations between settings and sectors in terms of EHR availability and usage. In the UK, for example, computerisation is almost universal in primary care, while progress in secondary care has lagged behind and for this reason has been a strategic priority for the NHS IT implementation programme. In contrast, the US healthcare IT strategy is geared towards changing the historically low rates of EHR use in primary care.65 The purposes for which EHR are used also tend to vary by user group, which is reflected in the different functionalities adopted. Box 6.4 provides examples different EHR components commonly used by healthcare staff (doctors, nurses, radiologists, pharmacists, laboratory technicians and radiographers), and by patients and their carers.12 As discussed in due course, other users are more concerned with the ‘secondary uses’ of aggregated or linked and anonymised data for health surveillance, planning, audit or research. 181 Box 6.4: Components of EHRs used by different clinical stakeholders Nurses: daily charting, medication administration, physical assessment, admission nursing notes and nursing care plans. Doctors: referral, present complaint, e.g. symptoms, past medical history, life style, physical examination, diagnoses, tests, procedures, treatment, medication and discharge. Pharmacists: medication. Secretarial staff: procedures, problems, diagnoses, findings and immunisation records. Multi-professionals (nurse, doctor, laboratory staff, radiology staff, clerk or administrative staff, pharmacy personnel, healthcare professionals): referral, present complaint, e.g. symptoms, past medical history, lifestyle, physical examination, diagnoses, tests, procedures, treatment, medication, discharge, administration of medication, admission nursing note and daily charting. Patients: history, diaries and test results. Parents/carer: history. Source: Adapted from Häyrinen K, et al, 200812, with permission Patient information can be stored in different formats in different systems. Systems that require data to be coded using agreed taxonomic schemes prior to entry can facilitate Health Information Exchange and interoperability. Box 6.5 illustrates the main international terminologies applicable to different data types. Data may also be entered into EHR as free text. This may be important for describing complex cases or issues for patient handover, however ‘secondary uses’ of this type of data are far more difficult to realise. Future developments in natural language processing may help to bridge this gap. Box 6.5: Examples of international terminologies used for coding data in EHRs Data component International terminology Diagnoses International Classification of Diseases (ICD) Read codes International Classification of Primary Care (ICPC) Oxford Medical Information Systems (OXMIS) International Classification of Primary Care (ICPC) 182 International Classification of Health Problems in Primary Care (ICHPPC) Procedures Current Procedural Terminology (CPT) Medication Anatomical Therapeutic Chemical Classification Index (ATC) Pathological findings Systematized Nomenclature of Medicine (SNOMED) Nursing problems North American Nursing Diagnoses (NANDA) International Classification of Nursing Practice (ICNP) Nursing interventions Iowa Nursing Intervention Classification (NIC) International Classification of Nursing Practice (ICNP) Nursing outcomes Iowa Nursing Outcome Classification (NOC) Source: Adapted from Häyrinen K, et al, 200812 and Thiru K, et al, 200323; with permission The long-term vision for a fully integrated (cross-sectoral), longitudinal (“cradle to grave”) patient record, supported by universal data and interoperability standards and high-level communication and decision support technologies is shared by England’s DH,59 in common with agencies such as the US Institute of Medicine, although this is still some way from being realised.60 The current DH commitment to paving the path towards this “information revolution” is summarised in Box 6.6 below. 183 Box 6.6: England’s Department of Health has committed to: “move • away from information belonging to the system, to information enabling patients and service users to be in clear control of their care; • away from patients and service users merely receiving care, to patients and service users being active participants in their care; • away from information based on administrative and technical needs, to information which is based on the patient or service user consultation and on good clinical and professional practice; • away from top-down information collection, to a focus on meeting the needs of individuals and local communities; • away from a culture in which information has been held close and recorded in forms that are difficult to compare, to one characterised by openness, transparency and comparability; • away from the Government being the main provider of information about the quality of services to a range of organisations being able to offer service information to a variety of audiences; and • in relation to digital technologies, away from an approach where we expect every organisation to use the same system, to one where we connect and join up systems.” Source: Department of Health, Liberating the NHS: An Information Revolution, executive summary;59reproduced with permission As already noted, electronic record systems are well-embedded in some parts of the NHS, particularly in general practice. Almost all general practices in the UK are now computerised, with many practices now ‘paperless’ or ‘paper-lite’. Such systems are available from many vendors and although there is some variation in functionality between these, all primary care systems generally cover seven broad areas (see Box 6.7). 184 Box 6.7: Classes of data captured within EHR systems in UK primary care • Administrative and demographic information, such as name, age, date of birth, sex, NHS number, address with post code, telephone number, and increasingly ethnicity and end-of-life care preferences. • Information on prevention of disease and screening, such as immunisations, cervical screening, smoking status and alcohol intake. • Allergies and adverse reactions. • Free text entered during consultations. • Clinical and diagnostic information in the form of READ and OXMIS coded data. • Measurements such as height, weight, Body Mass Index (BMI), and blood pressure. • Issuing of prescriptions. • Results of laboratory and radiological investigations. • Image files of documents scanned after being received from external agencies such as hospitals and social services. Although most primary care data in the UK are held by practice-based systems, certain patient information may also be extracted and stored in larger databases, such as the NHS ‘Spine’ on which the Summary Care Record is stored, or in research-focused repositories such as the General Practice Research Database (GPRD), QRESEARCH and the Doctors Independent Network (DIN) database in the UK, and the MediPlus database in Germany, etc. In addition to general practice computing systems, detailed clinical information (Detailed Care Records) may also be held in specialist clinical databases and disease registers based in secondary and tertiary care, with more basic information being held in hospital patient administration systems (PAS). Such records have historically been maintained separately, due a combination of incompatible systems (which the interoperability addenda is seeking to address), different data formats and taxonomies (addressed by the coding standards strategy) and the absence of a single, reliable means of recording and sharing current patient demographic information.9 The latter is being addressed through the introduction of the NHS Number—a unique patient identifier, which will be applied to all healthcare transactions in order to facilitate data linkage and thus integrated 185 care records. Scotland has had this facility for some time, enabling the production of a national integrated care record for diabetes. 6.2.3 Scope of EHR At the heart of the EHR are individual patient records–often referred to as EPR. These vary on multiple dimensions, including level of detail (from summary to detailed care records), data source (single- or multiple-provider) and timeframe (e.g. episodic or longitudinal). As well as patient histories and details of recent care (which may include free text and diagnostic codes) these records may incorporate digital images and scanned documents. The broader EHR also includes non-medical data relevant to healthcare administration and/or planning. Increasingly, EHR are incorporated within complex systems that integrate a range of other functions for supporting communication, decision-making and task management, in addition to documentation (e.g. order entry, clinical decision support, clinical messaging, results reporting, scheduling and referrals). In some countries functions for billing patients or insurers are a key part of EHR systems. Although this is not the case in the NHS, relevant functionality addresses financial requirements around local commissioning of services and national incentive schemes linking payments to preventive activities and evidence-based care processes. In the UK, single-provider EPR are still the most common format of clinical records, these being generated, for example, in general practice and hospital clinics . However, the advent of the NHS Care Records Service (NHS CRS)61 as part of England’s National Programme for Information Technology (NPfIT) (see Chapters 3 and 18) promises increasingly greater integration through, for example, the SCR.33 These EHRs can also aid patients more directly, by making information about their health much more readily accessible, thus allowing them to become more active in the provision of their own healthcare and the improvement of their health.62 186 The NHS in England envisages a patient-centric, integrated, record, designed to improve access by health and social care professionals where and when they are needed, provided they have legitimate relationships. As noted previously, the English strategy for EHR also involves giving individuals secure access to a summary of their own health record via the Internet, using the HealthSpace personal health organiser application, akin to the Personal Health Record concept illustrated earlier in this chapter. HealthSpace is set within a complex architecture of centralised (summary) and localised (detailed) care records, as described in more detail in Chapter 3.33 47 By May 2010, 1.5 million SCR records were created, out of 29.8 million people (73% of population aged >16 years) who were informed of their SCR creation unless they opted out, in 113 of 152 primary care trusts (PCTs) in England.63 In contrast, in the US, Kaiser Permanente has reported much greater usage of its customisable portal “My Health Manager”, which allows members to access part of their records, including medications, past health visits, main diagnoses, allergies and immunisations and also facilitates the sending of email communication with their doctors.64 65 As already noted, while the EHR lies at the heart of IT implementation plans in healthcare systems around the world, EHR implementation varies across countries depending on priority areas and the expected benefits, identified by the respective national governments.66 While some countries have focused on primary care (e.g. Denmark, New Zealand and Spain), others have concentrated on secondary care (e.g. England and China) and others on emergency care (e.g. Scotland and The Netherlands).61 National approaches for healthcare information exchange also vary across countries, whereby system standardisation is promoted in England but in other countries systems interoperability standards are encouraged e.g. Canada, Hong Kong and the US. Box 6.8 summarises examples of international EHR strategies considered in a House of Commons Report on Electronic Health Records.67 187 Box 6.8: Examples of countries embarking on important EHR projects • Canada: A Private Lifetime Record for each citizen is being created and is being co-ordinated by Canada Health Infoway: “By 2010, every province and territory and the populations they serve will benefit from new health information systems that will help transform their health care delivery system. Further, by 2010, the electronic health records of 50 per cent of Canadians and by 2016, those of 100 per cent of Canadians, will be available to their authorized health care practitioner.”68 • France: A Dossier Médicale Personnel69 was passed in June 2004, which will include a collection of health information to be viewed online. Patients will have their own access and will legally own their record. Over a million Dossier Médicale Personnel are created and internet based (beta national version) Rollout to citizens aged over 16 years and covered by the State Health Insurance System will take place 2011. The record will include diagnoses and treatment details provided by public and private health professionals. Clinical images are to be added in future. • United States: Several integrated electronic records systems already exist and in 2004, President Bush established that EHRs would be available for “most” US citizens by 2014. President Obama’s administration provided $19 billion as a stimulus fund to hospitals and healthcare facilities to digitise patient records and for the “meaningful use” of IT. Source: Adapted from House of Commons Report (2007) 67; permitted under open license agreement for non-commercial research The primary use of EHR is to support direct patient care through the effective management and timely provision of information about individual patients’ health history, conditions and treatments. EHR that integrate clinical decision support also facilitate evidence-based practice, while integrated and interoperable EHR systems support HIE and interdisciplinary care. Primary uses also include the support of administrative and financial processes, as well as uses by patients for self-management. In addition to supporting healthcare delivery, electronic data and databases derived from EHR have many secondary uses; for example they can improve the administration, management and quality control of healthcare through facilitating clinical audit and the monitoring of service use; contribute to public health by identifying inequalities and trends; support patient safety through pharmaco-vigilance and disease surveillance; aid population-based research into the impacts of new health technologies or policies and provide a source of data for scientific studies of disease aetiology, genetic influences or drug effects (see also Chapter 8).70 71 72 73 74 188 6.3 Theoretical benefits and risks of EHR 6.3.1 Theorised benefits The substantial international investment in EHR and related eHealth applications is predicated on the assumption that the adoption of accurate and readily accessible electronic records will improve the quality, safety and efficiency of healthcare (Table 6.1). Table 6.1: Key theoretical benefits of electronic health records Attribute Benefit Immediate and universal access to Increased efficiency (e.g. reduced time the patient record spent pulling charts, and duplicate history-taking etc). Increased quality (better information at the point of care) Easier and quicker navigation More efficient point of care assessment through the patient record and data abstraction Increased legibility and Better quality information to aid clinical comprehensiveness, through decision-making and shared care; fewer computer-aided history taking errors in patient management (e.g. missystems and better formatting (e.g. prescribing) templates) Secure record keeping No lost records, fewer unnecessary waits or missed appointments, aiding informed patient care. Patient satisfaction. Standardisation of care among Through better recording and sharing of providers within the organisation information and linkage to computerised decision support systems (CDSS) Reduction of paperwork, Removes duplication, reduces documentation errors, filing activities processing time, decreases personnel costs Coding efficiency and accuracy Improved data quality Alerts for medication errors, drug Safer patient care interactions, patient allergies Ability to electronically transmit Fewer delays, more efficient and information to other providers integrated patient care. Enhanced (assessments, history, treatments patient satisfaction. ordered, prescriptions, etc.) Availability of non-clinical data Easier management of costs, performance and workflow Availability of data for research With downstream benefits for patient care Source: Healthcare Information and Management Systems Society (HIMSS) 2003;75 reproduced with permission The fundamental premise is that clinicians require comprehensive and accurate data on patients at the point of care if they are to provide high quality health services. Electronic health records can aid this objective by dispensing 189 with the need to use difficult to access, and often illegible, paper-based records. The improved record keeping, legibility and access afforded by EHR can help to avoid the inconvenience of lost records, data recapture and reentry and mitigate the risk of recording and prescribing errors, thus improving the appropriateness and safety of patient management. By facilitating the secure exchange and sharing of patient information, an EHR can support shared and continuous care both within single healthcare episodes (e.g. hospital admission and supported discharge) and over time (as in the management of long-term illnesses). An EHR with integrated decision support can help providers improve the quality of clinical decision-making, guideline compliance, aid preventive care and potentially reduce utilisation of services and costs of care (see also Chapters 8 and 10).76 77 78 The success of EHR can be assessed by the following criteria, as proposed by Hayrinen et al,12 using the framework originally developed by DeLone and MacLean.79 “System quality assesses the information processing system itself, and its attributes ... include ease of use, ease of learning or usefulness of system. Information quality measures both the output and input of the information system; attributes…here include completeness, accuracy, legibility, reliability and format. Information use measures end-users’ consumption of the output of an information system, with attributes…including amount of use and number of queries. User satisfaction measures the end-users’ response to the use of the output of an information system, and attributes … include overall satisfaction and decision-making satisfaction. Individual impact measures the effect of information on the behaviour of the end-user, and attributes …include improved individual productivity and information understanding. Organizational impact measures the effect of information on organizational performance, and its attributes….include return on investment and increased work volume...”12 190 6.3.2 Theoretical risks of EHR It is important to recognise that the introduction of an EHR may also have unintended consequences, some of them negative. As illustrated in Chapter 4, the EHR and other eHealth applications have the potential to introduce error and risk at many levels, including error associated with the user (e.g. via inappropriate system operation and interpretation of outputs), the technology (e.g. poor interface usability leading to missed readings or alerts; system or network unreliability leading to missed or delayed exchanges; cyber-attacks), and the underlying knowledge bases (e.g. faulty decision support algorithms). One of the major sources of risk is related to the quality of the data held within an EHR. As previously indicated, EHR lie at the heart of eHealth systems, with related functionality such as ePrescribing, decision support, and computerised order entry, utilising the data in the EHR. The correctness and completeness of the data stored in the EHR is thus a key factor in ensuring that it will be used effectively. There are currently large variations in the accuracy and completeness of information recorded in EHR by clinicians and associated administrative staff. For example, in the UK, disease prevalence and specialist referral rates, based on information derived from an EHR, vary widely between general practices. The scale of the variation is so large that a significant proportion is likely to be due to differences in the accuracy of the data recorded.80 A second issue with data quality arises with the information stored in ‘legacy’ paper records. Unless this information is summarised and entered onto the EHR system, the EHR will not give an accurate and complete record of a patient’s medical history. Patients’ access to their EHR has the potential to mitigate some of these risks by enabling individuals to identify errors in their record. In the UK record access systems which allows patients to view their GP record online, NHS HealthSpace, which provides a portal to the SCR and disease-centred applications offer potentially important benefits in this direction. As with the use of computers in general, the use of EHRs in the context of the clinical consultation also as potential to influence patient perceptions of care quality and the clinician-patient relationship, 191 6.4 Empirically demonstrated benefits and risks of EHR: a summary of the evidence We have identified and appraised 21 reports of 20 systematic reviews (SRs) addressing the associations between EHR and healthcare quality and safety (see Appendices 4 and 5) and these comprise the knowledge base for this section,11-23 27 81-86 of which Clamp 2007 is a second publication,14 in a journal, of the same piece of work that had been published as a report in 2005.87 These SRs cover a broad range of outcomes including prevalence of adverse drug events, compliance with guidelines, consultation time, documentation time of nurses, doctor-patient relationship, quality of morbidity coding in primary care, data quality, exchange of information, costs, process change and a synthesis of all the disparate findings. They also include a SR of barriers perceived by doctors to adopting EHR, which can be intervened with IT.11 16 6.4.1 Impact on clinically-relevant outcomes Clinical endpoints Relatively few SRs have looked at the impact of EHRs on clinical endpoints, reflecting the paucity of primary studies which have done so. Delpierre et al.’ included studies of patient outcomes in their 2004 SR of the impact of computerised patient record systems on medical care quality.17 They found that, overall, EHR use did not improve measures of cardiovascular risk or clinical outcomes of acute respiratory distress syndrome, although one study included in their review reported a significant decrease in diastolic blood pressure,17 When outcomes were assessed by patients themselves, no clear benefit was found in relation to the management of asthma, angina or depression. It is worth noting, however, that all of the relevant studies had included the EPR as part of a more complex intervention involving clinical decision support and it is therefore not possible to attribute the effects to the EHR alone. Given the paucity of SRs evaluating the impact of EHRs on morbidity or mortality, any claims about their clinical effectiveness need to be viewed with caution. While a more detailed review of primary studies is warranted to shed 192 further light on this important question, there are likely to be few studies which have robustly evaluated these impacts in light of the distance in time between when EHR implementation takes place and the emergence of clinically or statistically significant changes in measures of hypothesised clinical endpoints. In other words, the size and length of study required to convincingly demonstrate impacts of EHR on mortality and, to a lesser extent morbidity, are far larger than a typical academic research grant would support. Instead it is likely that the most convincing evidence of this kind will be revealed if large-scale implementations are set up with evaluation as a key objective, so as to enable longer-term changes to be captured. Most of the evidence captured by the systematic reviews on which this report draws is therefore concerned with intermediate or surrogate outcomes associated with care quality or processes or user perceptions of benefit. These are summarised next. Surrogate endpoints: preventive care A comprehensive SR by Chaudhry et al. revealed benefits of EHRs in the form of increased adherence to guidelines or protocol-based care, especially in the domain of preventive health.13 The effectiveness of EHRs was demonstrated in terms of improved monitoring and surveillance, reduction of medication errors and decreased utilisation of inappropriate or potentially redundant care. However the authors cautioned against generalising from these results, as the majority of studies demonstrating such benefits had taken place in centres of excellence using ‘home-grown’ systems that had been extensively customised to local needs and practices.13 In contrast, the majority of institutions procure commercial systems with considerably less opportunity for local tailoring. A SR by Jerrant et al., focusing mainly on primary care studies published between 1996-1999, found evidence that using EHR in the context of screening interventions improved compliance with certain tasks by both patients and providers.18 193 Delpeirre et al, in a SR mainly based in secondary care, found EHR to be associated with increased pneumococcal vaccination (from 1% to 36%), influenza vaccination (from 1% to 51%), prophylactic heparin (from 19% to 32%) and aspirin prescription (from 28% to 36%).17 6.4.1 Impact on process indicators Satisfaction Shachak and Reis’ SR on the effects of EHR on doctor-patient communication draws primarily on data from primary care.19 The authors identified four key themes in the literature, including the use of a computer for EHR by the doctor during consultation; the impact of EHR use on doctor-patient communication; factors affecting doctor-patient communication in EHR settings; and behavioural styles of doctors that affect their interaction with patients.19 In the studies covered, computers were used by doctors to look up patients’ general medical records, to check patients’ medications, to retrieve test results, to enter information and write prescriptions and letters, and at times to demonstrate to patients changes in clinical data over time and to provide printouts to inform patients. Compared with paper-based systems, patients in clinics with EHR received more communication with respect to their medication, in both written and oral format, but received less information on medication indication, which led to more questions to doctors about medicines. Electronic health records have been shown to affect patient’s interaction with their doctor. In the studies reviewed by Boonstra et al., doctors spent on average 25-50% of the consultation time looking at the computer screen.16 Computers have also in some cases been perceived by doctors as a physical barrier to effective doctor-patient interaction.17 Different approaches to managing data during the consultation were also reported; for example some doctors entered data during the consultation, others did so after the patient had left and some did not type into the computer at all, but recorded notes on dictating machines which were then transcribed into the record by clerical support staff. It was found that some doctors sometimes used their computer to send non-verbal cues to patients, by looking up at the screen, holding the 194 mouse or typing, to change the topic or flow of conversation and to indicate an end to the meeting. Doctors who looked less at the screen and more at the patient, were more prone to ask psychosocial questions and to respond emotionally to their patients.19 However some doctors using computers had improved verbal and non-verbal communication with patients, and could furthermore manage visits and data better, despite having more to do in terms of computer use and patient management.19 The layout of the doctor’s clinic can also have an impact on the doctor’s use of EHR. For example, the SR by Shachak et al. found that when computers were fixed in one place, doctors had to turn their body or move their chair and thus eye contact with patients was hampered.19 It was in some cases possible to obviate some of these adverse effects by repositioning the screen or using flat monitors on mobile arms. According to Shachak et al, once successfully implemented, patients in primary care report being satisfied with the corollary gains associated with EHR use such as improved exchange of information between healthcare professionals;19 Others have noted indifferent patient responses to EHRs.86 A SR of predominantly secondary care-based studies found that patients were usually satisfied and unlike doctors, did not perceive the computer to be a physical barrier.17 Whilst interesting, these data should be interpreted with caution, as satisfaction levels can be difficult to compare across studies conducted in a range of settings and studying an array of computing devices; furthermore, the fact that patients in general tend to rate doctors highly is an important potential confounding variable.86 Overall, the evidence suggests that doctors tend to have a mixed response to EHR use, but generally patients are satisfied. It is difficult from this body of evidence to assess the extent to which any clinician sense of dissatisfaction is temporary, reflecting the problem of poor early integration that is widely recognised in studies assessing health informatics innovations in the early post-implementation phase. Improved training in consultation skills, particularly in relation to the use of computers in an inclusive way, and more 195 attention to screen sizes and room layout could help to obviate some of the adverse effects associated with use of EHR. Impact on professionals’ time In a SR that synthesised the findings from 23 studies, predominantly from secondary care, Poissant et al. found that documentation time was decreased for nurses, but not doctors. This was thought to be related to differences in the types of information the two groups record and their different approaches to data recording. Importantly these differences did not attenuate with increased or more established use.15 Furthermore, the time taken for doctors to retrieve information was found to increase.15 A more recent SR by Thompson et al, but this time focusing on nursing efficiency, found a more mixed picture in relation to nurses’ documentation time.20 It appeared that any overall reductions in documentation time were negated by increased patient specific documentation time.20 Somewhat unexpectedly, studies that have looked at the impact of EHR on process times shortly after system implementation, have tended to find a reduction in documentation time, whilst studies with a longer time period between implementation and evaluation have tended to find the reverse.15 Whilst the reasons for this are not entirely clear, this could reflect the introduction of greater functionality of the EHR system over time, and possibly users’ greater willingness to engage more meaningfully with the more integrated record systems that develop over time. Overall, it seems that time savings are by no means guaranteed and it may lead to disappointment if such a promise is used as a means of encouraging EHR implementation.15 The impact on time may also vary between professional groups, with evidence suggesting that the time trade-offs may be particularly unfavourable for doctors. 196 6.4.3 Impact on data quality A high standard of data quality is essential to deliver the promises made relating to EHRs. The most commonly used measures of validity found in the SR by Chan et al. were correctness and completeness, which were defined thus: “Data completeness is defined as the level of missing data for a data element. Data accuracy is conceptualized as the extent to which data captured though the EHR system accurately reflects an underlying state of interest (e.g., whether a medication list accurately reflects the number, dose, and specific drugs a patient is currently taking). This includes the timeliness or “currency” … of the data as well as the clinical specificity or “granularity” in how individual clinicians or clinical staff document within an EHR system. A high level of missing data and/or data errors will substantially reduce measure reliability and validity.”83 In a review of a substantial body of international studies (n=89) covering the period 1982 to 2004 and conducted in mixed settings, mostly in secondary and tertiary care, Hayrinen et al found that the information systems studied helped clinicians to record information more completely.12 Although documentation varied across record fields, these EHR systems enabled more detailed recording. Documentation was judged to be accurate, even when data were entered by patients. Although EHR also enabled more comprehensive recording, there were still shortcomings when records were judged against professional regulations and guidelines.12 Overall, their SR concluded that EHR can improve the comprehensiveness and accuracy of data recording, but in some studies the overall quality of data recording remained sub-optimal. As discussed above, time trade-offs may also be encountered with use of EHR systems resulting in an increase in recording time. Thiru et al, systematically reviewed studies examining the scope and quality of data recording in primary care, published between 1980 and 2001 (31 out of 52 were from the UK). This revealed that data quality has been measured 197 in different ways, most commonly using comparisons of rates derived from the EHR with an external standard.23 Data validity was examined using a range of terms (e.g. completeness, correctness, accuracy, consistency, and appropriateness), but the precise terms were often undefined in many studies.23 The most commonly used measure of validity was sensitivity or the proportion of total cases of a disease or events such as prescriptions that were recorded by the EHR system (i.e. completeness of recording). Prescriptions had the highest rate of recording (sensitivity), probably because prescribing is a core function of many EHR systems and is used even in circumstances when EHRs are not used for other functions such as morbidity recording.23 Positive predictive values were generally high for all recorded elements. As in most studies the component of EHR (numerator) and the component of the reference standard (denominator) were not clearly stated and, where these were specified, there were inconsistencies between studies leading the authors to recommend that in future data quality studies the numerator, denominator and confidence intervals should be clearly stated.23 Thiru et al further stated that as verifying data would not be possible with disappearance of paper notes in EHR practices, the triangulation method (EHR data, clinical diagnoses and/or notes) could help to ascertain the true quality of data. This SR also suggested that terms used to measure data quality, such as accuracy and completeness, should be clearly defined to minimise the risk of ambiguity. This issue has also been highlighted in a SR by Jordan et al, where data quality in EHRs in UK primary care was assessed from studies published up to 2002, using the following four criteria: (1) completeness–if there is a morbidity code every time the patient visited the GP; (2) correctness–if appropriate codes were given; (3) completeness of a morbidity register–if relevant patients are included in the register; and (4) correctness of the morbidity register–if patients on that register should be there. While Thiru et al. also included non-UK studies using any gold standard for data quality, Jordan et al. looked at UK primary care data quality studies only referring to specific gold standards.22 As with the findings from Thiru et al’s SR, recording 198 of consultations was generally high (typically greater than 90 per cent) in the 24 studies, but assigning a morbidity code during each consultation was more variable (66-99% complete). That the results from Jordan et al. are similar to those reported by Thiru et al. is not altogether unsurprising given that these SRs largely reviewed the same evidence-base. Jordan et al. however also pointed to the observation that quality of recording was higher for conditions with clear diagnostic features such as diabetes when compared to conditions with possibly more subjective diagnostic criteria such as asthma.22 Coronary heart disease was the most commonly assessed condition recorded in disease registers in previous studies and completeness of recording was generally moderate (typically around 70%).22 The positive predictive value of coronary heart disease registers was generally high (typically around 83–100 %).22 Other diseases that were examined in the 24 studies (such as asthma and epilepsy) showed similar patterns of completeness of recording and positive predictive value of recorded diagnoses, but rates were generally lower than for coronary heart disease.22 Herrett et al. examined the methods used to validate data and the quality of the validations between 1987-2008 in the GPRD, which is one of the largest primary care databases in the world. They found 212 publications that reported 357 validations to verify 183 distinct diagnoses. Eighty-five percent (n=303) of validations were done outwith GPRD, by sending questionnaires to GPs and requesting anonymous patient records from GPs or by comparing rates of disease with a non-GPRD UK-based data source. The remaining validations were done using GPRD by comparing diagnoses with codes, reviewing manually the free texts on anonymised records or by sensitivity analysis. These validation exercises revealed that in most studies a high proportion of cases were confirmed as having the diagnoses, with an overall median of 89% (range 24-100%). The estimates of incidence and prevalence of diseases in GPRD were comparable to other UK population-based sources, except in a few cases; for example, the incidence rate of rheumatoid arthritis from GPRD was 50% higher than previous estimates88 while prevalence estimates of musculoskeletal diseases in GPRD were lower in comparison to 199 other GP databases.89 This could be due to presence of clearer case identification in the former than in the latter, as Jordan postulated. There were however a number of potential limitations to these attempts at validating recorded data, these including the small sizes of the samples used for validation, missing data because of transfer and/or death, and in some cases difficulties with ascribing codes to some diagnoses. Completeness and correctness of coding has also been shown to be dependant on the level of enthusiasm of the GPs and practices and GP’s preference for certain codes.22 Like Thiru et al., Jordan et al. noted the lack of well-defined data quality standards and the need to correct this if better measurement of data quality in primary care EHRs was to be established. An older SR by Hogan and Wagner, covering the period 1978 to 1996, mostly reviewing studies undertaken in US secondary care (in contrast to the Thiru et al., Jordan et al. and Herrett et al. SRs, which were primary care-focused), drew similar conclusions and noted the wide variability in the accuracy of EHRs and the lack of standard measures of data accuracy, as was also reported by Hayrinen et al. Errors in EHR data did not just concern data entry; rather, the use of EHRs at various points in the care pathway made it prone to multiple errors and it was thus insufficient to implement a single intervention in an attempt to improve data accuracy.27 Hoerbst and Ammenwerth published a systematic review of quality requirements for EHR in 2010, based on both published literature and expert interviews. They considered that both functional and non-functional requirements of EHRs could potentially create problems before, during or after EHR implementation.84 A repository of 1,191 possible requirements were assigned to 59 categories and sub categories.84 Functional categories were regarded as having the highest quality requirements, followed by data security and content (22.8%, 22.7%, 19.0%, 12.6% respectively), whilst non-functional categories such as portability, performance, maintainability, etc, were regarded as having the lowest requirements.84 200 Overall, the evidence in relation to data quality considerations indicates that with EHR use there tends to be greater recording of information, but the accuracy of what is recorded can vary quite considerably. This partly depends on the ease with which it is possible to detect and classify diseases but also on doctors’ preferences for using certain codes. Organisations such as EuroRec in Europe and the Certification Commission for Health Information Technology (CCHIT) in the USA are working towards establishing EHR quality certification, to promote and support data quality improvement. 6.4.4 Economic impacts Successful EHR implementations require considerable capital investment and can result in major changes to workflows if these are to be adopted. Boonstra et al. found that worries about the costs of adopting and maintaining EHRs can be a major barrier to their use.16 The SR by Chaudhry et al. on US studies between 1995 to 2005, found little information on basic cost or Return on Investment, and estimating cost-effectiveness remains largely reliant on statistical modelling, which is not usually available beyond the research arena.13 Despite reviewing a comprehensive body of studies, the authors concluded that there was insufficient evidence available to guide stakeholders on the financial effects of moving to EHR systems. In a recent update to this influential report, which covered the period 20042007, the authors reported that although patient focussed applications are rapidly growing in the healthcare service, little economic evaluation has been undertaken.90 They suggested that greater partnership between public-private organisations, policies on financial incentives and a robust evidence base for EHR implementation are all important prerequisites to speed up the adoption process. In an earlier SR published in 2007, Clamp and Keen examined the value of EHR reported in studies between 1998 and 2005. They also found limited clear evidence on the economic outcomes of EHR systems.14 They urged caution when interpreting pronouncements in relation to the health economic benefits of EHR. Even the studies that they regarded as being of reasonable 201 quality, and had presented cost data, had not included health economists in the research team. There was technically no sound evidence about cost changes associated with the EHR, apart from a paper by Bryan et al. on PACS.91 Most papers failed to include all reasonable costs, irrespective of whether they would tend to increase or decrease overall cost changes associated with EHR use.14 Most papers made linear projections of costsavings in future years, typically up to five years ahead.14 However, there was often no explicit basis for these projections.14 There were no studies which actually sought to capture all of the costs and benefits associated with an EHR at the level of a process or within a single hospital setting.14 Although there is some suggestive evidence that information networks lead to greater economies of scale and scope in non-health care fields and interoperability can enhance EHR efficiency, no good studies were found in this SR that reported on this aspect.14 In contrast, the 2008 SR by Uslu et al, which reviewed mainly US studies between 1991 to 2003, concluded that there were economic benefits from reductions in time spent in administrative work, some savings in documentation costs or savings in nursing costs or savings in non-personnel costs.85 Bearing in mind the various caveats stated above, particularly in relation to the difficulties in ascribing costs, the claims made in this somewhat uncritical review however need to be interpreted with caution. With recent policy developments, more indirect costs and economic benefits also need to be considered. For example, in UK primary care, financial incentives, driven by the Quality and Outcomes Framework (QOF), have encouraged the recording of clinical processes and intermediate outcomes using EHR. While this is acknowledged to have improved certain measures of quality, there has been some evidence of the use of gaming strategies to maximise payments and attention has been paid to methods of ameliorating the problem of ‘perverse incentives’.92 Of more direct relevance to EHR is the introduction of the “Meaningful Use” criteria in the US, developed as part of the Health Information Technology for Economic and Clinical Health Act (HITECH). Under this scheme, stimulus funds are being employed to 202 incentivise the ‘Meaningful Use’ of ‘certified EHRs.93 This releases payments to Medicare and Medicaid providers for capturing electronically coded clinical data that will “enable tracking of key clinical conditions, communicating that information for care coordination purposes, and initiating the reporting of clinical quality measures and public health information.”94 While the Meaningful Use incentives are expected to have many benefits, the US Department of Health and Human Services has taken the precaution of commissioning an expert panel to consider their possible unintended consequences (negative and positive) and methods of addressing these, in the context of the HITECH act.95 Overall, there remain very few primary studies and SRs that have formally studied the cost-effectiveness of EHR. The AHRQ, and its update, remains the most comprehensive attempt to study this question. The authors found that the current evidence-base has emerged from a limited number of benchmark institutions that have evaluated “home-grown” EHR systems, which were developed over a period of time iteratively to suit the purpose. This limits the generalisability of the results, as many institutions do not have the means to develop their own EHR and so will be implementing commercial systems. Even when studying these benchmark institutions, it has not been possible to demonstrate clear return on investments. The AHRQ has the following caveats as in Box 6.9. 203 Box 6.9: Important caveats outlined by the Agency of Health Related Quality • All studies are predictive analyses that are based on many analytical assumptions and limited empirical data. Hence, the strength of evidence is often weak. • In all studies, the EHR system was assumed to have multiple functionalities and the functional capability of the EHR system is critical to the benefit that may occur from using it. • The organisations that were the subjects of four studies were all large. The literature review did not identify cost benefit studies for EHR implementation in small organisations. • The costs of implementing an EHR system may be underestimated. Only one of the five cost benefit analyses included the cost of the implementation process, and found that this cost was 1.5 times the cost of the EHR system. • Implementing an EHR system requires extensive changes in the organisational processes, individual behaviours, and the interactions between the two. These resulting costs are often omitted or not reported from studies but can be substantial. • The financial benefits depend on the financing system. As shown in the sensitivity analysis of one study, the benefit estimates are most sensitive to the assumption of the proportion of capitated patients. Realising all quantifiable benefits of EHR implementation would require changes to the current healthcare financing system. • Both the cost and the benefit of attaining interoperability among EHR systems are directly proportional to the level of data exchange achieved. For example, although the cost of achieving machine-organisable or machine-interpretable interoperability is greatest, it offers the most potential for increased efficiency, improved healthcare utilisation, and reduced costs. Adapted from: AHRQ (2006)78 (permission to reproduce applied for) 6.4.5 Human and organisational factors affecting the adoption of EHR Realising the potential efficiency and quality gains of EHR is dependent on their effective implementation and adoption, which can prove challenging (see also Chapters 17 and 18).13 16 96 A SR by Boonstra and colleagues, mainly covering studies conducted in secondary care settings, identified eight barriers, namely: 1) financial, 2) technical, 3) time, 4) psychological, 5) social, 204 6) legal, 7) organisational type and 8) change process, of which 1-3 were more often identified than the rest.16 Financial, technical and time barriers were classified as primary barriers while the remainder were categorised as secondary barriers.16 According to the authors, organisational and change process barriers are the “mediating factors in the success of EMR project”, as the former determine the importance of other barriers (1 to 6 above) before EHR adoption, while the latter can mediate the effect of the other (1 to 6 above) barriers during the EHR implementation process.16 The technical barriers faced by doctors were further explored in a SR by Castillo et al, with some suggested interventions.11 Technical barriers were typically a lack of data standards that permit effective, accurate and timely exchange of electronic clinical data between healthcare providers.11 The six enablers of EHR adoption identified from the studies reviewed were: user attitude towards information systems, workflow impact, interoperability (compatibility), technical support, communication among users and expert support.11 These factors were influential irrespective of practice/clinic size, type of setting (inpatient or outpatient), primary or secondary care, or national policy.11 The authors suggest that if EHR usage was logged and interpreted with reference to the above factors, tailored assistance could be provided; for example by providing access to frequently asked questions, notification messages, electronic bulletins, electronic workflow and process documentation, documents on standards, support and knowledge directories, social networking, etc.11 Strategies for addressing such barriers, suggested by Boonstra et al in 2010, are summarised in Table 6.2. Other barrier-reduction strategies include the certification and standardisation of applications and systems to permit exchange of electronic clinical data; removal of legal barriers; and greater security of the medical data stored in EHRs.11 205 Table 6.2 Perceived barriers to EHR adoption and related interventions Perceived barriers A Finance B Technical Possible barrier-related intervention strategies • • • • • • • • C Time • • E Social • • • • • • F Legal • • G Organization H Change process • • • • • • • • Provide documentation on return on investment. Show profitable examples from other EMR implementations. Provide financial compensation. Educate doctors and support ongoing training. Adapt the system to existing practices. Implement EMR on a module-by-module basis. Link EMR with existing systems. Promote and communicate reliability and availability of the system. Acquire third party for support during implementation. Provide support during implementation phase to convert records and assist Provide training sessions to familiarize users. Implement a user friendly help function and help desk. Redesign workflow to achieve a time gain Discuss advantages and disadvantages for doctors and patients. Information and support from doctors who are already users. Ensure support, leadership, and communication from management. Develop requirements on safety and security in cooperation with doctors and patients. Ensure EMR system meets these requirements before implementation. Communicate on safety and security of issues. Redesign workflow to realize a better organizational fit. Adapt EMR to organization type. Adapt EMR to type of medical practice Select a project champion, preferably an experienced doctor. Let doctors (or representatives) participate during the implementation process. Communicate the advantages for doctors. Use incentives. Ensure support, leadership, and communication from management. Source: Boonstra and Broekhuis, 2010 16; reproduced under open third party use agreement 206 6.4.6 Empirically-demonstrated risks of EHR As noted earlier, different types of human error in the use or interpretation of EHR may generate patient safety risks. These may be associated with “...cognitive overload, loss of overview, errors in data entry and retrieval, excessive trust in electronically-held data, and the tendency to conflate data entry with communication within and between care teams”.21 We did not find a systematic review specifically examining the outcomes of such factors for patient safety. To a large extent, the theoretical risks of EHR to patient safety remain unsubstantiated. However one area of risk has commanded greater attention; namely privacy breaches and their implications for patients. EHR contain sensitive patient information and carry risks to patient privacy due to unauthorised or inappropriate access. Such privacy breaches have potential to harm patients in a number of ways, including the embarrassment of having sensitive medical secrets disclosed, financial implications associated with access by interested third parties such as insurers or employers, or even health impacts, for example if the record were to be falsified and inappropriate treatments administered. . Although there have been few reported cases of patient harm due to data breaches, high profile cases highlighted in the media and the availability of enforceable sanctions, including dismissal, have raised concerns amongst some clinicians about their legal liability. Indeed it has been suggested that some doctors are more concerned about patient privacy and confidentiality than their patients are.16 Concerns over information security and confidentiality are prominent amongst reasons for poor engagement with the EHR (see Chapter 4), as has been evidenced by the strong reaction from the British Medical Association (BMA) and others in NHS England. Although we did not identify any systematic reviews specifically concerned with the impacts of privacy breaches on patients, the topic was explored in a recent SR of HIE in primary care practices by Fontaine et al .97 The authors report that 76% of Americans are concerned about the privacy of their health information, particularly in the unauthorised sharing of ‘super confidential’ information about mental health, chemical dependency and genetics, medical identify theft and fear of 207 discrimination based on health related conditions. They also note the historical resistance of providers to relinquishing patient information to a centralised database and its role in the failure of community health information networks. Whilst acknowledging that federated models of HIE–where records remain with the provider and are shared only when needed–offer advantages for privacy and security, they also note that the greater transparency offered by HIE can make providers reluctant to use EHR for fear of negative exposure (e.g. of non-compliance with guidelines or poor patient outcomes). 6.5 Implications for policy, practice and research Although establishing EHR systems is a key priority for NHS England, prior evaluations have been of variable quality and have shown limited benefits in terms of increased efficiency, patient safety, and quality of care, and little evidence of improved patient outcomes. Large, methodologically robust studies which include sound economic evaluations and well-theorised qualitative studies are needed to fill this evidence-gap. The evidence base on EHR is muddied by the variable use of EHR-related terms and concepts as well as differences in the nature of the interventions described in published research, many of which are largely driven by the clinical decision support functionality that EHR can offer. This complexity can make it difficult to generate theory about the mechanisms through which the intervention will have influence and the types of outcome that are significant, or to back-track from effect to cause when interpreting the results of evaluation studies. As noted earlier, a further source of difficulty is the timescale between intervention and anticipated outcome, which may be too large to demonstrate clinical or cost-effectiveness in the course of a typical research project, although intermediate outcomes provide useful indicators for projection. Likewise longer-term studies may help to establish whether the early benefits noted in some studies are maintained once initial enthusiasm has worn off or where usage of the system has evolved to incorporate additional tasks to those originally envisaged. Integrating a theoreticallyinformed evaluative research agenda into large-scale projects holds the greatest promise for revealing conclusive evidence of longer-term effects and 208 will help to address the so-called ‘evaluation paradox’ that can act as a barrier to progress.4 It also provides greater opportunities to embed contextualised qualitative and observational studies that may help to advance our understanding of the complex and dynamic interactions between users, stakeholders and technologies,21 and of how organisational factors (e.g. workflow management) can influence the realisation of benefits. In considering the value of evidence, it is also worth reflecting on the role of EHR as a marker of a modern healthcare system. As with other aspects of the digital society, the adoption of EHR is inevitable and necessary if we are to move forward to a more connected, knowledge-driven and flexible healthcare environment and there are many associated benefits for providers, organisations, governments and citizens that will never be captured by simplistic Return on Investment calculations or clinical impact studies. While robust scientific evidence is highly desirable and may provide insights into the types of EHR functionalities that may be most useful for meeting particular objectives in particular groups or settings, the limitations of the current evidence-base should not be regarded as an overwhelming barrier to progress, provided that we can operate under the remit of ‘do no harm, but modernise’. Given the current financial climate, however, and the vast cost of implementing such systems across health systems, the need to justify investment will nevertheless remain a priority. We agree with the authors of the AHRQ review in advocating more research that evaluates commercially-procured, or ‘off-the-shelf’ systems, in order to balance the evidence derived from studies of home-grown systems in centres of excellence, which may be atypical. Whilst independent academic studies may be valuable, a requirement for evaluation within vendors’ business plans may be one way of ensuring the routine capture of relevant metrics. We also agree that there may be value in establishing standard reporting criteria for 4 This refers to the fact that it can be hard to motivate the adoption of systems without being able to cite evidence of their impacts, but it can also be hard to generate this evidence without the systems first being adopted. 209 studies of EHR implementation to aid interpretation of results across contexts.78 A recurring theme in the literature is the importance of clinicians’ attitudes as a mediator of effective implementation and benefits realisation, and thus the value of clinician engagement at a strategic and operational level (see Chapter 17). This message has clearly been taken into account within the NHS IT strategy, but it remains worthy of repetition. As discussed in Chapter 17, there is some evidence (mostly from the US) that clinicians from larger medical practices have higher rates of adoption of EHR when compared to smaller practices, possibly reflecting the greater technical support and training facilities available in institutions with larger IT budgets.16 Similar observations have been made in relation to larger hospitals versus smaller specialist care settings16 There are also indications that market forces may contribute to rates of EHR adoption in different settings. For example Abdolrasulnia et al attributed the greater adoption of EHR in urban, as compared to rural ambulatory care practices, to inter-provider competition and the perceived advantages of EHR for relative quality improvement.98 They suggest that policy makers direct financial incentives to areas where providers face fewer competitive pressures to adopt EHR.98 Although such factors will be less influential in nationally-funded health systems like the UK, where primary care computing is widespread, market forces nevertheless play a role in shaping GPs choice of off-the-shelf systems through the perceived functionality or usability of alternative vendor solutions.99 The organisational and workflow changes required to move from paper-based records to EHR are often substantial and support from the organisation, in the form of leadership, encouragement, incentives or training opportunities, can all facilitate EHR adoption; conversely a lack of these can hinder the change process. The introduction of EHR systems may offer indirect benefits for the NHS, including a reduction in clinical errors and improved patient safety, through the 210 associated introduction of CDSS and ePrescribing systems, but further evidence of these impacts is required. EHR systems may also make information more readily accessible by patients, thus enabling them to become more involved in the management of their own healthcare, and this is again an area that warrants further study. The ‘secondary uses’ of EHR data for clinical audit, health surveillance and research also offer considerable potential to improve the quality and safety of care through highlighting compliance with evidence-based practice recommendations, facilitating population-based evaluations of health policies or interventions, supporting the monitoring of health conditions and inequalities, maintaining pharmacovigilance and enabling medical innovation. For example, electronic patient records have been used in recent seminal research employing a range of methods including randomised controlled trials, observational, cohort, and case-control studies; surveys; and qualitative research.70-72 100-107 The impact and economic value of such ‘secondary uses’ is difficult to quantify, but in the case of research leading to biomedical innovation the benefits to the scientific community and the national economy are likely to transcend those for the health system. In addition to ensuring that requests to use EHR for ‘secondary uses’ are appropriate, it will also be important to ensure that access to data for such purposes is minimally clinically disruptive.. Linked to this is the need for further studies to establish the appropriate balance between providing secure EHR to healthcare providers, consumers, primary and secondary users of EHR (see also the safety map in Chapter 4) and the risks of breaches of confidential information. A range of authorisation mechanisms, including role-based access, incorporating cryptographic and biometric techniques are being considered.108 Research to evaluate the impacts and outcomes of these and other privacy-enhancing solutions is warranted. Value would also be gained from further work to link legislative and regulatory frameworks, governing primary and secondary uses of electronic patient records, with evidence on public attitudes towards data sharing, particularly in light of the opportunities and risks presented by cross-sectoral 211 data linkage and market drivers for third party (including) commercial uses. It will also be important to ensure that conventions and procedures for data sharing are built around ethical principles relating to trust, equity and benefits sharing.21 83 Key areas for further work include the development and evaluation of data quality standards for use in EHR, and the evaluation of methods for improving data quality. Accurate and complete data is an essential prerequisite for effective EHR use and this should therefore be seen as a priority for NHS England. For EHR-based healthcare quality assessment, data correctness, completeness, granularity and comparability have to be taken into consideration.83 Further research would be worthwhile to help build an understanding of the contexts in which EHR are used, and to examine which aspects of EHR are perceived favourably or unfavourably by healthcare professionals and why. This should include studies of contextual factors affecting the implementation of large-scale EHR initiatives and rigorous, longitudinal, qualitative studies to examine how EHR is assimilated into routine practice over time.21 109 Understanding how EHR are used to support HIE between within organisations is also important.21 Research exploring the soft and hard impacts of “home-grown” versus “off the shelf” systems and examining under what conditions small EPR systems are fit for purpose, would also be useful.13 21 Additional work to derive methods of incorporating patients’ experiences of healthcare and patient reported outcomes into EHR would also be valuable both in terms of supporting clinician-patient partnership and generating new types of data for clinical or secondary uses. As already noted, there is also a real need for more rigorous longitudinal realtime evaluations of national-scale EHR implementations of the kind that are now either underway or being planned in many parts of the world.61 212 Policymakers worldwide will be interested to observe the outcomes of the HITECH act in the US; particularly the impact of the ‘Meaningful Use’ incentives in promoting adoption and best practice in EHR and in realising predicted benefits for care quality and patient outcomes. This raises the more general need for international-level research on EHR adoption, use and impacts, bearing in mind national differences in healthcare systems, population health needs and technological infrastructure, and the demand for electronic health data and services in a globalised society. 213 References 1. World Health Organization, Global eHealth Observatory. http://www.who.int/goe/publications/goe_atlas_2010.pdf; accessed 26 January 2011. 2. Wikipedia. http://en.wikipedia.org/wiki/Electronic_health_record; last accessed 16 January 2011. 3. International Organization for Standards. Health informatics — Electronic health record — Definition, scope, and context. ISO publication 2005;TR 20514. 4. Department of Health, NHS Executive. Information for health: an information strategy for the modern NHS 1998-2005; http://www.dh.gov.uk/en/Publicationsandstatistics/Publications/PublicationsPo licyAndGuidance/DH_4007832. 5. US National Institute of Standards and Technology, Information Technology Laboratory. Electronic Health Record; http://www.itl.nist.gov/div897/docs/EHR.html; accessed 26 January 2011. 2006. 6. European Commission, Information Society and Media. The conceptual framework of interoperable electronic health record http://www.ehrand ePrescribing systems. impact.eu/downloads/documents/EHRI_D1_2_Conceptual_framework_v1_0. pdf 2008. 7. World Health Organization. Electronic Health Records: Manual for Developing Countries; http://whqlibdoc.who.int/wpro/2006/9290612177_eng.pdf; accessed 26 January 2011. 2006. 8. Healthcare Information and Management Systems Society. Electronic Health Record; http://www.himss.org/ASP/topics_ehr.asp; accessed 27 January 2011. 9. US Office of Communications. EMR vs EHR – What is the Difference? http://healthit.hhs.gov/blog/onc/index.php/2011/01/04/electronic-medicalrecord-vs-electronic-health-record-clarifying-the-ehremrdifference/#comments Accessed 25 March 2011. 2011. 10. ACS/Xerox. Electronic Health Records and Health Information Exchange: The Rx for an Ailing Healthcare System http://www.acsinc.com/wp_electronic_health_records_and_health_information_exchange.as px Accessed 25 March 2011. 2009. 11. Castillo VH, Martinez-Garcia AI, Pulido JR. A knowledge-based taxonomy of critical factors for adopting electronic health record systems by physicians: a systematic literature review. BMC Med Inform Decis Mak 2010;10(60):60. 12. Häyrinen K, Saranto K, Nykänen P. Definition, structure, content, use and impacts of electronic health records: A review of the research literature. International Journal of Medical Informatics 2008;77(5):291-304. 13. Chaudhry B, Wang J, Wu S, Maglione M, Mojica W, Roth E, et al. Systematic Review: Impact of Health Information Technology on Quality, Efficiency, and Costs of Medical Care. Annals of Internal Medicine 2006;144(10):742-752. 14. Clamp S, Keen J. Electronic health records: is the evidence base any use? Med Inform Internet Med 2007;32(1):5-10. 15. Poissant L, Pereira J, Tamblyn R, Kawasumi Y. The impact of electronic health records on time efficiency of physicians and nurses: a systematic review. J Am Med Inform Assoc 2005;12(5):505-16. 16. Boonstra A, Broekhuis M. Barriers to the acceptance of electronic medical records by physicians from systematic review to taxonomy and interventions. BMC Health Services Research 2010;10:231. 214 17. Delpierre C, Cuzin L, Fillaux J, Alvarez M, Massip P, Lang T. A systematic review of computer-based patient record systems and quality of care: more randomized clinical trials or a broader approach? International Journal for Quality in Health Care 2004;16(5):407-416. 18. Jerant AF, Hill DB. Does the use of electronic medical records improve surrogate patient outcomes in outpatient settings? J Fam Pract 2000;49(4):349-57. 19. Shachak A, Reis S. The impact of electronic medical records on patient-doctor communication during consultation: a narrative literature review. J Eval Clin Pract 2009;15(4):641-9. 20. Thompson D, Johnston P, Spurr C. The impact of electronic medical records on nursing efficiency. J Nurs Adm 2009;39(10):444-51. 21. Greenhalgh T, Potts HW, Wong G, Bark P, Swinglehurst D. Tensions and paradoxes in electronic patient record research: a systematic literature review using the meta-narrative method. Milbank Q 2009;87(4):729-88. 22. Jordan K, Porcheret M, Croft P. Quality of morbidity coding in general practice computerized medical records: a systematic review. Family Practice 2004;21(4):396-412. 23. Thiru K, Hassey A, Sullivan F. Systematic review of scope and quality of electronic patient record data in primary care. BMJ 2003;326(7398):1070. 24. Gray J, Majeed A, Kerry S, Rowlands G. Identifying patients with ischaemic heart disease in general practice: cross sectional study of paper and computerised medical records. BMJ 2000;321(7275):1519. 25. Lowrie R, Morrison J, McConnachie A. A cluster randomised controlled trial of pharmacist led Statin Outreach Support (SOS) in primary care: Design and baseline characteristics. Contemporary Clinical Trials 2010;31(4):303-311. 26. Bleich HL, Slack WV. Reflections on electronic medical records: When doctors will use them and when they will not. International Journal of Medical Informatics 2010;79(1):1-4. 27. Hogan WR, Wagner MM. Accuracy of data in computer-based patient records. J Am Med Inform Assoc. 1997;4(5):342-55. 28. Quaas J, Lanigan M, Newman D, McOsker J, Babayev R, Mason C. Utility of the digital rectal examination in the evaluation of undifferentiated abdominal pain. The American Journal of Emergency Medicine 2009;27(9):1125-1129. 29. Lehmann CU, Kim GR, Johnson KB, Sable C, Reyna M, Holbrook PR. Telemedicine Applications in Pediatrics. Pediatric Informatics: Springer New York, 2009a:279-292. 30. Gackstetter GD, Hooper TI, Al Qahtani MS, Smith TC, Memish ZA, Schlangen KM, et al. Assessing the potential health impact of the 1991 Gulf War on Saudi Arabian National Guard Soldiers. International Journal of Epidemiology 2005;34(4):801-808. 31. Department of Health and Human ServicesHealth Records and Office of the Assistant Secretary for Planning and Evaluation, Task Force on Privacy, Social Needs and Personal Privacy. http://aspe.hhs.gov/pic/reports/ahrq/4441.pdf, accessed 26 January 2011. 1993. 32. US General Accounting Office, Report to the Chairman, Commiittee on Governmental Affairs, US Senate. http://archive.gao.gov/t2pbat8/143217.pdf. 1991. 33. Greenhalgh T, Wood GW, Bratan T, Stramer K, Hinder S. Patients' attitudes to the summary care record and HealthSpace: qualitative study. Bmj 2008;336(7656):1290-5. 34. Coiera E. Do we need a national electronic summary care record? Med J Aust 2011;194(2):90-2. 35. Morris LM, Cameron J, Brown C, Wyatt JC. Sharing summary care records. Results from Scottish emergency care summary. Bmj 2010;341(10):c4305. 215 36. Cresswell KS, Aziz. The NHS Care Record Service (NHS CRS): recommendations from the literature on successful implementation and adoption. Informatics in Primary Care 2009;17(3):153-160. 37. Wikipedia. Continuity of Care Record; http://en.wikipedia.org/wiki/Continuity_of_Care_Record, accessed 26 January 2011. 2011. 38. Lehmann CU, Kim GR, Johnson KB, Kim GR, Lehmann CU. Electronic Health Records and Interoperability for Pediatric Care. Pediatric Informatics: Springer New York, 2009b:257-264. 39. Gill J, Chearman T, Carey M, Nijjer S, Cross F. Presenting patient data in the electronic care record: the role of timelines. JRSM Short Reports 2010;1(4). 40. Scottish Government. Partnership for Care: Scotland's Health White Paper; http://www.scotland.gov.uk/Publications/2003/02/16476/18737. 2011. 41. Satoguina J, Walther B, Drakeley C, Nwakanma D, Oriero EC, Correa S, et al. Comparison of surveillance methods applied to a situation of low malaria prevalence at rural sites in The Gambia and Guinea Bissau. Malar J 2009;8(274):274. 42. Ross DA, Hinman AR, Saarlas KN, Lloyd-Puryear MA, Downs SJ. The NearTerm Future for Child Health Information Systems. Journal of Public Health Management and Practice 2004;10:S99-S104. 43. Höerbst A, Ammenwerth E. Quality and Certification of Electronic Health Records. ACI. 2010 May;1(2):149-164. Applied Clinical Informatics 2010;1(2):149-164. 44. Lee D, Lau F, Quan H. A method for encoding clinical datasets with SNOMED CT. BMC Medical Informatics and Decision Making 2010;10(1):53. 45. Halamka JD, Mandl KD, Tang PC. Early Experiences with Personal Health Records. Journal of the American Medical Informatics Association 2008;15(1):1-7. 46. Gregory KE, Radovinsky L. Research strategies that result in optimal data collection from the patient medical record. Applied Nursing Research 2010. 47. Greenhalgh T HS, Stramer K, Bratan T, Russell J. Adoption, non-adoption, and abandonment of a personal electronic health record: case study of HealthSpace. BMJ 2010b;341:c5814. 48. Pagliari C, Detmer D, Singleton P. Potential of electronic personal health records. BMJ 2007b;335(7615):330-333. 49. National Health and Hospitals Reform Commission, Australia. Person-controlled Electronic Health Records; http://www.health.gov.au/internet/nhhrc/publishing.nsf/Content/mediaRelease 300409, accessed 26 January 2011. 2009. 50. Rowlands S, Callen J, Westbrook J. What information do general practitioners need to care for patients with lung cancer? A survey of general practitioners perceptions. Health Information Management Journal 2010;39(1):8-16. 51. Cooper C, Steinbuch M, Stevenson R, Miday R, Watts N. The epidemiology of osteonecrosis: findings from the GPRD and THIN databases in the UK. Osteoporosis International 2009;21(4):569-577. 52. Austin T, Kalra D, Lea NC, Patterson DL, Ingram D. Analysis of Clinical Record Data for Anticoagulation Management within an EHR System. Open Med Inform J 2009;3:54-64. 53. Department of Health. Delivering 21st Century IT support for the NHS: National specification for Integrated Care Records Service - consultation draft; http://webarchive.nationalarchives.gov.uk/+/www.dh.gov.uk/en/Publicationsan dstatistics/Publications/PublicationsPolicyAndGuidance/DH_4108952. 2002. 54. Groen PJ, Goldstein D, Jaime N. Personal Health Record (PHR) Systems: An Evolving Challenge to EHR Systems. http://www.hoise.com/vmw/07/articles/vmw/LV-VM-08-07-26.html 2007. 216 55. National Committee on Vital and Health Statistics, U.S. Department of Health and Human Services. http://www.ncvhs.hhs.gov/0602nhiirpt.pdf; accessed 25 March 2011. 2006. 56. Ross SE, Lin CT. The effects of promoting patient access to medical records: a review. J Am Med Inform Assoc 2003;10(2):129-38. 57. Fisher B, Bhavnani V, Winfield M. How patients use access to their full health records: a qualitative study of patients in general practice. J R Soc Med 2009;102(12):539-44. 58. Canada Health Infoway. Blueprint -an interoperable EHR framework. 2006. 59. Department of Health. Liberating the NHS:An Information Revolution; executive summary http://www.dh.gov.uk/prod_consum_dh/groups/dh_digitalassets/@dh/@en/document s/digitalasset/dh_120601.pdf. 2010. 60. Institute of Medicine. To Err Is Human: Building a Safer Health System National Academy Press, Washington, DC 1999. 61. Robertson A, Cresswell K, Takian A, Petrakaki D, Crowe S, Cornford T, et al. Implementation and adoption of nationwide electronic health records in secondary care in England: qualitative analysis of interim results from a prospective national evaluation. BMJ 2010;341:c4564. 62. Pagliari C, Detmer D, Singleton P. Electronic Personal Health Records, Emergence and Implications for UK, http://www.nuffieldtrust.org.uk/ecomm/files/Elec%20Personal%20Records%2 0II.pdf. The Nuffield Trust 2007a. 63. Greenhalgh T SK, Bratan T, Byrne E, Russell J, Potts HW.Greenhalgh T, Stramer K, Bratan T, Byrne E, Russell J, Potts HW. Adoption and nonadoption of a shared electronic summary record in England: a mixed-method case study. BMJ 2010a;341. 64. Silvestre AL, Sue VM, Allen JY. If you build it, will they come? The Kaiser Permanente model of online health care. Health Aff (Millwood) 2009;28(2):334-44. 65. Chen C, Garrido T, Chock D, Okawa G, Liang L. The Kaiser Permanente Electronic Health Record: Transforming And Streamlining Modalities Of Care. Health Affairs 2009;28(2):323-333. 66. Morrison Z, Robertson A, Cresswell K, Crowe Sa, Sheikh A. Understanding contrasting approaches to nationwide implementations of electronic health record systems: England, the USA and Australia. Journal of Healthcare Engineering 2010. 67. House of Commons Health Committee, House of Commons London. The Electronic Patient Record Sixth Report of Session 2006-07 http://www.publications.parliament.uk/pa/cm200607/cmselect/cmhealth/422/4 22.pdf. The Stationery Office Limited 2006-07;1(HC 422-II). 68. Canada Health Infoway, Electronic Health Information and Communication Technology Projects, March 2010 Performance Evaluation. Final Report, http://www2.infowayinforoute.ca/Documents/Performance%20Evaluation%20Report_English.pdf, accessed 26 January 2011. 69. Dossier Médicale Personnel, France. http://www.ehealtheurope.net/Features/item.cfm?docId=194, accessed 26 January 2011. 70. Simpson CR, Sheikh A. Trends in the epidemiology of asthma in England: a national study of 333,294 patients. J R Soc Med 2010;103(3):98-106. 71. Gray L, Batty GD, Craig P, Stewart C, Whyte B, Finlayson A, et al. Cohort Profile: The Scottish Health Surveys Cohort: linkage of study participants to routinely collected records for mortality, hospital discharge, cancer and offspring birth 217 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. characteristics in three nationwide studies. International Journal of Epidemiology 2010;39(2):345-350. Ford I, Murray H, Packard CJ, Shepherd J, Macfarlane PW, Cobbe SM. LongTerm Follow-up of the West of Scotland Coronary Prevention Study. New England Journal of Medicine 2007;357(15):1477-1486. Gnani S, Majeed A. A user's guide to data collected in primary care in England; http://www1.ic.ac.uk/resources/579D8B09-C1C1-4026-A7BEC3E936EE9567/, accessed 26 January 2011. Eastern Region Public Health Observatory, Cambridge 2005. Centres for Disease Control and Prevention, Atlanta, USA. Assessing the potential of national strategies for electronic health records for population health monitoring and research; http://www.cdc.gov/nchs/data/series/sr_02/sr02_143.pdf; accessed 26 January 2011. 2006. Healthcare Information and Management Systems Society. HIMSS Electronic Health Record, Definitional Model, Version 1.0; http://www.providersedge.com/ehdocs/ehr_articles/HIMSS_EMR_Definition_ Model_v1-0.pdf; accessed 9th March 2011. 2003. Evans RS, Pestotnik SL, Classen DC, Clemmer TP, Weaver LK, Orme JF, et al. A Computer-Assisted Management Program for Antibiotics and Other Antiinfective Agents. New England Journal of Medicine 1998;338(4):232-238. Jones DL, Kroenke K, Landry FJ, Tomich DJ, Ferrel RJ. Cost savings using a stepped-care prescribing protocol for nonsteroidal anti-inflammatory drugs. JAMA 1996;275(12):926-30. AHRQ. Costs and Benefits of Health Information Technology, Evidence Report/Technology Assessment Number 132. 2006;6(E006). DeLone WH, McLean ER. Information Systems Success: The Quest for the Dependent Variable. INFORMATION SYSTEMS RESEARCH 1992;3(1):6095. Majeed A. Sources, uses, strengths and limitations of data collected in primary care in England. Health Stat Q 2004;21:5-14. Herrett E, Thomas SL, Schoonen WM, Smeeth L, Hall AJ. Validation and validity of diagnoses in the General Practice Research Database: a systematic review. Br J Clin Pharmacol 2010;69(1):4-14. Jamal A, McKenzie K, Clark M. The impact of health information technology on the quality of medical and health care: a systematic review. Him J 2009;38(3):26-37. Chan KS, Fowles JB, Weiner JP. Review: Electronic Health Records and the Reliability and Validity of Quality Measures: A Review of the Literature. Medical Care Research and Review 2010;67(5):503-527. Hoerbst A, Ammenwerth E. Electronic health records. A systematic review on quality requirements. Methods Inf Med 2010;49(4):320-36. Uslu AM, Stausberg J. Value of the electronic patient record: An analysis of the literature. Journal of Biomedical Informatics 2008;41(4):675-682. Irani JS, Middleton JL, Marfatia R, Omana ET, D'Amico F. The Use of Electronic Health Records in the Exam Room and Patient Satisfaction: A Systematic Review. J Am Board Fam Med 2009;22(5):553-562. Clamp S, Keen J. The Value of Electronic Health Records: A Literature Review. 2005. Watson DJ, Rhodes T, Guess HA. All-cause mortality and vascular events among patients with rheumatoid arthritis, osteoarthritis, or no arthritis in the UK General Practice Research Database. J Rheumatol 2003;30(6):1196-202. Jordan K, Clarke AM, Symmons DP, Fleming D, Porcheret M, Kadam UT, et al. Measuring disease prevalence: a comparison of musculoskeletal disease 218 using four general practice consultation databases. Br J Gen Pract 2007;57(534):7-14. 90. Goldzweig CL, Towfigh A, Maglione M, Shekelle PG. Costs And Benefits Of Health Information Technology: New Trends From The Literature. Health Affairs 2009;28(2):w282-w293. 91. Wang SJ, Middleton B, Prosser LA, Bardon CG, Spurr CD, Carchidi PJ, et al. A cost-benefit analysis of electronic medical records in primary care. Am J Med 2003;114(5):397-403. 92. McDonald R, Cheraghi-Sohi S, Tickle M, Roland M, Doran T. The impact of incentives on the behaviour and performance of primary health care professionals. NHS Service Development and Organisation Programme. http://www.sdo.nihr.ac.uk/files/project/158-final-report.pdf; accessed 25 March 2011. 2010( ). 93. Centers for Medicare & Medicaid Services. CMS finalizes definition of meaningful use of certified electronic health records (EHR) technology; http://www.cms.gov/apps/media/press/factsheet.asp?Counter=3794&intNum PerPage=10&checkDate=&checkKey=&srchType=1&numDays=3500&srchO pt=0&srchData=&keywordType=All&chkNewsType=6&intPage=&showAll=&p Year=&year=&desc=&cboOrder=date; accessed 28th February 2011. 2010. 94. HITECH Law Blog. HHS Releases Proposed EHR “Meaningful Use” Standards http://healthitlawblog.wordpress.com/2010/01/07/hhs-releases-proposed-ehrmeaningful-use-standards/; accessed 26 March 2011. 2010. 95. Department of Health and Human Services. Anticipating Unintended Consequences of Health Information Technology and Health Information Exchange. Presolicitation notice; https://www.fbo.gov/index?s=opportunity&mode=form&id=22633354ce0c9b300b832f f16c2658a7&tab=core&_cview=0; Accessed 25 March 2011. 2010. 96. Coiera E. Building a National Health IT System from the Middle Out. Journal of the American Medical Informatics Association 2009;16(3):271-273. 97. Fontaine P, Ross SE, Zink T, LM. S. Systematic review of health information exchange in primary care practices. The Journal of the American Board of Family Medicine 2010;23(5):655-670. 98. Abdolrasulnia M, Menachemi N, Shewchuk RM, Ginter PM, Duncan WJ, Brooks RG. Market effects on electronic health record adoption by physicians. Health Care Manage Rev 2008;33(3):243-252. 99. NHS Connecting for Health. GP Systems of Choice (GPSoC) http://www.connectingforhealth.nhs.uk/systemsandservices/gpsupport/gpsoc; accessed 26 March 2011. 100. Buckley BS, Simpson CR, McLernon DJ, Murphy AW, Hannaford PC. Five year prognosis in patients with angina identified in primary care: incident cohort study. BMJ 2009;339. 101. Sanderson H, Adams T, Budden M, Hoare C. Lessons from the central Hampshire electronic health record pilot project:evaluation of the electronic health record for supporting patient care and secondary analysis. BMJ 2004;328(7444):875-878. 102. Black C, Kaye JA, Jick H. Relation of childhood gastrointestinal disorders to autism: nested case-control study using data from the UK General Practice Research Database. BMJ 2002;325(7361):419-421. 103. Smith AG, Fear NT, Law GR, Roman E. Representativeness of samples from general practice lists in epidemiological studies: case-control study. BMJ 2004;328(7445):932. 104. Mackay D, Haw S, Ayres JG, Fischbacher C, Pell JP. Smoke-free Legislation and Hospitalizations for Childhood Asthma. New England Journal of Medicine 2010;363(12):1139-1145. 219 105. Hippisley-Cox J, Coupland C. Unintended effects of statins in men and women in England and Wales: population based cohort study using the QResearch database. BMJ 2010;340:2197. 106. Helms PJ, Ekins Daukes S, Taylor MW, Simpson CR, McLay JS. Utility of routinely acquired primary care data for paediatric disease epidemiology and pharmacoepidemiology. Br J Clin Pharmacol 2005;59(6):684-90. 107. Gray J, Orr D, Majeed A. Use of Read codes in diabetes management in a south London primary care group: implications for establishing disease registers. BMJ 2003;326(7399):1130. 108. Win KT. A review of security of electronic health records. Him J 2005;34(1):138. 109. Robert G, Greenhalgh T, MacFarlane F, Peacock R. Adopting and assimilating new non-pharmaceutical technologies into health care: a systematic review. J Health Serv Res Policy 2010;15(4):243-50. 220 Chapter 7 Computer-assisted history taking systems Summary • Most computer-assisted history taking systems are designed for use by professionals, often as templates incorporated into the electronic health record; these are however increasingly becoming available for completion by patients and carers, either on-site in healthcare settings or remotely before or after the clinical assessment. • Computer-assisted history taking systems can be used in a range of contexts, these include, primary prevention (e.g. population screening programmes), to support the extraction of potentially sensitive information (e.g. a detailed sexual or alcohol intake history) and to support disease management of those living with long-term conditions (e.g. collection of dietary intake information for diabetes care). • Most of the applications are at present supported only by face validity and modest or weak empirical evidence. The clearest evidence of benefit at present relates to the elicitation of sensitive information. • There is however inconclusive evidence that computer-assisted history taking systems translate into time savings for professionals or improved important patient outcomes. • Key areas in which to consider the wider use of computer-assisted history taking systems include situations in which patients and/or carers might be apprehensive about disclosing potentially embarrassing information and situations in which a fixed repertoire of questions need to be asked of large numbers of patients such as in the context of screening programmes. 221 7.1 Introduction Medical history taking lies at the centre of clinical diagnosis and decisionmaking. As described by Pringle over a decade ago, computer-assisted history taking systems (CAHTS) are tools that aim to aid clinicians in gathering data from patients to inform a diagnosis and/or formulating a treatment plan.1 Of relevance is the increasing potential for patients to complete aspects of their history, prior to or post the consultation, whilst onsite in surgeries and clinics, or remotely. Online systems can also help with data collection for screening programmes, such as England’s national vascular risk assessment and management strategy.2 Despite the many possible applications and even though CAHTS have been available for nearly three decades,3 these are however not very commonly used in routine clinical practice. 7.2 Definition, description and scope 7.2.1 Definition The medical history refers to the information obtained by a clinician or other healthcare professional by listening to the patient/carer’s history and then asking questions which can provide further insight and clarification about the condition. A history taking system is a tool that aids clinicians in gathering such information from patients to help formulate the diagnosis or inform a treatment plan. If computerised, these tools are referred to as CAHTS. 7.2.2 Description Computer-assisted history taking systems can be used by healthcare professionals, or directly by patients, as in the case of, for example, preconsultation interviews.4-6 They can be used remotely, for example via the Internet, telephone or on-site. They draw on a range of technologies such as personal computers (PCs), personal digital assistants (PDAs), mobile phones and electronic kiosks; data input can be mediated via, amongst others, keyboards, touch screens and voice-recognition software (see Table 7.1). 222 Data Collection Modes On-site Telephone Administration The patient Modes present on-site with Internet is a health practitioner The patient responds who facilitates the to process completed technological with a practitioner who N/A records the (audio responses system (audio oral, presentation; presentation; patient is The patient responds The patient is given present on-site and to an automated access to an online conducts the history telephone by oral, keyed input) keyed input) The telephone interview by a health history-taking Professional- a himself/herself that through system survey to complete, records the which can then be a responses for access accessed Patient- technological completed system, by such laptop, a as practitioner desktop presentation; health health a practitioner (audio and linked to other keyed similar patient computer, or PDA input) records (audio audio & by visual presentation; keyed (visual & presentation; keyed input) input) Table 7.1 Modes of data collection for computer-assisted history taking systems 7.2.3 Scope As noted above, there are two main approaches to using these tools to facilitate history taking, these being either for use by healthcare professionals or where these involve obtaining data directly from patients. 223 Health professional completed This is where health professionals input patient details into computers, typically via templates; this can include remote telephone interview and onsite interview. Examples include: • Emergency services: Where mobile systems are used in ambulatory services and/or in accident and emergency departments. Sometimes computer-assisted history taking and computer decision support systems are used in combination. For example, nurses in the out-ofhours, emergency service type in patient responses to questions generated by the computer programme and then direct patients to answer further relevant questions leading ultimately to a diagnosis and/or a management plan.7;8 • General practice systems: Where CAHTS record aspects of the patient’s history into a computerised template, which informs part of the patient record.9 Patient/carer completed (self-administered) This refers to instances in which the computer software prompts or enables the patient to input relevant information themselves; this can include on-site self-interview, remote, automated telephone self-interview and remote, online self-interview. Examples of these include: • Pre-consultation questionnaire: This enables patients to complete the history either on a computer in the waiting room or at home.4-6;10;11 This has, for example, been shown to help professionals rapidly to appraise psychiatric referrals and prioritise patients.12;13 • Electronic and on-line health records: Where the patient types their own history into sections of their electronic health record (EHR), which can then link to other similar patient-held records. With the addition of diagnostic and reminder functionalities, CAHTS may influence all stages of the patient care pathway before, during and after the consultation. For example, with the addition of a diagnostic platform such as 224 probabilistic advice and question-prompting, CAHTS may become instrumental in the decision-making process. 7.3Theoretical benefits and risks 7.3.1 Potential benefits Saving professionals’ time on documentation Professionals have limited time in consultations and, in a traditional face-toface clinical encounter, it is not always possible to obtain a complete or, in some cases, even relevant medical history.14 In an evaluation of ambulatory practices, clinicians were, for example, found to spend 20% of their day writing.15;16 In an Ohio family practice, dictation and charting outside of examination rooms occupied 56 minutes of an eight-hour working day.17 In an antenatal clinic, it was estimated that two-thirds of the working day was spent recording information.18 These studies demonstrate the potential importance of CAHTS as a time-saving device for clinicians, particularly if these devices can also sift and present relevant data in an easily accessible format. Collection of more comprehensive and valid information Clinicians need to remember many questions relating to the management of each condition. Omitting an important question can have considerable implications for diagnosis and treatment. For example, studies show that 50% of psychosocial and psychiatric problems are missed 19 and that 54% of patient problems and 45% of patient concerns are neither elicited by the clinician nor disclosed by the patient.20 Furthermore, studies have demonstrated fewer errors in PDA data records than in paper diaries and that PDA datasets were correctly completed in 93% of cases21 while one study reported patient compliance of 100%.22 Patient level impacts Computer-assisted history taking systems can be used in several clinical settings and are particularly useful in eliciting potentially sensitive information, for example, on alcohol consumption,23; 24psychiatric care,25-28 sexual health29 and gynaecological health.30 Using CAHTS before the consultation would 225 also, in principle, allow more time for the patient to discuss their actual health problem rather than routine aspects of medical history with their physician. There are important potential benefits of CAHTS in the care of people with special needs. For example, computers can allow questions to be asked in a number of different languages. They can also provide multi-media forms for patients who cannot read and write through making computers voice questions and digitally record spoken responses.4; 31 7.3.2 Potential risks A central limitation of CAHTS may be that not every question can be meaningfully answered in a questionnaire; such issues are likely to be particularly relevant when using general questionnaires which may include questions irrelevant to individual patient concerns and needs.32 Computerassisted history taking systems are inherently limited with respect to nonverbal communication. Computers are currently unable to detect non-verbal behaviour since a computer cannot, for example, sense a patient’s mood that might easily be picked up in a consultation.4 Another related concern is that computers may depersonalise the doctor-patient relationship, although this has not actually been demonstrated in the clinical setting.6 7.4 Empirically demonstrated benefits and risks 7.4.1 Benefits We found five completed published systematic reviews (SRs) on this subject. 32; 33; 34; 35; 36 We have furthermore initiated three additional related Cochrane reviews in key areas.37-39 We synthesise this body of evidence together with additional selected evidence to illustrate key findings in relation to CAHTSs. Saving professionals’ time on documentation The length of time associated with documenting information was noted almost four decades ago and was one of the reasons that Mayne et al.40 initiated their research into the scope for patient interviews being aided by computers. Our review of the literature found that evidence on the effectiveness of CAHTS on saving professionals’ time on documentation is non-conclusive. 226 A study comparing PDA use by patients for documenting acute pain assessment time and completeness and comparing this to pen-and-paper showed that PDA usage decreased the overall consultation time (such as chart review, assessment and documentation) by 22 per cent.35 However, as Poissant et al. point out, although over 90 patient encounters were considered in this study, it was a single clinician study so the results may not be generalisable to other settings.35 Further, Poissant et al. demonstrated that the use of central station desktops for CAHTS is slightly less time-consuming (with a weighted average reduction of 8 per cent) for clinicians, when compared to bedside or point-of-care computer systems.35 Dale and Hagen point out that PDA use for CAHTS benefit healthcare professionals as there is a ‘reduction in the data management and processing time’.20-22; 41; 42 However, barriers still exist as all five (out of a total of nine) of the studies in this SR that reported data concerning feasibility of collecting data using PDAs identified persistent ‘technical difficulties with the PDA technology’.20; 21 ‘Power problems’ and ‘PDA malfunction’ were the most common problems. Collection of more comprehensive and valid information Kerkenbush et al. found increased compliance with data entry45 and noted that Invivodata™ (a company specialising in electronic diary technology) showed that patients responded timely to 93% of all the electronic data gathering prompts.46 Also, Kamarack et al. found 99% compliance with assessments that needed to be completed every 45 minutes during waking hours over a six-day period.45 Studies have demonstrated fewer errors in PDA data records than in paper diaries and that PDA datasets were correctly completed in 93% of cases22 while one study reported patient compliance of 100%.21 A recent randomsied controlled trial (RCT) by Richens et al46 has found that the odds of disclosure for seven key sensitive behaviours were found to be 40% greater when computer-assisted interviewing was compared with penand-paper interviewing. De Jarlais47 reported that CAHTS enable substantially more complete reporting of HIV risk behaviour than face-to-face interviewing. Furthermore, if patients use a PDA for a CAHTS, they are less likely to falsify 227 data when compared with those generated by pen-and-paper, as demonstrated, by four RCTs.21; 41; 42; 48 Similarly, Tiderman et al,49 found that women undertaking the CAHTS interviewing reported a significantly higher median number of male partners for the preceding 12 months than those undertaking face-to-face interviewing. However the overall responses to sexual history related questions using CAHTS and face-to-face interviewing were similar. Patient level impacts There is good evidence, that patients prefer to record aspects of their history using PDA technology when compared with recording using pen-and-paper.21; 41; 42 Furthermore, if patients use a PDA for CAHTS, they are less likely to falsify data when compared with those generated by pen-and-paper. 50 One RCT for example, reported that ‘young people find computerised questionnaires equally or more acceptable than the usual clinical interview or a written questionnaire’.23 In another study, parents originally assumed that an interview using a computer was not as ‘friendly and personal’, but became more optimistic after the interview was completed.25 7.4.2 Risks We should not expect CAHTS to pose risks to patients’ wellbeing. However, technical problems, lack of training and poorly integrated systems can result in frustration for patients and professionals. A SR assessed the use of PDAbased templates for CAHTS by medical trainees.34 The review highlighted that while most medical trainees who use handheld computers for patient history taking appear comfortable and are generally satisfied with them, barriers still exist, these including: 1) lack of technical experience; 2) a preference for pen-and-paper; 3) difficulty handling the small device; and 4) concerns about data loss and security.34 Also, what may work in some clinical settings (such as an infertility, antenatal or allergy clinic) may not work as well in other areas where patients more typically present with undifferentiated problems (such as general practice or a general medical or gynaecology clinic, for example).33; 35 228 7.5 Implications for policy, practice and research 7.5.1 Implications for policy The progressive change in focus from hospital care to community-based care means that staff will become more mobile and therefore need to access and input data at the point-of-care. There is also a drive within health systems globally to focus on health promotion, disease prevention, and early detection of disease. For example, in England, a vascular risk assessment programme for people aged 40-74 is now starting; CAHTS can be helpful in collecting risk factor and behavioural data for this.51; 52 Gathered histories can be stored locally or in a user-friendly online space that can be accessed by patients supporting tailored advice and self-management.6 7.5.2 Implications for practice Several studies have found that self-administered computer-assisted interviewing is perceived favourably by patients because computer systems cannot be judgemental towards sensitive behavioural data such as sexual history taking and domestic violence.53 Clinician and patient-operated CAHTS data are potentially important additions to the EHR as they can help to improve data quality through: data entry forms with data validation checks; encoding of data; legibility; easier access to past records; attribution of entries; easier availability; and facilitating patient checks of their own data. Consideration also needs to be given to incorporating patient completed diaries online, thereby allowing information on key complaints and selfgenerated data (for example, blood pressure or peak expiratory flow) to be made available to clinicians before the actual consultation. This may result to more complete history taking and more time available to spend on the actual consultation. However Greenhalgh et al.,54 found that patients perceived HealthSpace (a personal electronic health record accessed via the Internet) as neither useful nor easy to use and its functionality aligned poorly with their expectations and self-management practices. The same authors conclude that ‘unless personal EHRs align closely with people's attitudes, self management practices, 229 identified information needs, and the wider care package (including organisational routines and incentive structures for clinicians), the risk that they will be abandoned or not adopted at all is substantial. Conceptualising such records dynamically (as components of a socio-technical network) rather than statically (as containers for data) and employing user centred design techniques might improve their chances of adoption and use’. In general, the adoption of healthcare informatics is complex, multidimensional, and influenced by a variety of factors that relate to individuals and organisations. A recent systematic review of mixed methods studies55 found that successful implementation strategies should include: ‘a) involving users at different development and implementation phases, b) using project champions or other key staff, c) providing adequate training and support, and d) monitoring system use in the early stages of implementation’.55 7.5.3 Implications for research Seminal reports on quality and safety of healthcare56 invariably recognise information technology as one of the main vehicles for making radical improvements in delivery of healthcare. Although CAHTS have been available for around 30 years, successful use in routine healthcare is still variable. A recent SR36 found that results of CAHTS effectiveness depend on the sensitivity of the question, the tool used as well as the population assessed. Certainly, any interpretation of results is based on two sets of assumptions. Firstly, no self-report tool can be proven to be more accurate than another, given that no gold standard exists with which to compare alternative methods. Secondly, many assumptions are made about people’s social desires when answering questions that may not always go in the direction expected. In further developing methods for collecting patient histories multiple methods assessing self-reported behavioural risks should be complemented with appropriate biological markers as appropriate, depending on the type of study.36 230 Important changes should take place in how we conceive, plan and conduct primary and secondary research on CAHTS so that we provide the framework for a comprehensive evaluation that will lead to an evidence base to inform policy and practice. Findings of such comprehensive evaluation should also inform a sustainability model that views CAHTS as an integral component and working platform for EHRs (see Chapter 6). Relatively few studies reported on safety outcomes57 when evaluating CAHTS, whilst others sometimes failed to assess the most salient dimensions of quality such as access, accessibility and equity.58 Although CAHTS are frequently promoted as being “cost-saving”, 59-62 this attribute was rarely evaluated rigorously.35 Most of the technologies are at present supported only by face validity and modest or weak empirical evidence. Unless these systems are adequately studied, they may not “mature” to the extent that is needed to realise their full potential when deployed in everyday clinical settings.58 Additional research on the reliability of different methods of data collection would also be useful in assessing the value of collecting data using CAHTS in healthcare settings.56 Moreover, even after successful interventions, accuracy may not be maintained over time. Medical processes are complex and changing, and data error, as well as procedural changes may occur due to the high turnover of personnel. Hence, there is a pressing need for regular evaluations of CAHTS, analogous to techniques used in continuous quality improvement.35; 51; 63 References 1. Pringle M. Using computers to take patient histories. BMJ 1998; 297: 698-698. 2. Department of Health: Putting prevention first. Vascular checks: risk assessment and management. http://www.dh.gov.uk/prod_consum_dh/groups/dh_digitalassets/@dh/ @en/ documents/digitalasset/dh_083823.pdf 3. Slack WV, Hicks GP, Reed CE, Van Cura LJ. A computer-based medical-history system. New England Journal of Medicine 1966; 274: 194-198. 4. Healthspace. https://www.healthspace.nhs.uk/ (last accessed 20/09/2009). 2009. 231 5. RelayHealth. https://www.relayhealth.com/rh/general/news/newsArchive/news54.aspx (last accessed 20/09/2009). 2009. 6. Bachman JW. The patient-computer interview: a neglected tool that can aid the clinician. Mayo Clin Proc 2003; 78: 67-78. 7. Parkin A. Computers in clinical practice: applying experience from child psychiatry. BMJ 2000; 321: 615-618. 8. Lilford RJ, Guthrie K, Kelly M. History-taking by computer. Baillieres Clin Obstet Gynaecol 1990; 4: 723-742. 9. NHS Direct in England. Report by the controller and auditor general. HC 505 Session 2001-2002: 25 January 2002. 25-1-2002. 10. http://www.nhsdirect.nhs.uk/ (last accessed 20/09/2009). 2007. 11. Millman A, Lee N, Brooke A. ABC of medical computing. Computers in general practice--I. BMJ 1995; 311: 800-802. 12. Jones RB, Atkinson JM, Coia DA, Paterson L, Morton AR, McKenna K et al. Randomised trial of personalised computer based information for patients with schizophrenia. BMJ 2001; 322: 835-840. 13. Morrison-Beedy D, Carey MP, Tu X. Accuracy of audio computer-assisted selfinterviewing (ACASI) and self-administered questionnaires for the assessment of sexual behavior. AIDS Behav 2006; 10: 541-552. 14. Taylor J. http://www.e-health-insider.com/news/item.cfm?ID=1162. Last accessed 28/05/07 2007. 15. Tang PC, Jaworski MA, Fellencer CA, LaRosa MP, Lassa JM, Lipsey P et al. Methods for assessing information needs of clinicians in ambulatory care. Proc Annu Symp Comput Appl Med Care 1995; 630-634. 16. Tang PC, Jaworski MA, Fellencer CA, Kreider N, LaRosa MP, Marquardt WC. Clinician information activities in diverse ambulatory care practices. Proc AMIA Annu Fall Symp 1996; 12-16. 17. Jancin B. Two hours a day of unreimbursed time. Fam Pract News . 2001. 18. Wong WS, Lee KH, Chang MZ. A microcomputer based interview system for antenatal clinic. Comput Biol Med 1986; 16: 453-463. 19. Davenport S, Goldberg D, Millar T. How psychiatric disorders are missed during medical consultations. Lancet 1987; 2: 439-441. 20. Palermo TM, Valenzuela D, Stork PP. A randomized trial of electronic versus paper pain diaries in children: impact on compliance, accuracy, and acceptability. Pain 2004; 107: 213-219. 21. Tiplady B, Crompton GK, Dewar M.H, Bollert FGE, Matusiewicz SP, Campbell LM. The use of electronic diaries in respiratory studies. Drug Inf J 1997; 31: 759764. 22. Dale O, Hagen KB. Despite technical problems personal digital assistants outperform pen and paper when collecting patient diary data. J Clin Epidemiol 2007; 60: 8-17. 23. Supple AJ, Aquilino WS, Wright DL. Collecting sensitive self-report data with laptop computers: Impact on the response tendencies of adolescents in a home interview. Journal of Research on Adolescence 1999; 9: 467-488. 24. Skinner HA, Allen BA. Does the computer make a difference? Computerized versus face-to-face versus self-report assessment of alcohol, drug, and tobacco use. J Consult Clin Psychol 1983; 51: 267-275. 25. Carr AC, Ghosh A, Ancill RJ. Can a computer take a psychiatric history? Psychol Med 1983; 13: 151-158. 26. Sawyer MG, Sarris A, Baghurst P. The use of a computer-assisted interview to administer the Child Behavior Checklist in a child psychiatry service. J Am Acad Child Adolesc Psychiatry 1991; 30: 674-681. 27. Sawyer MG, Sarris A, Baghurst P. The effect of computer-assisted interviewing on the clinical assessment of children. Aust N Z J Psychiatry 1992; 26: 223-231. 232 28. Erdman HP, Klein MH, Greist JH. Direct patient computer interviewing. J Consult Clin Psychol 1985; 53: 760-773. 29. Jaya, Hindin, M. J., & Ahmed, S. 2008, "Differences in young people's reports of sexual behaviors according to interview methodology: a randomized trial in India", Am.J Public Health, vol. 98, no. 1, pp. 169-174. 30. Brownbridge G, Lilford RJ, Tindale-Biscoe S. Use of a computer to take booking histories in a hospital antenatal clinic. Acceptability to midwives and patients and effects on the midwife-patient interaction. Med Care 1988; 26: 474-487. 31. Slack WV, Slack CW. Patient-computer dialogue. N Engl J Med 1972; 286: 13041309. 32. Wu RC, Straus SE. Evidence for handheld electronic medical records in improving care: a systematic review. BMC Med Inform Decis Mak 2006; 6: 26. 33. Garritty C, El EK. Who’s using PDAs? Estimates of PDA use by health care providers: a systematic review of surveys. J Med Internet Res 2006;8:e7. 34. Kho A, Henderson LE, Dressler DD, Kripalani S. Use of handheld computers in medical education. A systematic review. J Gen Intern Med 2006; 21: 531-537. 35. Poissant L, Pereira J, Tamblyn R, Kawasumi Y. The impact of electronic health records on time efficiency of physicians and nurses: a systematic review. J Am Med Inform Assoc 2005; 12: 505-516. 36. Phillips AE, Gomez GB, Boily MC, Garnett GP.A systematic review and metaanalysis of quantitative interviewing tools to investigate self-reported HIV and STI associated behaviours in low- and middle-income countries. Int J Epidemiol. 2010 Dec;39(6):1541-55. 37. Whiteford A, Allen A, Pappas Y, Car J, Minozzi S, Pagliari C, Sheikh A, Majeed A (Protocol). Computer-assisted versus oral-and-written alcohol intake history taking for diagnosing alcohol misuse or abuse in screened participants. Cochrane Database of Systematic Reviews. 38. Pappas Y, Wei I, Car J, Majeed A, Sheikh A. Computer-assisted versus oral-andwritten family history taking for identifying people with elevated risk of type 2 diabetes mellitus. Cochrane Database of Systematic Reviews 2010, 4. Art. No.: CD008489. DOI: 10.1002/14651858.CD008489. 39. Wei I, Pappas Y, Car J, Majeed A, Sheikh A. Computer-assisted versus oral-andwritten dietary history taking for diabetes mellitus. Cochrane Database of Systematic Reviews 2010, 4. Art. No.: CD008488. DOI: 10.1002/14651858.CD008488 40. Mayne JG, Martin MJ, Morrow GW Jr, Turner RM, Hisey BL. A health questionnaire based on paper-and-pencil medium individualized and produced by computer, I: technique. JAMA. 1969;208: 2060-2063. 41. Gaertner J, Elsner F, Pollmann-Dahmen K, Radbruch L, Sabatowski R. Electronic pain diary: a randomized crossover study. J Pain Symptom Manage 2004; 28: 259-267. 42. Lauritsen K, Degl' IA, Hendel L, Praest J, Lytje MF, Clemmensen-Rotne K et al. Symptom recording in a randomised clinical trial: paper diaries vs. electronic or telephone data capture. Control Clin Trials 2004; 25: 585-597. 43. Kerkenbush NL, Lasome CE. The emerging role of electronic diaries in the management of diabetes mellitus. AACN Clin Issues 2003; 14: 371-378. 44. Shiffman S, Hufford M. Methods of measuring patient experience: paper versus electronic patient diaries. White Paper. Scotts Valley, Calif: invivodata, Inc; 2001. 2001. 45. Kamarack TW, Shiffman S, Smithline L. Effects of task strain, social conflict, and emotional activation on ambulatory cardiovascular activity: daily life consequences of recurring stress in a multiethnic adult sample. Health Psychology 17, 17-29. 1998. 46. Richens J, Copas A, Sadiq ST, Kingori P, McCarthy O, Jones V, Hay P, Miles K, Gilson R, Imrie J, Pakianathan M. A randomised controlled trial of computer- 233 assisted interviewing in sexual health clinics. Sex Transm Infect. 2010 Aug;86(4):310-4. 47. Des Jarlais DC, Paone D, Milliken J, Turner CF, Miller H, Gribble J, Shi Q, Hagan H, Friedman SR. Audio-computer interviewing to measure risk behaviour for HIV among injecting drug users: a quasi-randomised trial. Lancet. 1999 May 15;353(9165):1657-61. 48. Bulpitt CJ, Beilin LJ, Coles EC. Randomised controlled trial of computer-held medical records in hypertensive patients. BMJ 1976; 677-679. 49. Tideman RL, Chen MY, Pitts MK, Ginige S, Slaney M, Fairley CK. A randomised controlled trial comparing computer-assisted with face-to-face sexual history taking in a clinical setting. Sex Transm Infect. 2007 Feb;83(1):52-6 50. Hogan WR, Wagner MM. Accuracy of data in computer-based patient records. J Am Med Inform Assoc 1997;4:342–355. 51. Friedman RH. Automated telephone conversations to assess health behavior and deliver behavioral interventions. J Med Syst 1998; 22(2): 95-102. 52. Revere D, Dunbar PJ. Review of Computer-generated Outpatient Health Behavior Interventions: Clinical Encounters “in Absentia””. J Am Med Inform Assoc 2001; 8(1): 62-79. 53. Farzanfar R. When computers should remain computers: a qualitative look at the humanization of health care technology. Health Informatics J 2006; 12(3): 239254. 54. Greenhalgh T, Hinder S, Stramer K, Bratan T, Russell J. Adoption, non-adoption, and abandonment of a personal electronic health record: case study of HealthSpace. BMJ. 2010 Nov 16;341:c5814. doi: 10.1136/bmj.c5814 55. Gagnon MP, Desmartis M, Labrecque M, Car J, Pagliari C, Pluye P, Frémont P, Gagnon J, Tremblay N & Légaré F. Systematic Review of Factors Influencing the Adoption of Information and Communication Technologies by Healthcare Professionals. J Med Syst doi 10.1007/s10916-010-9473-4 56. Auerbach AD, Landefeld CS, Shojania KG. The tension between needing to improve care and knowing how to do it. N Engl J Med 2007; 357: 608–613. 57. Rigby M. Essential prerequisites to the safe and effective widespread roll-out of e-working in healthcare. Int J Med Inform 2006; 75: 138–147. 58. Stevens A, Milne R, Lilford R, Gabbay J. Keeping pace with new technologies: systems needed to identify and evaluate them. BMJ 1999; 319: 1291. 59. Stroetmann K, Jones T, Dobrev A, Stroetmann V. eHealth is worth it: The economic benefits of implemented eHealth solutions at ten European sites. 2006. Luxembourg, Office for Official Publications of the European Communities. 60. Corley ST. Electronic prescribing: a review of costs and benefits. Top Health Inf Manage 2003; 24: 29–38. 61. Eisenstein EL, Ortiz M, Anstrom KJ, Crosslin DR, Lobach DF. Economic evaluation in medical information technology: why the numbers don’t add up. AMIA Annu Symp Proc 2006; 914. 62. First Consulting Group. Costs, benefits and challenges of CPOE. 2003. 63. Sidorov J. It ain’t necessarily so: the electronic health record and the unlikely prospect of reducing health care costs. Health Aff (Millwood) 2006; 25: 1079– 1085. 234 Chapter 8 Computerised decision support systems Summary • Improved access to relevant clinical information for healthcare professionals can translate into improvements in practitioner performance, and possibly also the quality, safety and efficiency of care. • Computerised decision support systems are software applications that use patient data, a repository of clinical information (the knowledge-base) and an inference mechanism to generate patient specific recommendations for care. These applications are highly variable in sophistication and in the extent to which they integrate with other clinical information systems. • There is particular interest in using computerised decision support systems in the context of delivering preventive care, ordering and interpreting investigations, diagnostics, monitoring of patients, and guideline-based and prescribing-related decisions on care provision. • Although numerous evaluations of these applications have taken place, very little consistent and generalisable evidence exists on their ability to improve patient outcomes; there is however emerging evidence that they can, particularly if they involve provision of point-of-care recommendations that integrate well with clinician workflows, result in modest improvements in practitioner performance. • The use of computerised reminders for preventive care has been empirically demonstrated to be of the most benefit in impacting on relevant proxy measures. However, there are limited data on patient outcomes; this to a large extent reflects the prohibitive size and/or duration of study needed to demonstrate an effect on clinically relevant outcome measures. • These applications are largely unregulated in the US and UK, as they fall outside the remit of the Federal Drug Administration and Medicine and Healthcare Products Regulatory Agency, respectively. 235 Summary continued… • Without formal quality and safety assurances for these applications, the potential risks to patient safety need to be considered as they may in certain situations inadvertently introduce new risks and errors. • As evidence of benefit is clearest and risk of harm is lowest in relation to support for preventive healthcare, the NHS should consider introducing a range of computerised health promotion and prevention-related tools into primary care and, in the context of the ongoing roll-out of the National Health Service Care Records Service, also into secondary care. • Future research should focus on understanding the contexts in which these applications are most likely to prove effective and cost-effective in improving patient outcomes and this should be a priority consideration for NHS Connecting for Health as it introduces new eHealth applications with built-in decision support functionality. 8.1 Introduction Computerised decision support systems (CDSSs) are software applications that integrate patient data with a knowledge-base and an inference mechanism to produce patient specific output in the form of care recommendations, assessments, alerts and reminders to actively support practitioners in clinical decision-making.1 These systems can take a number of forms; this depends on the particular task that they seek to support clinicians with, the approach to drawing on patient data (which may have been input by professionals or which may involve automated interrogation of existing patient records), the type of knowledge-base that is drawn upon and the inference mechanism that is employed, the types of outputs that are generated and the ways in which these are communicated to healthcare providers. Finally, whilst patient or consumer orientated CDSSs now exist (and may well have a lot of potential for health gains) these will not be considered in this chapter as these fall outside the scope of this report. 236 This chapter aims to provide an overview of CDSSs aiming to support healthcare providers with delivery of care. It should be read in conjunction with Chapter 9, which focuses in detail on the specific case of CDSS support for image interpretation in the context of mammography screening, and Chapters 10, 11 and 12, which are concerned with CDSS for prescribing decisions and medicines management considerations. 8.2 Definition, description and scope for use 8.2.1 Definition There is no universally agreed definition of CDSS, this in part at least reflecting the continuing evolution of the scope, reasoning approach and application of artificial intelligence (AI) based computerised clinical support systems. Wyatt and Spiegelhalter’s definition of CDSS as ‘…active knowledge systems which use two or more items of patient data to generate case-specific advice’2 is widely used and is useful, but as this definition currently stands it excludes simple memory aids to clinicians (such as the more basic computer-assisted history taking systems discussed in the previous chapter). It further also excludes the more novel patient support tools (sometimes known as decision support systems), systems that make clinical decisions on population level data and systems that provide additional information to a clinician (such as prognosis) without explicitly supporting the decision-making process. A related, but more inclusive definition has more recently been proposed, defining these tools as providing ‘…access to knowledge stored electronically to aid patients, carers and service providers in making decisions on health care’.3 There is thus, as with most other eHealth applications, need for greater clarity in what exactly is being referred to when using the term CDSS. In the context of this chapter, we will be focusing on studies investigating computerised systems that draw on specific patient data to support the management of individual patients by professionals. 8.2.2 Description of use Computerised decision support systems vary in design and function.4 Applications can be used by any healthcare worker involved in the provision of healthcare. They can be “stand-alone” or, as is increasingly the case, 237 “integrated” with other clinical information systems, in particular electronic health records (EHRs). Patient data can be inputted by manual entry, transferred from other clinical information systems or transmitted from medical devices. Engagement by professionals with a CDSS may be “active” and/or “passive”, this referring to whether the information is “pushed” to the user– often in real-time so as to be able to support care decisions at the point-ofcare (POC)–or whether professionals need to retrieve or “pull” the advice from the CDSS. In generating the recommendations, the patient data (input) are compared against a knowledge-base (the collection of clinical knowledge) and made sense of by an inference mechanism (the logic). These knowledge-bases may have been developed in-house or, as is increasingly the case, procured from a commercial provider. The underlying reasoning approach or inference mechanism can be highly variable in sophistication and approach, these ranging from the relatively simple clinical algorithms and Bayesian-based approaches to the more advanced rules-based expert systems and machine learning-based approaches.1 The output can also take different forms and can be delivered to a number of destinations occurring at any time before, during or after the patient-professional interaction. Several helpful classification approaches to CDSSs have been proposed, but none appears as yet to be able to do justice to the complex multi-dimensional nature of CDSSs.3-7 Perhaps most interesting in this respect is Sim et al.’s framework, in which they distinguish between evidence-adaptive applications (i.e., those in which the underlying knowledge-base of the CDSS is based on evidence that is continually updated) and those that draw on more static knowledge sources.7 8.2.3 Scope for use As discussed in Chapters 1 and 4, there are concerns now being expressed internationally about major deficiencies in the quality, safety and efficiency of healthcare.8 This is at least in part due to the difficulties that clinicians have in accessing and processing the vast amounts of research now being published, the challenges in applying this to decisions relating to the care of individual 238 patients, and the increasing data that can now be generated in relation to individual patients (e.g. a patient in an intensive care unit who is being continually monitored), which can rapidly lead to cognitive overload.9 The general premise underlying the use of CDSSs is that they support clinicians in making more informed decisions. As such, knowledge-bases can fill the gaps in clinicians’ knowledge and inference mechanisms can aid in the interpretation of patient data to improve clinical decision-making. By improving clinical decision-making, CDSSs have the potential to impact on the quality—the appropriateness, timeliness and effectiveness—and safety of healthcare; they are also widely seen as having the potential to remove some of the inefficiencies of care through rationalising expenditure on diagnostic tests and treatment decisions. CDSSs can be used for a variety of clinical activities such as preventive care, diagnostics, therapeutics, disease management, prognostics and public health surveillance. Theoretically, CDSSs can be used for any speciality of clinical care and in any setting where the requisite knowledge-base and technological infrastructure exists. They can potentially also be used to support research through, for example, identifying eligible patients and supporting data collection according to pre-defined protocols in enrolled patients. The plasticity of this application thus suggests a diverse range of opportunities to impact on care provision. 8.3 Theoretical benefits and risks 8.3.1 Benefits Whilst the range of ways in which CDSSs can be used is potentially endless, the greatest interest currently relates to supporting management decisions rather than the more difficult to model processes of diagnostics and prognostication. The immediate impacts of these tools are likely to be the understanding, confidence and competencies of clinicians, which will it is hoped translate into more appropriate, evidence-based decisions and improved outcomes for patients. Through promoting best practice, supporting monitoring of patients and alerting professionals to potential risks and errors, these clinician-support tools may also result in a reduction of adverse events. 239 It is based on the presumed improvements in both the quality and safety of care that claims about the cost-effectiveness of decision support tend to be made. 8.3.2 Risks There is a risk that the introduction of CDSSs may have a detrimental effect on care processes and outcomes, this resulting either from faulty algorithms resulting in erroneous advice, degradation in practitioner knowledge and resulting over-reliance on the system and the disruption resulting from a change in workflow processes resulting from poorly integrated tools. The known data inaccuracies in patient EHRs are also potentially important considerations in the context of the introduction of tools that draw on such information.10 11 There are also potential legal considerations associated with clinicians ignoring appropriate advice and also institutions investing in poor quality CDSSs that also need to be considered.12 The costs of investing in CDSSs also need to be considered, as these can be substantial. It is therefore important that these potential adverse outcomes, whether at the level of the practitioner, patient or the organisation are considered and, where relevant, studied.13 8.4 Empirically demonstrated benefits and risks We found 24 relevant systematic reviews (SRs) investigating various aspects of the effectiveness and safety of CDSSs (see Appendix 5 for details of these reviews).14-37 These reviews had a range of foci, these including: providing an overview of the impact of eHealth and/or CDSSs; investigating their impact in specific clinical settings (e.g., nursing, primary care, intensive care settings and pathology services) and disease contexts (e.g., diabetes and hypertension); test ordering; and the promotion of preventive and guidelinebased care. In a few cases, these represented substantive updates of earlier reviews,38-41 in which case, we focused on the most recent reports. We also identified two meta-analyses investigating the impact of CDSSs on aspects of preventive care (vaccination and screening) that reported before 1997, but which nonetheless provided some relevant insights.42 43 For the sake of brevity, we do not detail reviews on the use of CDSS for image interpretation. 240 We do, however, undertake a detailed case study of the use of CDSS in the context of mammography screening in the following chapter, which illustrates the most pertinent issues. Reviews specifically focusing on the use of CDSSs in prescribing decisions (ePrescribing) are discussed in Chapter 10, with other aspects of medicine management consideration explored in the case studies presented in Chapters 11 and 12. Given the range of CDSS interventions, study designs, settings, contexts, outcomes and time periods studied, synthesising this body of evidence was in many respects challenging. Also of relevance was that these SRs tended not to report on the full range of practitioner, patient and organisational outcomes discussed above; in particular there was a deficiency in reporting on possible adverse events. 8.4.1 Benefit Practitioner performance A number of SRs have demonstrated that CDSSs can impact positively on practitioner performance, such as the promotion of preventive care and guideline adherence. The SR by Garg et al. is the most important in this respect (and in general in that it evaluates a substantial body of evidence from 100 trials), revealing that CDSSs had positive impacts in 16/21 (76%) of the studies investigating the issuing of reminders for cervical cancer screening, mammography and vaccination.19 Similar findings have also been demonstrated in other systematic reviews for aspects of practitioner performance about relatively simple prevention decisions. Interestingly, a broadly similar finding was observed in much earlier systematic reviews.42 43 For example, Austin et al. found in their meta-analysis of three cervical screening trials that reminders resulted in an increased number of procedures: OR=1.18 (95%CI 1.02, 1.40); similarly, the pooled estimate from the three tetanus vaccination trials demonstrated an increase in vaccination rate: OR=2.82 (95%CI 2.66, 2.98).42 Practitioner performance is more usually improved in relation to guideline adherence and specific tasks in relation to chronic disease management, this 241 having been demonstrated across a range of care settings and disease contexts. Garg et al. for example found improvements in practitioner performance in 23/37 (62%) of the trials studying this outcome.19 The review by Heslemans is interesting in that it focused on studies investigating the impact of CDSSs on adherence with more complex guidelines (which they characterise as “multidimensionality”), but this found improvements in measures of practitioner performance in 7/17 (41%) studies, this suggesting that CDSSs might be more effective with the relatively simple tasks that can easily be modelled.21 This may also be why these tools are less effective in the context of supporting clinical decision making (positive impacts demonstrated in 4/40 (10%) of the diagnostic studies.19 Patient outcomes A recurring theme in this body of work is that patient outcomes tend not to be studied by investigators and when they are the results are on the whole disappointing. For example, fifty-two of the 100 trials retrieved by Garg et al reported on patient outcomes and of these only seven (13%) showed positive patient level outcomes (two of which were in the context of medicines management interventions, which will be considered further in Chapter 10).19 Similar disappointing patient-level outcomes have also been reported by, amongst others, Georgiou et al.,20 Heselmans et al.,21 Jamal et al.,22 Mador et al.27 and Main et al.28 Cost-effectiveness Another important concern highlighted in the literature relates to the issue of cost-effectiveness. Garg et al were unable to uncover much data to estimate the cost-effectiveness of CDSSs, and most of the other SRs we found are silent in this respect, this perhaps also reflecting the lack of available data.19 The recently reported review by Main et al.,28 which set out to compare the effectiveness of a CDSS-based approach when compared to simpler order communications support (also know as computerised physician order entry) in the context of test requests only found two studies yielding economic data: one showed a cost reduction with CDSS, whilst the other found an increase in costs for the tests requested. 242 8.4.2 Risks The SRs reviewed have, as noted above, tended not to report adverse outcomes. This may reflect the fact that these interventions are largely safe or, alternatively, it may represent a failure adequately to consider such risks, whether in the primary studies or the secondary reviews. Looking more widely, concerns have been expressed about the effects of poor data quality—accuracy and completeness—on CDSS functioning. For instance, Berner et al. explored the effect of incomplete patient data on CDSS accuracy and found that when the available data were input into the CDSS, the missing data elements resulted in inappropriate and unsafe recommendations in almost 77% of encounters.44 8.5 Implications for policy, practice and research 8.5.1 Policy Computerised decision support systems are currently outwith the remit of the Medicines Health Regulation Authority (MHRA) in the UK and the Federal Drugs Authority (FDA) in the USA. However, in view of the potential for the introduction of new risks, it is important that consideration be given to the need for regulation of CDSSs.12 As discussed in Chapter 11, such regulation could also address the issue of “defensive” practices by software developers (e.g. the factoring into the algorithms of any conceivable possibility), which may through “over-alerting” inadvertently increase the risk of adverse events. The policy challenge here however is not to over-regulate so it stifles innovation in this field. 12 Given the considerable costs of developing and maintaining CDSSs, Wright et al have made an interesting suggestion in relation to the possible benefits from sharing the content of CDSSs both within and, more contentiously in the US context, between organisations.45 This has considerable potential applicability to the NHS context where competing partners could be brought together in the national interest to work on collaborative projects. We discuss this issue further in Chapter 11 when considering ePrescribing. It also seems 243 prudent in this respect for the NHS to develop an inventory–that is periodically updated–of decision support tools that are available for and in use in the NHS.28 46 8.5.2 Practice Based on the evidence reviewed, it seems likely that most CDSSs will not result in measurable patient-level benefits; they may however translate into practitioner or process level improvements, particularly in the context of CDSSs aiming to improve aspects of preventive care, guideline adherence and disease management; achieving even this level of benefits seems more challenging in the context of tools aiming to improve diagnosis. It is therefore important that claims made about these decision support tools are carefully scrutinised before investing in these technologies. The research to date has also offered several other pointers as to which tools may prove particularly effective in improving these practitioner/process end-points, which it may also prove helpful to consider when investing in CDSSs, these including: from Garg et al.,19 automatic prompting to use the system (adjusted OR 3.0, 95%CI 1.2, 7.1) and designers of the systems also being involved with the evaluation (adjusted OR 6.6, 95%CI 1.7, 26.7); from Kawamoto et al.25: automatic provision of decision support as part of clinician workflow (adjusted OR 105.0, 95%CI 10.4, ∞); and from Damiani et al.,16 automatic provision of recommendations that are integrated with workflow: (adjusted OR 17.5, 95%CI 1.6, 193.7) and publication year (adjusted OR 6.7, 95%CI 1.3, 34.3). More fundamentally, there is the need for clinical engagement to identify the areas in which CDSSs are most likely to prove useful to clinicians, rather than serving as a technology that is introduced for technology’s sake.47 48 8.5.3 Research There are a number of important unanswered questions in relation to CDSSs that warrant further investigation. Arguably, the most important is to need to establish which contexts, if any, these decision support tools can translate into patient level benefits. Neither the technology or its implementation or maintenance is cheap–it is therefore also crucially important that we obtain a 244 better idea of the cost-effectiveness of CDSSs, particularly in contrast to potentially cheaper and less disruptive approaches. Such economic considerations are particularly important in these times of economic difficulties facing the UK and many other countries. It seems more likely however that the impact of CDSSs aimed at professionals will, at best, be on those professionals, rather than translating into demonstrable downstream improvements in patient outcomes. This is because the causal chain of events to impact on patients is (relatively) long i.e. technology → practitioner action → patient actions → patient outcomes, and also because many of the clinical outcomes of interest are uncommon events (e.g. cervical cancer), which would necessitate studies involving large numbers of people who may need to be followed up over many years in order to have adequately powered studies. There is therefore the need for academic consensus building work on the context in which proxy measures are suitable surrogate markers to study; clinical endpoints should however continue to be studied in order to facilitate future meta-analysis of the types undertaken by Austin et al.42 and Shea et al.43 It is also important that researchers study and report on the full range of potentially relevant outcomes associated with CDSSs, whether showing evidence of benefit or harm, using an appropriate range of experimental and naturalistic designs. 7 13 24 Finally, it is important that future work focuses on the effectiveness of commercial “offthe-shelf” rather than “home-grown” solutions.1 19 33 38 References 1. Coiera E. Guide to Health Informatics. 2nd ed. London: Hodder Arnold, 2003. 2. Wyatt J, Spiegelhalter D. Field trials of medical decision-aids: potential problems and solutions. Proc Annu Symp Comput Appl Med Care 1991; 3-7. 3. National Electronic Decision Support Taskforce (NEDST). Electronic Decision Support for Australia’s Health Sector. Canberra: Australian Government Department of Health and Ageing, 2003. 4. Berlin A, Sorani M, Sim I. A taxonomic description of computer-based clinical decision support systems. J Biomed Inform 2006; 39: 656-67. 5. Perreault L, Metzger J. A pragmatic framework for understanding clinical decision support. J Healthc Inf Manag 1999; 13: 5-21. 6. Broverman CA. Standards for clinical decision support systems. J Healthc Inf Manag 1999; 13: 23-31. 245 7. Sim I, Gorman P, Greenes RA, Haynes RB, Kaplan B, Lehmann H et al. Clinical decision support systems for the practice of evidence-based medicine. J Am Med Inform Assoc 2001; 8: 527-34. 8. Institute of Medicine. Crossing the Quality Chasm: A New Health System for the 21st Century. Washington, D.C., National Academy Press, 2001. 9. Smith R. Strategies for coping with information overload. BMJ 2010; 341: c7126. 10. Hogan WR, Wagner MM. Accuracy of data in computer-based patient records. J Am Med Inform Assoc 1997; 4: 342-55. 11. Jordan K, Porcheret M, Croft P. Quality of morbidity coding in general practice computerized medical records: a systematic review. Fam Pract 2004; 21: 396412. 12. Fox J, Thomson R. Clinical decision support systems: a discussion of quality, safety and legal liability issues. Proc AMIA Symp 2002; 265-269. 13. Ash JS, Sittig DF, Campbell EM, Guappone KP, Dykstra RH. Some unintended consequences of clinical decision support systems. AMIA Annu Symp Proc 2007; 26-30. 14. Balas EA, Krishna S, Kretschmer RA, Cheek TR, Lobach DF, Boren SA. Computerized knowledge management in diabetes care. Med Care 2004; 42: 610-21. 15. Bryan C, Boren SA. The use and effectiveness of electronic clinical decision support tools in the ambulatory/primary care setting: a systematic review of the literature. Inform Prim Care 2008; 16: 79-91. 16. Damiani G, Pinnarelli L, Colosimo SC, Almiento R, Sicuro L, Galasso R, et al. The effectiveness of computerized clinical guidelines in the process of care: a systematic review. BMC Health Serv Res 2010; 10: 2. 17. Delpierre C, Cuzin L, Fillaux J, Alvarez M, Massip P, Lang T. A systematic review of computer-based patient record systems and quality of care: more randomized clinical trials or a broader approach? Int J Qual Health Care 2004; 16: 407-16. 18. Dexheimer JW, Talbot TR, Sanders DL, Rosenbloom ST, Aronsky D. Prompting clinicians about preventive care measures: a systematic review of randomized controlled trials. J Am Med Inform Assoc 2008; 15: 311-20. 19. Garg AX, Adhikari NK, McDonald H, Rosas-Arellano MP, Devereaux PJ, Beyene J, et al. Effects of computerized clinical decision support systems on practitioner performance and patient outcomes: a systematic review. JAMA 2005; 293: 122338. 20. Georgiou A, Williamson M, Westbrook JI, Ray S. The impact of computerised physician order entry systems on pathology services: a systematic review. Int J Med Inform 2007; 76: 514-29. 21. Heselmans A, Van de Velde S, Donceel P, Aertgeerts B, Ramaekers D. Effectiveness of electronic guideline-based implementation systems in ambulatory care settings - a systematic review. Implement Sci 2009; 4: 82. 22. Jamal A, McKenzie K, Clark M. The impact of health information technology on the quality of medical and health care: a systematic review. HIM J 2009; 38: 2637. 23. Jerant AF, Hill DB. Does the use of electronic medical records improve surrogate patient outcomes in outpatient settings? J Fam Pract 2000; 49: 349-57. 24. Kaplan B. Evaluating informatics applications--clinical decision support systems literature review. Int J Med Inform 2001; 64: 15-37. 25. Kawamoto K, Houlihan CA, Balas EA, Lobach DF. Improving clinical practice using clinical decision support systems: a systematic review of trials to identify features critical to success. BMJ 2005; 330: 765. 26. Lisboa PJ, Taktak AF. The use of artificial neural networks in decision support in cancer: a systematic review. Neural Netw 2006; 19: 408-15. 246 27. Mador RL, Shaw NT. The impact of a Critical Care Information System (CCIS) on time spent charting and in direct patient care by staff in the ICU: a review of the literature. Int J Med Inform 2009; 78: 435-45. 28. Main C, Moxham T, Wyatt JC, Kay J, Anderson R, Stein K. Computerised decision support systems in order communication for diagnostic, screening or monitoring test ordering: systematic reviews of the effects and cost-effectiveness of systems. Health Technol Assess 2010; 14: 1-227. 29. Montgomery AA, Fahey T. A systematic review of the use of computers in the management of hypertension. J Epidemiol Community Hlth 1998; 52: 520-25. 30. Randell R, Mitchell N, Dowding D, Cullum N, Thompson C. Effects of computerized decision support systems on nursing performance and patient outcomes: a systematic review. J Health Serv Res Policy 2007; 12: 242-49. 31. Shiffman RN, Liaw Y, Brandt CA, Corb GJ. Computer-based guideline implementation systems: a systematic review of functionality and effectiveness. J Am Med Inform Assoc 1999; 6: 104-14. 32. Rosado B, Menzies S, Harbauer A, Pehamberger H, Wolff K, Binder M et al. Accuracy of computer diagnosis of melanoma: a quantitative meta-analysis. Arch Dermatol 2003; 139: 361-67. 33. Shekelle PG, Goldzweig CL. Costs and Benefits of Health Information Technology: an updated systematic review. London: The Health Foundation, 2009. 34. Shojania KG, Jennings A, Mayhew A, Ramsay C, Eccles M, Grimshaw J. Effect of point-of-care computer reminders on physician behaviour: a systematic review. CMAJ 2010; 182: E216-25. 35. Sintchenko V, Magrabi F, Tipper S. Are we measuring the right end-points? Variables that affect the impact of computerised decision support on patient outcomes: a systematic review. Med Inform Internet Med 2007; 32: 225-40. 36. Smith MY, DePue JD, Rini C. Computerized decision-support systems for chronic pain management in primary care. Pain Med 2007; 8: S155-66. 37. Tan K, Dear PR, Newell SJ. Clinical decision support systems for neonatal care. Cochrane Database Syst Rev 2005; 2: CD004211. 38. Chaudhry B, Wang J, Wu S, Maglione M, Mojica W, Roth E, et al. Systematic review: impact of health information technology on quality, efficiency, and costs of medical care. Ann Intern Med 2006; 144: 742-52. 39. Shekelle PG, Morton SC, Keeler EB. Costs and Benefits of Health Information Technology. Rockville: AHRQ, 2006. 40. Johnston ME, Langton KB, Haynes RB, Mathieu A. Effects of computer-based clinical decision support systems on clinician performance and patient outcome. A critical appraisal of research. Ann Intern Med 1994; 120: 135-42. 41. Hunt DL, Haynes RB, Hanna SE, Smith K. Effects of computer-based clinical decision support systems on physician performance and patient outcomes: a systematic review. JAMA 1998; 280: 1339-46. 42. Austin SM, Balas EA, Mitchell JA, Ewigman BG. Effect of physician reminders on preventive care: meta-analysis of randomized clinical trials. Proc Annu Symp Comput Appl Med Care 1994: 121-24. 43. Shea S, DuMouchel W, Bahamonde L. A meta-analysis of 16 randomized controlled trials to evaluate computer-based clinical reminder systems for preventive care in the ambulatory setting. J Am Med Inform Assoc 1996; 3: 399409. 44. Berner ES, Kasiraman RK, Yu F, Ray MN, Houston TK. Data quality in the outpatient setting: impact on clinical decision support systems. AMIA Annu Symp Proc 2005; 41-45. 45. Wright A, Bates DW, Middleton B, Hongsermeier T, Kashyap V, Thomas SM, et al. Creating and sharing clinical decision support content with Web 2.0: Issues and examples. J Biomed Inform 2009; 42: 334-46. 247 46. Sittig DF, Wright A, Simonaitis L, Carpenter JD, Allen GO, Doebbeling BN, et al. The state of the art in clinical knowledge management: an inventory of tools and techniques. Int J Med Inform 2010; 79: 44-57. 47. Catwell L, Sheikh A. Information technology (IT) system users must be allowed to decide on the future direction of major national IT initiatives. But the task of redistributing power equally amongst stakeholders will not be an easy one. Inform Prim Care 2009; 17: 1-4. 48. Handler JA, Feied CF, Coonan K, Vozenilek J, Gillam M, Peacock PR Jr, et al. Computerized physician order entry and online decision support. Acad Emerg Med 2004; 11: 1135-41. 248 Chapter 9 Case study: Computerised decision support for mammography screening Summary • Breast cancer is one of the leading causes of cancer death in the developed world. Screening mammography is recommended by many major institutions for women in order to reduce mortality from breast cancer. • Routine breast cancer screening begins in women from the age of 50 and continues every three years until the age of 70 years; ongoing scans are available to women who request this service beyond the age of 70 years. • In England, “gold standard” screening mammography involves four images: two views per breast, read by a reader trained in mammography; these readings are then, in most cases, independently double checked by a second reader. This second reading increases sensitivity and specificity of cancer detection and is at present usually performed by a radiologist. • Increases in the proportion of women being screened, the frequency with which they are screened and the lack of trained professionals places a considerable resource burden on health systems. Although the gold standard for screening mammography, readings made by professionals are subject to factors that risk the quality of readings and therefore patient safety. • Computerised decision support systems for image interpretation (most commonly referred to as computer-aided detection) denotes using the output of software analysis as a “second read” before making a diagnosis. • Computer-aided detection for screening mammography has the potential to improve the sensitivity and specificity of screening mammography and reduce the burden placed on health services. 249 Summary continued… • Recent studies suggest that computer-aided detection can result in comparable detection rates of breast cancer to that achieved by double human readings. • Computer-aided detection does however result in an increase in recall rate with an associated increase in the number of diagnostic procedures in healthy women. • There is as yet a lack of clear evidence from randomised controlled trials of the impact of computer-aided detection for screening mammography on improving long-term patient outcomes or on the cost-effectiveness of this approach when compared with double reading. • In the absence of such evidence, the NHS Screening Programme should continue with the current practice of double-reading of mammograms by humans. 9.1 Introduction Computerised decision support systems (CDSSs) have been employed for a variety of clinical activities including image interpretation, with the intention of improving the quality and safety of healthcare (see Chapter 8). Although a type of CDSS, these applications are generally referred to in the literature as computer-aided detection (CAD). In this case study, we consider CAD for screening mammography as an example of a CDSS, which illustrates some important issues typically encountered when considering the implementation and evaluating the impact of new eHealth applications. This case study demonstrates that, despite theoretically and some empirically proven benefits, novel technologies do not always translate into improvements in everyday healthcare practice. We use this case study to highlight some of the reasons underlying the lack of effectiveness of CAD in practice. The purpose of a screening programme is to identify as many people as possible with pathological findings in the early stages of disease and therefore have the opportunity to initiate treatment early. Breast cancer is one of the 250 leading causes of cancer death in women in the developed world.1 2 Detection of pathology and initiation of treatment in the early stages of breast cancer leads to reduced mortality, but mammography-based screening also may lead to over-diagnosis.3 Mammography is the “gold standard” procedure for breast cancer detection in screening programmes. Mammography is a technique that uses low dose xrays to visualise early malignant changes in breast tissue. In England, it involves four images: two views per breast, read by a reader trained in mammography; these readings are then, in most cases, independently double checked by a second reader.4 Although considered the gold standard, this procedure is not ideal as up to one-third of visible cancers are missed by screening programmes.5 There are several factors recognised as possibly contributing to suboptimal readers’ performances in screening programmes. These include:5-8 • Difficulties resulting from reading large numbers of normal radiographs whilst at the same time needing to remain vigilant about the very few abnormalities; this is because continuous reading of normal mammograms is monotonous, which can lead to reading fatigue with the resulting heightened risk of overlooking rare pathological findings • Low image quality • Inappropriate viewing conditions • Distractions • Oversight of pathology in one region of the breast resulting from focusing on more obvious findings in another region • Reliance on prior knowledge of the most probable locations of cancers • Insufficient “reading” skills and/or experience. With increasing numbers of women being screened, an increase in views per person, coupled with a lack of radiologists trained in mammography has created conditions that have in the UK necessitated a number of other healthcare professionals—such as breast clinicians and radiographers—to 251 become increasingly involved in reading mammograms. Their training and experience remain critical to the accuracy of diagnosis. Computer-aided detection in mammography refers to establishing a diagnosis with the help of specific pattern recognition software that marks suspicious features on mammograms thereby attracting readers’ attention to possible pathology.9 10 CAD analysis typically takes place after the initial reading by a radiologist thus taking the place of a second human reader.11 The widespread use of CAD in many centres in the US12 raises the possibility of this eHealth application being adopted in other healthcare settings. It is, however, important that prior to any such decision being taken that the benefits and risks associated with CAD for mammography be critically assessed. 9.2 Theoretical considerations 9.2.1 Potential benefits Computer-aided detection has the potential to aid radiologists by marking suspicious lesions thereby reducing the possibility of missing pathological findings.7 13 In other words, CAD could increase the sensitivity of screening mammography, i.e. reduce the number of false negatives. Increased sensitivity should in turn translate into fewer missed cancers and hence improved survival rates. Approximately 95 per cent of women with abnormalities on screening mammograms do not have breast cancer and a variable proportion of these women undergo unnecessary additional diagnostic procedures and/or treatment.3 Accurate algorithms applied to mammograms could potentially “recognise” only the true positive lesions with few or no false-positives thus increasing specificity with the computer-aided analysis. Increased specificity would then translate into reduced recall rates, reduced number of unnecessary diagnostic procedures and treatments and consequently, reduced risks associated with these interventions. 252 If high sensitivity and specificity were achieved this could then obviate the need for the current dual human reading of mammograms, with the potential for efficiency gains.14 Furthermore, CAD systems have the potential to increase the accuracy and consistency of less-skilled readers.15 9.2.2 Potential risks Flaws in CDSSs’ software design can however inadvertently lead to a deterioration of patient safety. For example, if CAD results in a decrease in specificity this will result in an increased number of false positives, resulting in unnecessary anxiety, testing and associated procedure-related complications.3 16-18 On the other hand, low sensitivity (i.e. the absence of clinically relevant prompts) might also detrimentally impact on readers’ decision making. Alberdi et al. emphasise that readers might base their decisions to not recall patients on the absence of computer prompts.19 Since CAD analysis is unlikely to be able to identify all lesions, a negative test result might induce a false sense of security. Theoretically, such reliance on the CAD support might also result in a deskilling effect for readers—reducing their ability to interpret images without the support of CAD. 9.3 Empirically demonstrated impact We found two relevant systematic reviews (SRs),20 21 which have shown that there is to date no clear evidence that CAD in screening mammography programmes improves patient outcomes. The first SR by Taylor et al.,20 based on an analysis of 10 prospective observational studies, found that a single human reading and CAD resulted in no increase in cancer detection rate (OR 1.04, 95%CI 0.96, 1.13), but an increase in recall rate (OR 1.13, 95%CI 1.05, 1.23) when compared to double human reading. Based on this analysis, they concluded that double human reading remains the preferred model. 253 The more recent SR by Noble et al.,21 which drew on largely the same body of evidence, came to a similar conclusion. The authors identified three studies that allowed sensitivity and specificity of CAD to be calculated, this finding a pooled sensitivity of 86.0% (95%CI 84.2, 87.6) and a pooled specificity of 88.2% (95%CI 88.1, 88.3). Five studies yielded data, which allowed the calculation of rates of correct cancer diagnosis and invasive investigational procedures in women without breast cancer. This found that CAD resulted in 50 (95%CI 30, 80) additional correct breast cancer diagnoses per 100,000 women screened, but there was an additional 1,190 (95%CI 1090, 1290) recalls in healthy women and 80 (95%CI 60, 100) biopsies in these women. They did not however find any data on whether these additional diagnoses translated into improved patient outcomes, which is important because it has been suggested that CAD might preferentially detect slow-growing cancers that might not have resulted in clinical problems during patients’ lifetime. The recent randomised controlled trial (RCT) from the CADET II Group has also generated broadly similar findings.22 This English multi-centre equivalence trial involving 31,057 women compared the effectiveness of single-reading with CAD and double reading. It found equivalent numbers of cancers detected by double reading and single reading with CAD (87.7% vs. 87.2%; P=0.89), but higher recall rates with CAD (3.9% vs. 3.4%; P<0.001). No data were however presented on patient outcomes associated with cancer detection or indeed on cost-effectiveness considerations. In a follow-on analysis, the authors have shown that cancers with different radiological features were detected using these two different approaches.23 The clinical implications of this observation is at present unclear. 9.4 Implications for practice, policy and research Overall, the evidence suggests that single reading with CAD is equivalent to double reading by humans in detecting cancer, but there is an increase in recall rate and associated diagnostic procedures in those screened using CAD. There is as yet no evidence on whether CAD-based screening translates into improved patient outcomes or indeed whether this is costeffective to health systems. Based on the available evidence, it is we feel 254 appropriate that in developed country settings such as the UK, where double screening is the routine, screening programmes should continue with dual human screening. Given the strong theoretical grounds for benefits associated with CAD, it is important to consider why this technology is currently failing to realise its potential. Different developers of CAD systems use different algorithms for image analysis and this might, in some cases, result in different prompts for the same lesions.24 Although the output given by various brands and versions of CAD software is very similar, possible small differences in output have not yet been properly evaluated. And whilst application invalidity (low specificity for example) might play a part in its relative lack of diffusion, it appears that the automation of image interpretation also proves problematic for other reasons. Providing reminders concerning preventive care and calculating drug doses are examples of activities that are readily automatable (see Chapters 8 and 10). The art of diagnosis or image interpretation is just that, and it is therefore extremely difficult to automate; though not well studied, but probably also important, is that professional autonomy might be perceived as being encroached by applications such CAD. Although there are difficulties associated with reading mammograms there have been no calls to greatly improve the accuracy of double-reading and the use of CAD for screening mammography screening thus seems to be technology driven rather than fulfilling a genuine clinical need.25 References 1. American Cancer Society. Breast cancer facts and figures. Available from: http://www.cancer.org/Research/CancerFactsFigures/BreastCancerFactsFigure s/breast-cancer-facts--figures-2009-2010 (last accessed 8/1/11). 2. Cancer Research UK, Cancer mortality - UK statistics. Available from: http://info.cancerresearchuk.org/cancerstats/mortality/ (last accessed 8/1/11). 3. Gotzsche PC, Nielsen M. Screening for breast cancer with mammography. Cochrane Database Syst Rev 2009; 4: CD001877. 255 4. NHS Breast Screening Programme. Available from: http://www.cancerscreening.nhs.uk/breastscreen/index.html (last accessed 8/1/11). 5. Yankaskas BC, Schell MJ, Bird RE, Desrochers DA. Reassessment of breast cancers missed during routine screening mammography: a community-based study. AJR Am J Roentgenol 2001; 177: 535-41. 6. Erickson BJ, Bartholmai B. Computer-aided detection and diagnosis at the start of the third millennium. J Digit Imag 2002; 15: 59-68. 7. Astley SM. Computer-aided detection for screening mammography. Acad Radiol 2004; 11: 1139-43. 8. Huynh PT, Jarolimek AM, Daye S. The false-negative mammogram. Radiographics 1998; 18: 1137-54. 9. Giger ML, Karssemeijer N, Armato SG, III. Computer-aided diagnosis in medical imaging. IEEE Transactions on Medical Imaging 1920; 1205-08. 10. Astley SM, Gilbert FJ. Computer-aided detection in mammography. Clin Radiol 2004; 59: 390-99. 11. Bywood P, Newton S, Merlin T, Braunack-Mayer A, Hiller J. Computer Aided Detection Systems in Mammography. Canberra: Australia and New Zealand Horizon Scanning Network. Horizon Scanning Report, 2004. 12. Doi K. Current status and future potential of computer-aided diagnosis in medical imaging. Brit J Radiol 2005: 78 (Spec No 1): S3-S19. 13. Taylor P, Champness J, Given-Wilson R, Johnston K, Potts H. Impact of computer-aided detection prompts on the sensitivity and specificity of screening mammography. Health Technol Assess 2005; 9: iii:1-58. 14. Dinnes J, Moss S, Melia J, Blanks R, Song F, Kleijnen J. Effectiveness and cost-effectiveness of double reading of mammograms in breast cancer screening: findings of a systematic review. Breast 2001; 10: 455-63. 15. Blue Cross Blue Shield Association. Technology Evaluation Center Assessment: Computer-aided Detection with Full-field Digital Mammography, 2006. 16. Stoate HG. Can health screening damage your health? J R Coll Gen Pract 1989; 39: 193-95. 17. Thornton H, Baum M. Should a mammographic screening programme carry the warning: screening can damage your health!? Br J Cancer 1999; 79: 691-92. 18. Brewer NT, Salz T, Lillie SE. Systematic review: the long-term effects of falsepositive mammograms. Ann Intern Med 2007; 146: 502-10 19. Alberdi E, Povyakalo AA, Strigini L, Ayton P, Hartswood M, Procter R et al. Use of computer-aided detection (CAD) tools in screening mammography: a multidisciplinary investigation. Brit J Radiol 2005; 78(Spec No 1): S31-40. 20. Taylor P, Potts HW. Computer aids and human second reading as interventions in screening mammography: two reviews to compare effects on cancer detection and recall rate. Eur J Cancer 2008; 44:798–807. 21. Noble M, Bruening W, Uhl S, Schoelles K. Computer-aided detection mammography for breast cancer screening: systematic review and metaanalysis. Arch Gynecol Obstet 2009; 279: 881-90. 22. Gilbert FJ, Astley SM, Gillan MG, Agbaje OF, Wallis MG, James J, Boggis CR, Duffy SW; CADET II Group. Single reading with computer-aided detection for screening mammography. N Engl J Med 2008; 359: 1675-84 23. James JJ, Gilbert FJ, Wallis MG, Gillan MG, Astley SM, Boggis CR, et al. Mammographic features of breast cancers at single reading with computeraided detection and at double reading in a large multicenter prospective trial of computer-aided detection: CADET II. Radiology 2010; 256: 379-86. 24. Kim SJ, Moon WK, Kim SY, Chang JM, Kim SM, Cho N. Comparison of two software versions of a commercially available computer-aided detection (CAD) system for detecting breast cancer. Acta Radiol 2010; 51: 482-90. 256 25. Catwell L, Sheikh A. Information technology (IT) system users must be allowed to decide on the future direction of major national IT initiatives. But the task of redistributing power equally amongst stakeholders will not be an easy one. Inform Prim Care 2009; 17: 1-4. 257 Chapter 10 ePrescribing Summary • There is now a growing body of data indicating the high incidence of medication errors that occur in a range of clinical settings; whilst the majority of these errors are relatively minor, some will translate into morbidity and, in a minority of cases, death. Many of these errors are now believed to be preventable. • A variety of ePrescribing systems have been developed as information technology-enabled responses to minimising of the risk of prescribingrelated harm and/or improving the organisational efficiency of healthcare practices in relation to prescribing. • These ePrescribing initiatives range from support for prescribers on placing medication orders and prescribing decision to the broader more visionary perspectives of cross-organisational integration often advocated in key policy documents. As technology has advanced, the scope of ePrescribing has also expanded and charting the evolution of definitions shows the emergence of a progressively more complex picture. In essence, however, this term embraces both the simpler computerised physician (or provider) order entry systems and the more sophisticated computerised decision support system functionality. • There is evidence that practitioner performance and surrogate prescribing outcomes can be improved through ePrescribing. Positive evidence on safer prescribing outcomes has tended to be reported in the more recent studies. However, overall the evidence showing improved prescribing safety has not been shown to lead to reduced patient morbidity and/or death. • Evidence of benefits from ePrescribing applications has in the main been derived from evaluations of “home-grown” applications from a few centres of excellence in the United States. Most applications in use are, however, commercially procured and typically lack the sophistication of the more tailored home-grown systems. 258 Summary continued… • Poorly designed applications and a failure to appreciate the organisational implications associated with their introduction may introduce unexpected new risks to patient safety and the evidence from evaluations of these home-grown systems is therefore not easily transferable to settings implementing commercial systems. • The persistent high rates of over-riding of alerts generated by the more advanced ePrescribing systems remains a major concern; finding ways of increasing the sensitivity and perceived relevance of alerts is a major issue that warrants further investigation. • This will, amongst other things, necessitate undertaking detailed medico-legal work to more accurately quantify the risks to system developers of changing from the current defensive practice in which they take a “belts and braces” approach to generating warnings to one in which there is more selective warning of major risks. • There is also a need to investigate–in a few carefully selected contexts– the role of “hard-stop” restrictions, which prevent the over-riding of alerts. • Given that electronic health record systems are now being introduced into NHS hospitals in England, there is a need to consider introducing ePrescribing systems, preferably in an evaluative context that allows the effectiveness and cost-effectiveness of these new systems to be established. 10.1 Introduction There is now a considerable body of evidence demonstrating that prescribing errors are common and that these are responsible for considerable– potentially avoidable–patient morbidity and mortality.1 2 For example, recent UK data indicate that medication-related harm is frequently implicated in admission to hospitals3 and furthermore that an estimated one in seven hospitalised patients experience one or more episodes of prescribing-related 259 harm.4 Many studies have now demonstrated that a large proportion of these adverse drug reactions (ADRs) are potentially preventable.5 6 Given the vast array of drugs now available and the considerable scope for their interaction with aspects of the patients’ history and/or other coprescribed treatments, it is simply no longer feasible for clinicians to know about, retain and judiciously draw on all such information from memory. Electronic prescribing (henceforth referred to as “ePrescribing”) has the potential to support professionals by helping them to identify and select potentially appropriate treatments and doses, and also by using patient specific and other local data to guide treatment decisions. In this chapter, we review the potential and empirically demonstrated benefits and risks associated with ePrescribing, building on the more generic discussions on computerised decision support systems (CDSS) in Chapter 8. A more focused critique of the literature on the potential of information technology (IT) to support prescribing in two exemplary particularly high risk contexts (i.e. the use of oral anticoagulants in primary care and approaches to minimising risk of repeat drug allergy in hospitals) is presented in the case studies in the following two chapters. 10.2 Definition, description and scope for use 10.2.1 Definition There is no agreed definition of ePrescribing. For example, Dobrev et al. have defined ePrescribing as “the use of computing devices to enter, modify, review, and output or communicate drug prescriptions”.7 In contrast, the definition used by NHS Connecting for Health (NHS CFH) is somewhat broader, including aspects of the governance of prescribing decisions i.e. “utilisation of electronic systems to facilitate and enhance the communication of a prescription or medicine order, aiding the choice, administration and supply of a medicine through knowledge and decision support and providing a robust audit trail for the entire medicines use process”.8 This definition embraces the use of technology to support the whole process of medication management and it also implies the integration of medication systems with existing electronic health record (EHR) systems (see Chapter 3). The 260 taxonomy of ePrescribing systems proposed by Teich et al. emphasises the importance of integration with EHRs (see Figure 10.1). 9 Figure 10.1. The level of sophistication of ePrescribing systems. Modified based on “Teich et al.” (2004)9 (permission obtained) Also of relevance is the degree of support that prescribers are offered. Many of the initially developed systems were, for example, computerised physician (or provider) order entry (CPOE) systems that provided clinicians with dropdown menus to support the prescribing decisions that were being made. More recently, however, the focus of developers has been to build on this basic prescribing support and offer prescribers functionality that takes into account other relevant contextual information about the patient using in-built decision support (see Chapter 8 for a more general discussion about computerised decision support systems (CDSSs). Table 10.1 provides a framework for considering the degree of decision support offered. It should be noted that both CPOE and CDSS can be used in other contexts, in particular the ordering of investigations; these other uses will not however be considered further in the context of this chapter. 261 Level 1 Standalone electronic prescription writer or CPOE Level 2 Electronic drug reference manual Level 3 Electronic prescription writing and electronic transmission of prescriptions—connectivity to dispensing site Level 4 Patient specific prescription creation or refilling Level 5 Basic decision support functionality (integrated or interfaced)— dosage (default and frequencies) and formulary support Level 6 Drug management—access to electronic medication administration record (eMAR) checks for allergies, drug interactions and duplicate therapies Level 7 Integration with an EHR Level 8 Integration with EHR and other clinical information systems (radiology, laboratory and pharmacy information systems) Level 9 Advanced decision support functionality (integrated or interfaced): adjusting dosages in light of patient characteristics (e.g. ethnicity), physiologic status (e.g. uraemia) and co-morbidities; other medications currently being taken; previous response to the drug, single, daily and life dose limits Table 10.1 Levels of system sophistication As is evident from the above discussion, the term “ePrescribing” thus encompasses a wide range of systems, these including both CPOE and CDSS and varying degrees of integration with other electronic record systems. In the absence of any agreed definition, ePrescribing is used in this chapter as an inclusive term referring to at minimum the electronic generation of prescriptions, but which may include point-of-care (POC) decision support and, amongst other things, electronic communication of the prescription information to other professionals and agencies involved with medicines management. 10.2.2 Description of use Prescribing is a complex organisational practice, including a range of processes spread across locations and involving diverse actors, so it is unsurprising that that ePrescribing systems are also organisationally complex; the choices available in their implementation and dimensions that can be included in their evaluation are hence also multifarious. Figure 10.2 depicts 262 the complexity of ePrescribing processes. It shows how ePrescribing can involve different healthcare professionals at different points of prescribing procedures and how these may require professionals to have access to patients’ healthcare or medical records to prescribe appropriately. Medication errors can occur at any point in the prescribing processes i.e. prescribing, transmitting the order, dispensing, administration and monitoring, and ePrescribing systems can therefore potentially support any of these functions. 10 Figure 10.2 High-level ePrescribing architecture Source: Bell et al.11 (permission obtained) ePrescribing systems have been developed for use in a range of healthcare settings, these ranging from primary care to hospital-based care. This issue of setting of use is emphasised in the taxonomy developed by Cornford et al.12 i.e.: • Pharmacy-based systems • Clinical specialty-based systems (e.g. those used in oncology, renal medicine and intensive care) • Components or modules of larger hospital information system packages • Home-grown software. 263 This framework also points to the importance of the genealogy of the system i.e. whether it is home-grown or commercially procured, this also having been highlighted by several other experts (discussed below). The term ePrescribing may therefore include systems with a range of functions and implemented in a wide range or organisational contexts. When critically reviewing the evidence on the impacts of ePrescribing systems it is important to be sensitive to this variation, as the benefits and risks may be influenced by the system functionality and implementation context. By treating all implementations as being commensurate and aggregating evidence across heterogeneous systems there is a risk that significant insights into system design and implementation will be overlooked. It follows from the discussion above that there are four dimensions that particularly need to be considered: 1. Interoperability (stand-alone vs. integrated with other health information systems) 2. Functionality (CPOE vs. CDSS-based) 3. The degree of customisation (home grown vs. commercially procured) 4. Setting of use (e.g. primary care vs. secondary care). These four dimensions are depicted in Figure 10.3. 264 1. Interoperability: Stand-alone vs. integrated 3. The degree of customisation: Home-grown vs. commercial 2. Functionality: Basic support vs. advanced support/ alerts 4. Setting of use: Primary care (including in-general, intensive care units, pediatric care) vs. Secondary care (including ambulatory and pharmacy settings) Figure 10.3 Four dimensional taxonomy of technological aspects of ePrescribing systems to understand local implementation landscape Dimension 1: Degree of inter-operability and integration This refers to the degree of integration of ePrescribing systems into other healthcare information technologies such as EHRs. ePrescribing systems may be modules within integrated IT systems, linking them to other functional systems, including patient records, accounting systems or inventory systems, or they may be stand-alone systems, with little or no integration with the data held on other systems. Some studies suggest that increasing integration with other systems is likely to be associated with the realisation of greater operational and other benefits.13 15 Dimension 2: Degree of decision support This relates to the extent of decision-making support embedded in the systems. An ePrescribing system may simply automate aspects of the preexisting paper-based system, but ePrescribing systems can also alert prescribers and pharmacists to prescribing decisions that break rules embodied in the systems. The categorisation of tiers of alerts fired, whether the alerts warn or block decisions, the rule-setting for alert messages, the extent to which decision support rules draw on individual patient records and the creation of knowledge-bases for prescription, may vary across countries, healthcare organisations, specialities and indeed clinical teams. Contextual 265 factors, such as patients’ health care records, drug of medical history, ethnicity, sex, age, weight and local agreements (as embodied for example in practice or hospital-based formularies) add further complexity to the decision making of clinicians. The growing functional complexity of ePrescribing systems is therefore likely to be correlated with their increasing integration with EHR and other local information systems. Dimension 3: System development This refers to the genealogy of the ePrescribing platform–i.e. whether it is a bespoke locally developed (or “home-grown”) system or a standardised commercial package. Locally developed systems may, once mature, demonstrate increased benefits because the systems are developed to fit with idiosyncratic working practices and also because clinicians tend to more tolerant of the shortcomings of a system that they have, even in a small way, contributed to the development of. However, due to the cost of maintaining local IT development resources and the cost of developing bespoke solutions, the trend across the IT sector has move away from bespoke development and become more dependent on standardised packages from major commercial suppliers. This axis is a continuum because systems may initially have been developed for local implementation, but their kernel then forms the basis for a commercial package that is sold onto users elsewhere. This is often witnessed in the major software development processes15 (see Box 10.1 as an example of this). Furthermore, this is so because commercial systems allow differing degrees of configurability, ranging from none or very little to substantial. Rothschild points out the limitation this creates in the current ePrescribing literature16 as there is possible limited generalisability of findings from studies focusing on locally developed ePrescribing systems, rather than “off-the-shelf” commercial projects that are more commonly found in nonresearch settings.17 Most early adopted ePrescribing systems were homegrown while commercial systems tend to be seen as more rigid and lacking the adoptability to meet individual organisational needs. 266 Box 10.1 Biography of a hospital ePrescribing system: evolution from bespoke system to commercial package The Prescribing Information and Communication System (PICS) is a portable rulesbased CDSS developed by staff at the University Hospitals Birmingham NHS Foundation Trust and is now available on the market following an implementation partnership agreement with CSE Healthcare systems. It is currently used for inpatients, but is being developed for outpatient and ambulatory care settings. PICS provides ePrescribing and medication functions supported by various patient management services, including laboratory/ radiology ordering and results reporting. For prescriptions, it can provide advanced decision support to check for drug interactions, contra-indications, dose limits and other clinical needs linked with prescribing. Advanced functions include an infection control module, druglaboratory warnings/ prescription-driven test requests, and thrombosis risk assessment and management. The decision support was developed in partnership with the British National Formulary (BNF) so that the drug dictionary is modifiable into a resource with widespread applicability. A number of system interfaces have been developed for users, from simple “pop-up” windows to bi-directional mobile email communications. The system uses standards to ensure interoperability with wider record systems. Dimension 4: Setting The setting of deployment can have obvious consequences in relation to the types and frequencies of errors that might be avoided. Systems may, for example, be implemented in small-scale organisational settings such as a single general practice or a ward or intensive care unit or may be implemented across a group of care providers in, for example, a geographical region. The more grand-scale plans for ultimate national coverage across both primary and secondary care settings envisioned by the National Programme for Information Technology (NPfIT) (see Chapter 3) is in many ways unique. These four factors–i.e. “stand-alone vs. integrated”, “basic vs. advanced”, “home-grown vs. commercial” and “setting”–mutually shape the ways in which 267 ePrescribing systems are implemented and appropriated, and this in turn may affect the quality, safety and efficiency of care. As will be clear from the discussion above, these dimensions are not necessarily mutually exclusive (see, for example, the description in Box 10.1), nonetheless examining the systematic review papers through the lens of this four-axis typology is potentially useful as it can help to interpret the at-times conflicting body of evidence. 10.2.3 Scope for use Accurately estimating the incidence of prescribing errors is complicated by the various definitions used and also be range of approaches to detect and measure errors. More importantly, however, there is a need to appreciate the degree of harm that these result in, this being particularly relevant given the repeated observation that many errors are relatively minor and do not necessarily translate into patient harm. A key question therefore is whether ePrescribing systems can improve the safety of care by reducing risk of errors that are particularly associated with risk of avoidable harm. If so, there are then related important follow-on questions of whether any particular type of ePrescribing system (see discussion above) is any more effective or costeffective than the other types. Medication errors are now known to occur at any point in the medicines management process. As discussed above, depending on the type of ePrescribing system used, any of these prescribing functions may to varying degrees be supported by the technology. Medication prescribing and administration are however the two areas of the delivery process with the highest incidence of error and ePrescribing systems are thus potentially particularly effective in supporting the task of generating and issuing prescriptions (see also Chapter 8 and Bates et al.18). Prescribing of medication is a high volume and high cost activity, with costs of medication in the same group sometimes varying several-fold. There are hence considerable cost savings to be achieved by equipping prescribers with relevant information about the effectiveness, costs and relative cost268 effectiveness of different medications, hence this is another important potential use of ePrescribing systems. In the UK context, ePrescribing systems of varying degrees of sophistication are now routinely used throughout primary care. A key question therefore is to establish whether their use should be extended into hospital in- and outpatient settings. 10.3 Theoretical benefits and risks 10.3.1 Benefits Quality of care The two generic domains of eHealth that are in theory supported by ePrescribing are the storing and managing of data, this support being provided irrespective of the level of functionality of the ePrescribing system, and the informing and supporting of decisions when applications have decision support capabilities (see Chapter 4). Box 10.3 details the range of potential benefits associated with ePrescribing systems. A number of more specific claims are made by NHS CFH on their website.19 269 Box 10.3 Main potential benefits of ePrescribing applications on healthcare quality Potential benefits to patients/carers: • Reduced under- and over-prescribing Professionals: • Standardisation of prescribing practices via the provision of guidelines • Improved communication amongst prescribers and dispensers (e.g. call back queries, instant reporting that item is out of stock, alerts for unfilled, prescriptions or those that have not been renewed) • Instant provision of information about formulary-based drug coverage including on-formulary alternatives and co-pay information • Data available for immediate analysis including post-marketing reporting, drug utilisation review, etc Healthcare systems/organisations: • Reduction in lost orders • Shorter process turn-around time such as the transit time to dispensing site, time until first dose, prescription renewal or refill • Generation of economic savings by linking to algorithms emphasising (offering as a first choice when a drug is selected) cost-effective drugs 270 Patient safety Although healthcare quality and patient safety are inextricably inter-linked (see Chapter 4), much of the premise underpinning the use of ePrescribing relates in particular to improving the safety of medicines management. Errors related to medicines management are probably the most prevalent type of medical error in both primary and secondary care within the UK. Of all types of medicines management errors—prescribing, dispensing, administration, monitoring, repeat prescribing—errors in prescribing decisions are typically the most serious.18 ePrescribing applications should facilitate improved communication between healthcare providers, patient identification, and improved decision and safety support. Improved communication is an inherent benefit of ePrescribing. Improved identification is on the other hand dependent on whether the system is integrated with other clinical information systems such as an EHR (see Chapter 6). Improved decision and safety support is in turn dependent on how alerts are configured and whether decision support is integrated (see Chapter 8); again, the degree to which this is improved is also likely to be dependent on integration with other clinical information systems. Most notably, ePrescribing has the potential to improve patient safety by decreasing errors in prescribing, monitoring and repeat prescribing. The reduction in these types of errors is clearly potentially dependent on the level of system sophistication, i.e. the degree to which the system is integrated with patient data and decision tools such as drug ontologies and the degree to which it is configured (customised) to the needs of individual prescribers. Table 10.1 (above) provides a schematic framework of the extent to which different applications are likely to improve prescribing safety. The types of drug errors potentially mitigated relative to the level of ePrescribing applications’ sophistication include: • Miscommunication of drug orders: due to poor handwriting, confusion between drugs with similar names, misuse of zeroes and 271 decimal points, confusion of metric and other dosing units and inappropriate abbreviations (Levels 1, 2, 3 and 7) • Inappropriate drug(s) selection: due to incomplete patient data such as contraindications, drug interactions, known allergies, current and previous diagnoses, current and previous therapies, test results etc (Levels 4, 5, 6, 8 and 9) • Miscalculation of drug dosage: incorrect selection of route of administration; mistakes with frequency or infusion rate (Levels 2 and 5) • Out-of-date drug information: for example, in references to alerts, warnings etc or information on newly approved drugs (Levels 2 and 6) • Monitoring failures: results of laboratory test monitoring and drug administration monitoring not being taken into account (Levels 6, 7 and 9) • Inappropriate drug(s) selection: due to clinical incompetence (Level 9). The use of ePrescribing facilitates identification of the prescribing clinician and the date of prescription thereby allowing quality control measures to be targeted at specific clinicians. It is also possible to configure a system so that it will not process certain orders that are considered dangerous, for instance the accidental prescribing of oral methotrexate for daily use when the intended prescription is for weekly use.20 Additionally, the applications are capable of linking to other clinical information systems for ADE monitoring and reporting21 and electronic-based representations of prescriptions can form the basis for additional safety measures related to dispensing and administration errors (e.g. automatic dispensing machines and bar-coding of drugs to ensure that patients receive the ordered drug in the correct dose at the specified time; see Chapter 12 for a fuller discussion of this issue in relation to approaches to minimise the risk of recurrent drug allergy). 272 Improved efficiency The more integrated systems should also in theory offer advantages in relation to the provision of drawing on data from a variety of sources and hence offer the potential for more advanced decision support functionality; they may furthermore also increase the efficiency of prescribing by, for example, reducing time to dispensing through better end-to-end communication in hospitals between wards and pharmacy. 10.3.2 Risks Patient safety How is the safety of these applications ensured? In the United States (US), the Food and Drug Administration (FDA) has classified medical software as a medical device since 1976 and therefore requires proof of software verification by demonstrating consistency, completeness and correctness of the software at each stage of the development life cycle. For the following three types of medical software, proof of software validation is also determined by the correctness of the final software product with respect to the users’ needs and requirements:22 • Software as an accessory • Software as a component or part • Stand-alone software. ePrescribing software is however exempted if it is ‘…intended to involve competent human intervention before any impact on human health occurs’.22 In the UK, the Medicines and Healthcare Products Regulatory Agency (MHRA; the UK FDA equivalent) does not consider medical software to be a medical device and therefore does not undertake quality assurance activities. In recognition of this regulative deficit, NHS CFH created a mechanism based on other safety critical software industries’ guidance for medical software products. This quality assessment and assurance however only applies to products developed for NHS CFH and no regulatory paradigm exists in either the US or the UK for commercially available medical software products, these being excluded by the “competent human intervention” clause. 273 This issue is important because although the use of ePrescribing applications for the ordering of drugs should in theory reduce the burden of some types of drug errors, these applications might also introduce new errors. These errors in system design and oversights in development might lead to:23 • Incorrect decision support provided → incorrect medicines ordered and administered → e-Iatrogenesis. Theoretically, risks to patient safety by ePrescribing applications could occur at any point in the use of applications due to errors made by the end-user, such as: • Incorrect patient data input → incorrect decision support → incorrect medicines ordered and administered → e-Iatrogenesis • Incorrect orders selected → incorrect medicines ordered and administered→ e-Iatrogenesis • Incorrect patient selected → inappropriate medicines ordered and administered → e-Iatrogenesis. Dependence on the support provided by the application can furthermore put patients’ safety at risk when support is not available as, for example, when general practitioners (GPs) prescribe in the context of home visits or hospital doctors change practices or hospitals. Similarly, not understanding the nature of the support provided, such as its limitations, can lead prescribers to misjudge the robustness of the support provided. In contrast, ignoring the advice generated may also threaten patient safety. Organisational inefficiency Although the use of ePrescribing is intended to improve the quality of healthcare processes by reducing complexity, the complexity of care often increases as a result of the incorporation of technology into health service delivery. This is primarily due to the significant process changes associated with ePrescribing implementation. Implementing ePrescribing applications 274 may therefore inadvertently impact on the efficiency of care by, for example, resulting in:24 • New or additional work • New training needs • Negative emotions/perceptions • Unwelcome changes to workflows • Parallel use of electronic and paper-based systems • Changes in relationships and/or power dynamics • Time-inefficiencies • Costs. 10.4 Empirically demonstrated benefits and risks We identified 32 reports13 16 25-52 121 122 of 30 systematic reviews (SR) 13 16 25-52 on the benefits and risks associated with ePrescribing systems. Two articles 121 122 are updates of previous SR studies30 31. A detailed description of these studies is presented in Appendices 4 and 5; here we consider the over-riding messages to emerge from this body of work. As an overview of the SR study, evidence on patient outcomes by ePrescribing has been reported by some of the SRs,13 16 25 28 30 31 35 39 40 48-51 121 122 but not all of them. Therefore, the evidence on reported impacts on patient outcomes is limited. For example, no impact on mortality was reported in any included study, but there are some studies demonstrating a limited impact on actual ADEs. A greater effect on potential ADEs and/or serious medication errors (MEs) is reported by many of the SR studies.28 31 32 35 36 39 44 46 49 50 52 122 Some SR studies30 33 34 37 44 46 reported projected cost savings from ePrescribing by extrapolating from the reductions in MEs, prescription dosages and patients’ hospital stays through the use of ePrescribing, but no evidence of direct cost-effectiveness was presented. Furthermore, the majority of studies used weak experimental designs which expose them to the risk of bias. 275 10.4.1 Empirically demonstrated benefits Many of these reviews focused on CPOE and CDSS supporting prescribing.13 16 25 28 30-34 39 44 46 51 121 122 Some of the SRs focused on CDSS for prescription.35 36 47-49 52 We detail most reviews below, omitting those where there is duplication of studies (and conclusions) with other reviews discussed either here or in Chapter 8 on CDSS. Impact on patient safety: mortality, morbidity and surrogate marker of medication errors Sub-standard prescribing practices, such as inappropriate drug selection due to allergies or contraindications and incorrect dosing, are frequently evaluated outcomes. Differences in the way errors are defined and measured make generalising across organisational settings difficult. and Metzger, citing Nebeker et al., write that: For example, Classen 53 ‘One of the ongoing controversies in medication safety is how to measure the safety of the medication system reliably and how to assess the effect of interventions designed to improve the safety of medication use. Clearly, common nomenclature, definitions, and an overall taxonomy for medication safety are essential to this undertaking and the lack thereof has significantly hampered the comparison of various medication safety interventions among different centres. 54 At the heart of an even more fundamental controversy is whether the focus of patient safety should be on errors or adverse events as a means of assessing and improving the safety of the healthcare system.’ Several SR papers 16 25 30 46 51 122 demonstrated limited evidence on the reduction of ADEs through the use of ePrescribing systems. For example, Wolfstadt, et al.51 conducted a SR evaluating ten studies to examine the effect of CPOE with CDSS functionality on a range of ADEs in a range of clinical settings and found that half of the studies found a significant reduction in ADEs. None of the studies however employed randomised controlled trial (RCT) designs, and seven of the 10 studies evaluated home-grown systems. The weak study designs and heterogeneity of patient settings, outcome 276 measures and system genealogies precluded any definitive overall conclusion on effectiveness. There was no discernible impact on other important outcomes such as death. The SR of 27 studies by Ammenweth et al.25 highlighted the complexity of interpreting this body of evidence. This SR, which reported at about the same time, included studies evaluating CPOE systems both without and with CDSS of varying degrees of complexity. The authors included studies that employed controlled and before-after designs undertaken in a range of in-patient settings and evaluating both home-grown and commercial systems. Overall, this evidence showed that 23/25 studies reported reduced rates of medication errors, the effect size ranging from 13-99%. Furthermore, six of the eight studies reporting on potential ADEs found a reduction in the incidence of this outcome (effect size 35-98%). More importantly, two-thirds of the six studies reporting an actual ADE also found a significant reduction of similar size magnitude for potential ADEs. Expressed in another way, however, only four of these 27 studies–which employed study designs that rendered it difficult to, in the authors’ words, ‘exclude a major source of bias’ –found a positive impact on clinical endpoints. To their credit, however, the authors undertook a range of subgroup analysis on the “ME” surrogate outcome: this revealed that systems that were home grown and had advanced CDSS functionality were more likely to be effective than commercial and basic CPOE systems. There were, however, no detectable differences by population studied, inpatient setting or study design. Clamp and Keen30 121 conducted an SR adopting Pawson’s Realist Synthesis approach. Their two reports examined the same 70 studies of healthcare IT systems with variable designs in a range of settings including general medical, surgical, intensive care, paediatric, tertiary, acute and subspecialty renal care settings. Twenty-seven of these studies were concerned with the evaluation of CPOE systems. Although there is no strong evidence that CPOE reduces preventable ADE rate, one study reported decrease in preventable ADEs of 17% 55 while another study showed one ADE would be prevented every 64 days by the use of CPOE in a paediatric unit, but there was no 277 reporting of statistical significance.56 They also concluded that all MEs were reduced significantly by the use of COPE by 40-80% with a significant decrease in serious MEs by 55% 55 and non-serious MEs by 86%.57 Rothschild also assessed ePrescribing in critical care, general inpatient and paediatric care settings16. Their review of 18 publications reported improvements in a range of process and surrogate markers. Three studies, two of which 55 57 have been already presented in the review by Clamp and 30 Keen , reported that the incidence of serious MEs/ADEs were significantly reduced by the use of CPOE, but the same effect was not identified in pediatric ICU study.58 Shekelle et al.46 conducted a SR study on the evaluation of costs and benefits of health IT mainly in US outpatient and inpatient paediatric settings, and in so doing identified 30 studies in relation to CPOE. The review showed consistent evidence that CPOE with CDSS has significant potential to reduce harmful MEs, particularly in inpatient paediatric and neonatal intensive care settings. Mullett et al.59 found that ePrescribing with CDSSs decreased pharmacist interventions for erroneous drug doses by 59%. CPOE (which is not combined with CDSS) was found to be effective in reducing medication dosing errors. Potts et al.56 conducted a prospective cohort study to examine medication prescribing errors and potential ADEs before and after implementation of a home-grown CPOE system in a paediatric intensive care unit. They found that the use of home-grown CPOE significantly reduced both MEs (30.1 to 90.2 %, p<0.001) and potential ADEs (2.2 to 1.3%, p<0.001).56 There are other SR studies which address the effect of COPE on serious in medication errors regarding safety, but apart from the above four studies, most of the studies failed to provide the evidence on ADEs or a few can provide the effect with low statistical power. For example, in a variety of settings with different types of patient populations, Shamliyan et al.45 systematically studied 12 studies in in-patient settings ranging from adult primary care, acute care to paediatric/ newborn intensive care unit settings. The review shows that prescribing errors amongst the majority of the studies 278 (8/10 studies) while they also showed a significant reduction in doing errors (3/7 studies) and in ADEs (3/8 studies), compared with handwritten orders. It also reported that one RCT and 5 uncontrolled interventions and four observational studies demonstrated that the implementation of CPOE is associated with the reduction of medication errors in both adults and paediatric patients without providing quantitative estimation of relative risk. Three studies also suggested that the implementation of CPOE systems had a positive impact on reducing ADEs, but without providing clear statistical evidence in support of this conclusion. For example, one study showed that the use of “CPOE would prevent 9 ADEs per 1,000 prescriptions in paediatric”56 populations while another study showed that “12 ADEs per 1,000 prescriptions in adult population” could be prevented.60 On the reduction in prescribing errors, the evidence shows that CPOE was associated with a 66% reduction in adults with odds ratio [OR] =0.34; 95%CI 0.22-0.52 and a positive tendency in children (P for interaction =0.028).45 From their wider review of 30 CPOE studies from 32 reports conducted in outpatient settings, Eslami et al. identified four studies evaluating the effect of ePrescribing on drug safety;35 all four systems had in-built CDSS functionality, but none of these four studies demonstrated any significant improvement in ADEs; the potential reasons for this may have varied across studies including non-use of systems and a small number of events and associated low power. Eslami et al. concluded that ‘…in spite of the cited merits of enhancing safety and reducing costs, published evaluation studies do not provide adequate evidence that ePrescribing applications provide these benefits in outpatient settings’. 35 Hider systematically reviewed the effectiveness of ePrescribing to improve practitioner performance and patient outcomes by covering 52 studies in a range of settings, including primary care, intensive care, inpatient and outpatient settings and most studies reported that CDSS could improve practitioner performance, especially for the prescribing of potentially toxic drugs and that alerts on prescription can reduce ADEs and improve other patient health outcomes including the risk of renal impairment.39 The inability 279 of some of the studies to find any improvement in health outcomes may be due to the small sample sizes, but the results from a meta-analysis of the studies evaluating electronic dose adjustment found that CDSS can reduce the frequency of adverse reactions and decrease the length of hospital stays. Schedlbauer et al. conducted a SR in which they identified 20 studies evaluate the impact of alerts on prescribing behaviour.44 The authors were found that the majority of alerts resulted in a reduction of MEs, but only a minority of studies reported on clinical outcomes. Apart from the reduction of ADEs and MEs, some SR papers reported that ePrescribing has shortened the length of patients’ hospital stay through more appropriate and effective prescription. For example, the SR study conducted by Durieux et al.31 122 assessed the beneficial effects of CPOE with CDSS on the process or outcome of health care with the focus on drug dosage in inpatient and outpatient settings by covering 26 comparisons in 23 articles of majority were RCTs (23 RCTs, 1 CT and 2CCT).31 122 Computerised advice on drug dosage had the effect to increase the initial dose of drug and serum drug concentrations, and this led to a more rapid therapeutic control, the reduction of the risk of toxic drug levels, as a result, shortening the length of patients’ hospital stay. The length of time spent in hospital was reported in six papers, finding a significant reduction in hospital stay overall in the computer group (SMD-0.35, 95%CI 0.52, 0.17).31 122 In one study a reduction in length of stay was found (SMD -0.04, 95%CI -0.07, -0.01), but Durieux et al. query the reliability of the confidence intervals due to a potential unit of analysis error.31 122 Eslami et al. conducted a SR study addressing the characteristics of CDSSs for tight glycemic control (TGC) and reviewed their effects on the quality of the TGC process in critically ill patients.32 Most of the CDSSs included in the studies were stand-alone. All of the controlled studies in Eslami et al.’s review reported on at least one quality indicator of the blood, but only one study reported a reduction in the number of hypoglycaemia events.32 280 Mollon et al. also conducted a SR study of 41 papers to evaluate the features of ePrescription with CDSS for successful implementation, prescribers’ behaviours and changes in patient outcomes.13 The authors identified five studies from 37 (12.2%) “successfully implemented” trials showing improvement in patient outcomes. It is interesting to note that all the five studies showing patient outcome improvements were published after 2005, implying that systems are becoming more effective in mitigating patient outcome risk. Yourman et al. assessed systematically the improvement of medication prescription in older adults, with the focus on CDSS intervention in outpatient and inpatient settings, mainly in the US.52 A majority of the studies were of systems providing direct support at the point of care and were not condition or disease-specific. Of 10 studies testing CDSS interventions for older adults, eight showed at least modest improvements (median number needed to treat, 33) in prescribing, as measured by minimising drugs to avoid, optimising drug dosage, or improving prescribing choices in older adults (according to each study's intervention protocols). The majority of studies reported medicationrelated process outcomes, for which CDSS generally showed positive effects, including lower rates of prescribing inappropriate drugs and closer adherence to better drug choices or dosages for older persons. The studies reviewed indicate that often straightforward point of care recommendations showed modestly effective results from a process outcomes perspective. Improved practitioner performance in prescription Reviews assessing the impact of ePrescribing on the quality of care include studies focusing on the ordering of prophylactic prescriptions, adherence to prescribing guidelines and organisational efficiency. The definitions and measurement of quality outcomes vary, so generalising across organisational settings is difficult. Garg et al. included 29 trials of drug dosing and prescribing, with single-drug dosing improving practitioner performance in 15 (62%) of 24 studies; another five applications used electronic order entry for multi-drug prescribing with four 281 of these applications improved practitioner performance.36 Nies et al.61 however, assessed the same studies included in the review by Garg et al.36, but came to a different conclusion, namely that ‘…drug dosage adjustment was less frequently observed in positive studies (29 per cent) than in negative studies (71 per cent).’61 Whilst Nies et al. noted that their conclusions differed to those made by Garg et al. they did interpret why this contradictory finding was made.61 This discrepancy may have resulted from differences in the definition of success, but merits further exploration. Time efficiency and improved work-flows One of the important themes for organisational implications of the implementation of ePrescribing systems highlighted in the literature is time efficiency and improved working practices. Tan, et al. systematically examined whether the use of CDSS has an effect on newborn infants’ mortality and morbidity and on healthcare practitioners performance by assessing three RCTs.49 One study,62 which investigated the effects of a database programme in aiding the calculation of neonatal drug dosages found that the length of time for the calculation was significantly reduced among resident paediatric staff, paediatricians and, to a lesser extent, for nurses by the use of the Neodosis spreadsheet program, while the system eliminated serious errors. However, there were insufficient data from the randomised trials to determine the patient benefit or harm from CDSS in neonatal care. Niazkhani systematically reviewed 51 papers covering 45 studies to evaluate the impact of CPOE on organisational efficiency and, in particular, clinical workflow.42 Most of the CPOEs studied were commercial and the majority were in adult inpatient settings in teaching hospitals but also included pediatric settings. The evidence shows that the implementation of CPOE resolved many disadvantages associated with the work-flow in paper-based processes. For example, 11 studies showed that CPOE systems improved work-flow efficiency by removing many intermediate and time-consuming tasks for healthcare professionals, while six before-after studies showed a 282 substantial decrease in the drug turnaround time, varying from 23-92%. Furthermore, three studies found a significant reduction of 24-69% in the time interval between clinicians’ radiology requests and the completion of the procedures pre- and post-implementation. The same three studies also showed that a shorter turnaround time was found for laboratory orders, varying from 21-50%. Clamp and Keen also noted that although there was no evidence of reduction in pharmacists’ time spent dealing with prescriptions, there were changes in their working patterns.30 The authors argued that pharmacists have an important quality control role in checking prescriptions, with one study finding that pharmacists only spent 5-20% of their time on direct clinical care.63 Prescription monitoring and adaptation was reduced to less than 10% in a UK hospital using ePrescribing, allowing pharmacists to spend around 70 % of their time on direct patient care.64 In a US study the pharmacists spent 46% more time on problem-solving activities and 34% less time filling in prescriptions.65 The authors noted that three studies—including one RCT— found that the total time for direct and indirect patient care increased due to the introduction of the ePrescribing system and there was a reduction in pharmacist interventions for prescriptions.66-68 Evidence for improved organisational efficiency was also found by Clamp and Keen in their review of turn-around times.30 Mekhijan et al. found a statistically significant reduction in turn-around times following the implementation of ePrescribing (64% reduction; P<0.001).69 Turn-around time from ordering to dispensing was found to decrease by up to 2.5 hours in a study by Lehman et al.70 Sintchenko et al. conducted an SR study of 24 papers (RCTs) to assess the importance of the type of clinical decisions and decision-support systems and the severity of patient presentation on the effectiveness of CDSS use in US and Europe.48 The study reported that control trials of CDSS indicated greater effectiveness in hospital settings than when applied to chronic care and CDSS improved prescribing practice and outcomes for patients with acute conditions, although CDSS were effective in changing doctors’ performance or outcomes in primary care. 283 Guideline compliance Another recurring theme in the literature is guideline compliance. A number of SRs have demonstrated the improved level of adherence to guidelines.16 33 34 37 40 47 This body of evidence suggests that achieving guideline compliance would increase cost effectiveness by reducing unnecessary prescriptions and laboratory tests. For example, Eslami et al.33 34 examined adherence to guidelines in outpatient and inpatient settings as part of their systematic reviews of ePrescribing. For outpatient settings, the authors concluded that there is evidence of the ability of ePrescribing applications to increase healthcare professionals adherence to guidelines and hypothesised that cost reduction can be achieved when guidelines are specifically geared towards this goal.33 The authors based their conclusions on 11 studies evaluating the impact of ePrescribing with a CDSS on the adherence to a guideline or another standard. Among these, four studies showed that there was a significant positive effect on adherence;71-74 two studies showed a positive effect without reporting on statistical significance;75 76 and five studies did not find a significant difference between the control and the intervention groups. Rothschild16 systematically evaluated the effects of CPOE on clinical and surrogate outcomes in hospitalised patients in both general and critical care settings, covering 18 papers. The review found that several process outcomes improved with CPOE, including increased compliance with evidence-based practices, reductions in unnecessary laboratory tests and cost savings in pharmaco-therapeutics. Guideline compliance (corollary orders) increased from 21.9% to 46.3% (P=0.01), but there was no effect on length of stay.77 10.4.2 Empirically demonstrated risks A main limitation to studies reporting negative consequences associated with the use of ePrescribing is that they tend to not indicate which of the many possible mechanisms might have resulted in the adverse effects. 284 Impact on patients Eslami et al. noted that recent studies suggested that errors, ADEs and even mortality may have increased after CPOE implementation.34 Van Rosse et al.50 also address the increase in mortality rates associated with the introduction of a CPOE system reported by Han et al..80 This study has been discussed extensively in the literature.81-83 Han et al. describe the most serious of risks to patient safety, mortality.80 The authors found that the unadjusted mortality rate increased from three per cent before ePrescribing implementation to seven per cent after ePrescribing implementation (P<0.001). Observed mortality was consistently better than predicted mortality before ePrescribing implementation, but this association did not remain after ePrescribing implementation.80 The Han et al. study demonstrated that increased mortality can be associated directly with modifications in standard clinical processes: With the ePrescribing system, order entry was postponed until after the processing of patient admission.84 Although accurate patient registration is important to patient safety, delaying care and treatment of severely ill patients due to the new work practices embedded in computer systems may adversely affect patient outcomes.85 However, van Rosse et al.50 also refer to the study conducted by Del Beccaro et al. 81 which evaluated the same CPOE system as Han et al.,80 but did not find a significant change in mortality rates. Only three hours of training were conducted during the three months before the implementation day. Ammenwerth et al. compared these studies and noted that there were important differences in design and implementation strategies.83 Han et al.80 studied CPOE use with a more critically ill and younger patient population than Del Beccaro et al.81 Furthermore, Han et al. studied outcomes only five months after CPOE implementation,80 whereas Del Beccaro et al. extended their post-implementation study period to 13 months.81 The longer study period of Del Beccaro et al. may have averaged out a potentially higher error rate in the first few months after CPOE implementation due to a learning curve effect.81 Keene et al.82 also studied the effect of CPOE introduction in a critically ill paediatric population with comparable results to those of Del 285 Beccaro et al.81 The study conducted by Keene et al. suggest that most of the possible factors which led to the increase of mortality after the implementation cannot be attributed to the CPOE system itself, but rather resulted from the implementation process.82 Furthermore, Rosenbloom et al. noted that the implementation process for the application described by Han et al. did not incorporate steps or elements known to ensure system dependability and usability.84 Bradley et al. has also noted that total error reports increased postimplementation of ePrescribing, but found that the degree of patient harm related to these errors decreased.86 Furthermore, Shulman et al.60 noted that the proportion of drug errors fell significantly from seven per cent before ePrescribing introduction to five per cent thereafter (P<0.05), but that this occurred against the backdrop of a strong declining linear trend of the proportion of drug errors over time (P<0.001).60 These authors, however, reported three important errors intercepted by ePrescribing which could otherwise have resulted in permanent harm or death; these errors were identified and then acted upon by pharmacist or nurse intervention60, i.e.: ‘A potentially fatal intercepted error occurred when diamorphine was prescribed electronically using the pull down menus at a dose of seven mg/kg instead of seven mg, which could have lead to a 70-fold overdose. In a separate case, amphotericin 180 mg once daily was prescribed, when liposomal amphotericin was intended. The doses of these two products are not interchangeable and the high dose prescribed would have been nephrotoxic. In the third case, vancomycin was prescribed one g intravenously daily to a patient in renal failure, when the appropriate dose would have been to give one g and then to repeat when the plasma levels fell below 10 mg/L. The dose as prescribed would have lead to nephrotoxicity.’ 60 Koppel et al. conducted a study on drug errors introduced by ePrescribing. The authors ‘…identified 22 previously unexplored drug error sources that 286 users reported to be facilitated by ePrescribing through their assessment.’87 The sources were grouped as: (1) information errors generated by fragmentation of data and failure to integrate the hospital’s several computer and information systems; and (2) human-machine interface flaws reflecting machine rules that do not correspond to work organisation or usual behaviours.87 However, this study has been criticised due to the high risk of bias with respect to their key findings. In response to this study, Bates,88 for example, notes that: ‘A main limitation of Koppel et al.’s study was that it did not count errors or adverse events, but instead measured only perceptions of errors, which may or may not correlate with actual error rates. Furthermore, it did not count the errors that were prevented. As such, it offers no insight into whether the error rate was higher or lower with ePrescribing. Unfortunately, however, the press interpreted the study as suggesting that ePrescribing increases the drug error rate. While the authors did not state this, a press release put out by the journal that published the article did so.’ 88 Risks to patient safety may arise indirectly from application use. For instance, a survey of UK GPs found that some respondents erroneously believed that their computers would warn them about potential contraindications or if an abnormal dose or frequency had been prescribed, highlighting how lack of knowledge and training in how ePrescribing systems function can compromise patient safety.89 Risks to patient safety can arise not only from system use but also from a lack of actual usage undermining the ability of ePrescribing applications to confer the envisaged benefits to patient safety. A sub-section of the review by Eslami et al. looked at system usage, the authors noted that there was wide variability in the degree of ePrescribing usage.33 Four studies found that of all prescriptions, 3–90 per cent were entered electronically.66 90-92 287 The SR by Chaudry29 included one study which used a mixed quantitative– qualitative approach to investigate the possible role of such a system in facilitating PEs reported that 22 types of ME risks were found to be facilitated by ePrescribing, relating to two basic causes: fragmentation of data and flaws in user-system interface.87 Ammenwerth et al. looked at the relative risk reduction on ME and ADE by CPOE covering 27 studies on ePrescribing mainly in the US in patient care settings with study designs including before-after studies/time-series analysis and two RCTs.25 Twenty-three of these studies showed a significant relative risk reduction for medication errors of 13-99%, but it is also worth highlighting one exceptional study which looked at the implementation of a commercial ePrescribing system with advanced CDSS at two study units between 2002 and 2003 for three months, which reported a significant increase of 26% for the risk of medication errors.93 Negative impact on professionals’ performance and organisational efficiency It should be noted that negative impact of ePrescribing systems on healthcare professionals’ performance and organisational efficiency can result in risks to patient safety. However, such negative evidence requires careful consideration to identify whether these risks are intrinsic to ePrescribing systems or are a part of the socio- organisational learning processes in the implementation. Eslami et al.33 and Eslami et al.34 conducted SR studies to evaluate studies of CPOE with/without CDSS on several outcome measures in outpatient and inpatient settings respectively. In outpatient settings, the authors found three studies65 67 94 (one RCT and two non-RCTs) that reported an increase in the total time for direct and indirect patient care due to the implementation of the CPOE system, while three studies (one RCT and two non-RCTs) also reported an increase in ordering time with the introduction of CPOE.95-97 Two of the studies94 96 were also assessed by Shekelle et al. who evaluated 30 studies on CPOE as a part of their evaluation of the costs and benefits of 288 health information technologies in various healthcare settings.46 The authors noted that two studies reported an increase of the clinicians’ time for order entry using CPOE compared to paper methods, and both studies demonstrated that CPOE took up slightly more clinician time.46 Poissant et al. reviewed 23 papers on EHRs to evaluate time efficiency of physicians and nurses and identify factors that may explain efficiency differences across studies.43 The authors found that the use of central station desktops for CPOE was inefficient, increasing the work time from 98.1% to 328.6% of physician’s time per working shift (weighted average of CPOEoriented studies, 238.4%).43 Tierney et al. found that interns in the intervention group spent an average of 33 minutes longer (5.5 minutes per patient) during a 10-hour observation period writing orders than did interns in the control group (P<0.001).79 Another BWH study published by Bates et al. using work study techniques found that for both medical and surgical house officers, writing orders on the computer took about twice as long as using the manual method, these differences being both clinically and statistically significant (P<0.001).78 However, medical house officers recovered nearly half the lost time due efficiency improvements in other administrative tasks, for example looking for charts.78 Additionally, a pilot of ePrescribing standards in the US found that providers noted that ‘…everything interacts with everything’ making for an overwhelming amount of alerting and therefore additional work.79 Other than writing orders, one observational study by Almond in the UK found that the time to complete the ward drug administration rounds doubled for healthcare assistants.80 Niazkhani conducted a SR study of 45 studies in 51 papers to evaluate the impact of CPOE on organisational efficiency, in particular clinical workflow.42 The author noted that the implementation of CPOE led to the creation of difficulties in work-flow mainly due to changes in the structure of preimplementation work and negative evidence was reported on time efficiency. Five studies showed time inefficiency due to the implementation of CPOE and 289 four out of the five studies reported a significant increase in time. The proceduralisation of order entry and the structuring of relationships between actors were also found to be a source of time inefficiency. Difficulties in choreographing the various actors and a reduction in team-wide discussions were also found. The scope for potential clinician process efficiency gains from the introduction of CPOE will be dependent at least in part on the inefficiency and thoroughness of the previous paper-based processes, combining retrieval, viewing of information, data entry, and in many cases, responses to alerts and reminders. The work practices of nurses have been found to be proceduralised, but clinicians may follow idiosyncratic practices.98 10.4.3 Implications of technological taxonomy to benefits We identified some of the SR papers assessed technical functionalities based on the three different types of taxonomy: “commercial versus home-grown” systems,25 29 30 42 45 46 “basic versus advanced” CDSS for prescription25 28 30 3234 38 44 45 47 51 52 and “stand-alone versus integrated” systems.25 29 30 36 40 47 The evaluation from these perspectives is extremely important to obtain the insight into what kinds of elements make contribution to successful implementation of ePrescribing systems for the improvement of patient safety and organisational efficiency. However, Wier et al.106 point out how often authors provided little information about the application, technical infrastructure, implementation process or other descriptive data. Such missing information leads to difficulties of obtaining accurate picture of the implementation sites and generalisability of evidence from studies. Bearing these issues in mind, this section demonstrates key evidence of benefits of ePrescribing from the body of literature. Commercial versus home-grown systems Not all of the systematic reviews papers evaluated relevant studies from a point of technological taxonomy and the evidence thus tends to be presented without distinguishing between the findings from in-house built and commercial systems. Some of the papers do however provide some insights 290 into findings when viewed through this lens. There are some SRs reporting positive results with home-grown systems25 29 30 46 but there is limited evidence reported regarding the benefits of commercial systems with highquality of studies. However they demonstrate that customising commercial systems to tailor them to local hospital environments can also bring benefits25 33 43 45 51 . Clamp and Keen30 121 found that there was no overall evidence that use of ePrescribing systems reduces the rate of preventable ADE, but pointed out that one study showed that the internally developed systems, with CDDS of menu of medications, default doses, range of potential doses, limited drugallergy checking, drug-drug-interaction significantly reduced serious MEs by 55% and 55 drug-laboratory checking as well as an 86% decrease in dose, frequency, route, substitution and allergies.57 The study also found an overall decrease in preventable ADEs of 17%.55 Shekelle et al.46 also cite a prospective cohort study investigating impact of a “home-grown” CPOE on MEs and ADEs in a paediatric intensive care.56 This study found a significant reduction of both MPEs (30.1 to 90.2%, P<0.001) and PADEs (2.2 to 1.3%, P<0.001).56 Another study by Cordero et al. in neonatal intensive care found that a CPOE system could eliminate gentamicin prescribing errors.99 Also another study which implemented a home-grown CPOE system with advanced CDSS (including allergy alerts, dose checking, drug interaction, clinical pathways, patient and place specific dosage, interfaces with clinical data repository–order related and laboratory alerts) reported the potential reduction of ADE, that is, the prevention of one ADE every 64 days by the use of the ePrescribing system in a paediatric setting but no statistical evidence was provided to support this estimation.56 After 2007 some SRs started to report the effect of commercial systems for patient safety in parallel to the studies of home-grown systems.25 33 45 In a recent review, Eslami et al. employed the taxonomy of “basic support” versus “advanced support/alerts” to evaluate the impact of ePrescribing 291 systems on safety; cost and efficiency; adherence to guideline; alerts; time; and satisfaction, usage, and usability in the outpatient setting while it also addressed the “stand-alone” versus “integrated systems” and “commercial” versus “home-grown” dimensions of ePrescribing systems.33 However, the evidence drawn from the study in relation to each dimension is not clearly stated and is obfuscated in the analysis. However, it is worth mentioning one observational study100 which showed “important weaknesses in generating alerts in four commonly used commercial systems in Britain’s GP offices”. Those systems were unable to generate “all 18 predefined established alerts for contraindicated drugs and hazardous drug-drug combinations”.33 Another important literature on this taxonomy is the study by Ammenwearth et al. who conducted a systematic review of 27 papers on ePrescribing implementation mainly in the US outpatient, inpatient and intensive care settings.25 The study included only two RCTs–most of the other studies employed before-after and in some cases time-series designs. The ratio of studies looking at commercial systems vs. home-grown systems was approximately 1: 1, with one study adopting both designs. Their sub-group analysis of 25 studies comparing reductions in medication errors between home-grown and commercial systems highlighted a greater risk reduction in errors with the home-grown systems.25 In spite of these reported results, the quality of those studies was not high as many of them did not fully specify the experimental design, did not describe the cohort or state whether the comparison and intervention groups’ treatment was commensurate and only two studies were randomised trials. Setting Shamliyan et al. reviewed 12 studies published from 1990 to 2005 evaluating the impact of ePrescribing systems on prescribing errors in in-patient settings.45 One study56 which implemented in-house developed system in a 20-bed paediatric intensive care unit setting with prospective, intervention study found 95.9% of overall errors reduction (P<0.05), 99.4% of total prescribing errors reduction and 88.8% of wrong drug reduction (P=0.07) as absolute change in rate while the study also found 7.6% (P=0.69) of wrong 292 dose increase. Another study99 examined the implementation of a commercial system with a retrospective study in a post-natal intensive care unit setting and found that medication errors to prescribe gentamicin reduction and to prescribe the wrong dose of gentamicin were eliminated, but with no statistical significances provided. Overall, although definitive evidence from systematic reviews of RCTs comparing home grown and package systems is lacking, the data suggest that home grown systems are more effective than commercial systems in reducing prescribing errors. There is however as yet no clear data available on whether these differences translate into improvements in important patient outcomes such as death. ePrescribing systems with “basic” vs. “advanced” decision support Overall picture of the taxonomy of “basic” versus “advanced” CDSS is ePrescribing systems with advanced CDSS showed a higher relative risk reduction compared to those with limited or no decision support.44 47 In particular, the evidence in the literature reported that the “patient-specific” alerts improve the quality of prescribing.25 This taxonomy is closely linked to the other one, “stand-alone” versus “integrated” systems. The following paragraphs refer to the key literature with CDSS elements of “basic” versus “advanced” decision support. The SR study by Schedlbauer et al. provides important evidence in relation to the typology of DSS/alerting systems and reminding systems in relation to ePrescribing systems.44 The study focused on the effects of those alerts and reminders on prescribing behaviour mainly in the US inpatient settings covering the relevant papers published between 1994 and 2007. Twenty studies, which have employed randomised and quasi-experimental designs were included. Categories of drug alerts comprise basic drug alerts, advanced alerts and complex alert systems (representing a set of CDSSs containing features of both basic and advanced alerts). Two papers in their study investigated the effects of four types of basic alerts, of which three reported statistically significant beneficial effects on prescribing. Drug allergy 293 warnings decreased allergy error events by 56% (P=0.009).55 It also found that providing default dosing via basic medication order guidance alerts resulted in reduced dose errors in two studies of 23% (P=0.02)55 and 71% (P=0.0013).101 Regarding medication errors, the 40% reduction in error events achieved by drug-drug interaction warnings did not reach statistical significance (P=0.89).55 The SR study confirms that advanced alert types tend to provide positive effects across the five categories, saying that all the 20 papers evaluated more advanced alert types and statistically significant effects were shown in 21 out of 23 types across five categories. Shiffman et al. looked at the impact of CDSS on practitioner performance, patient outcomes and satisfaction, with the focus on functionality and the effectiveness of the systems.47 The authors studied 25 RCTs, CT and TS which were published between 1992 and 1998. The SR study was conducted using a technological taxonomy, dividing studies between stand-alone and integrated systems and between basic and advanced DSS. All of the systems displayed relevant patient data, a menu of drugs and a choice of doses. Half of the studies used a system with advanced CDSS functionality while the other used systems with no or limited/basic CDSS. Also, in 14 studies with advanced decision support, the risk reduction was greater than in 11 studies without advanced decision support, but studies without advanced support were mostly compared to computer-based ordering whereas those with advanced support were compared to manual procedures. Stand-alone versus integrated systems None of the 30 SRs included in this analysis directly address differences between “stand-alone” and more “integrated” systems. However, some of the papers do indirectly address this issue offering some insights. We discuss below the salient findings from these studies. A study of ADEs found that having ADE detection and reporting capability in EHR can improve detection of, and potentially reduce ADE because the EHR system data can be used to identify patients experiencing ADEs.102 A RCT to 294 explore the impact of an EHR with integrated ePrescribing found positive effects on resource utilisation, provider productivity, and care efficiency.46 95 Eslami et al.32 looked at the characteristics of CDSS for tight glycemic control (TGC) and the effects on the quality of the TGC process in critically ill patient by categorising CDSS into the three features: 1) level of support (merely displaying the protocol chart or suggesting the specific amount of insulin to be administered); 2) the consultation mode (passive or active); and 3) the communication style (in the critiquing mode or in the non-critiquing mode). Most of the CDSS (14 out of 17) were stand-alone and only two papers studied more integrated system.103 104 One of these studies103 reported a reduction in the number of hypoglycaemia events, but without assessing statistical significance. Interest in the effectiveness of ePrescribing systems continues.105 The number of systematic review papers on e-Prescribing/CPOE has thus been growing. However, recent reviews have been inconclusive and shown wide variations in findings. For example, Wier et al., who conducted a systematic review of the scientific quality of empirical research on CPOE application, found that there are areas requiring improvement in research designs and analyses.106 There is a tendency for empirical studies of CPOE to lack adequate study designs and blinding, although there are several high-quality CPOE studies available: “Current concerns center in the prominent use of pre-post study designs less rigorous measurement techniques, failure to include key information about CPOE and informatics variables, failure to use blinding and inappropriate statistical analyses. These concerns must be addressed to allow the field to build a solid foundation of study generalisability for this area of inquiry in the future.”106 More importantly, implementation strategies significantly varied and this can lead to confounded results. Apart from the issues of internal validity (e.g. design type, testing of group differences, instrumentation bias and blinding), 295 construct validity including types of ordering functions, level of decision support available, electronic links to other departments, and whether usage is mandated, or measured, implementation strategies used and length of time from implementation to measurement of outcomes and statistical validity are important, but Wier et al. point out how often authors provided little information about the application, technical infrastructure, implementation process or other descriptive data.106 Such missing information leads to difficulties of generalisability of evidence from studies. Bearing these issues in mind, this section demonstrates key evidence of benefits and risks of ePrescribing, which are relevant to healthcare quality, patient safety and organisational issues, obtained from the relevant literature. 10.5 Implications for policy, practice and research 10.5.1 System integration Reviewing this body of work has revealed that ePrescribing systems have heterogeneous origins, scope and functionality and are furthermore implemented into diverse organisational settings, all of which may, along with the context of and approach to implementation, influence the risks and benefits that result. As the functionality of these systems extend, these are more appropriately seen as expert systems rather than data processing tools, so the integration or “fit” with organisational knowledge is increasingly seen as important (see Chapter 17). The integration of ePrescribing systems with EHRs is a technical bridge to allow patient-specific information to be used in the delivery of patient-centred care and to minimise the risks of MEs and ADEs as well as improving prescription efficiency. 10.5.2 Knowledge database sharing In parallel to the integration into EHRs, the integration of CPOE with CDSS is a logical development, which should be encouraged. The rule bases for decision support content can either be locally developed or created centrally, with clear implications for clinician autonomy. However, the open sharing and consolidation of complex rules on drug and diagnosis interactions across healthcare communities offers the possibility of further benefits and 296 efficiencies of scale from ePrescribing that may not have been identified in the more focused studies.107 Clinicians currently benefit from the development of guidelines on specific conditions, such as the British Thoracic Society’s asthma guidelines108 109 which can be embedded within CPOE. Guidelines gain legitimacy through the reputation of their sponsoring body, including the National Institute for Health and Clinical Excellence (NICE) and the UK’s National Service Framework (NSF).109 Implemented CPOE systems need to be able to be updated to keep their knowledge bases up-to-date with this evolving body of knowledge and ideally be able to provide information on the rules being applied to ensure clinician compliance. Aronson emphasises that it is not guidelines alone that influence prescribing behaviour, but also the education and financial incentives to ensure guideline compliance.109 10.5.3 System standards Interoperability with other healthcare information technology systems is a key factor for successful implementation of ePrescribing systems, as the systems are ideally drawing on patient data, updating patient records and integrating with consumable inventory systems. In practice, interoperability is achieved through standards, whether de facto local specifications, proprietary standards of system vendors or conformance with nationally agreed standards. At the heart of standardisation in ePrescribing, as in most areas of health informatics, is the EHR, as there is no point in building decision functionality into a CPOE system that depends upon patient data that are not accessible. As the market for advanced ePrescribing systems develops, the functionality offered by vendors will be shaped by the data on standardised EHRs, pushing vendors to offer functionally similar systems. There is therefore a need to ensure that EHR standards are extensible to include future patient data needs to prevent functional lock-in for ePrescribing systems. 297 10.5.4 Implementation of ePrescribing systems from human factors CDSS interventions may include alerting and reminder systems. Employing advanced guidelines is not in itself sufficient to make sure prompts are acted on, as alerts may be overridden or ignored.110 Consideration of human factors becomes crucial on this point (see Chapter 16). Human factors can be categorised into the following four categories: 1) physical and perceptual factors; 2) cognitive factors; 3) motivational factors; and 4) situational factors. The reasons repeatedly found for overriding alerts included: alert fatigue, disagreement; poor presentation; lack of time; knowledge gap.111 112 Procter et al. have argued that human factor considerations are the key to the achievement of effective and safe implementation of healthcare systems and that healthcare professionals’ involvement is crucial in system design and development.112 A “user-centred perspective can inform system design to ensure that individual technologies achieve their intended purpose and benefits”113 known as the human-tech approach.114 In order to achieve robust patient safety, the micro (user interfaces, ergonomics) the meso (inter-system communication and integration) and the macro (organisational design) perspectives all need to be addressed during system design.113 Taking these factors into account can increase clinicians’ trust and lead to greater system acceptance. By recognising that ePrescribing systems are fundamentally socio-technical systems and investing in addressing the human factors during design and implementation there will be longer-term gains in lower lifecycle staffing and training costs, reduced risk of errors and greater rule compliance.115 10.5.5 Database for CDSS and data standardisation As discussed above, to gain the greatest operational gains from ePrescribing requires the systems to draw on knowledge bases of complex rules which can be applied to specific patients. To develop parochial rule bases or rule bases developed by each system vendor is potentially inefficient. Part of the rule base, for example drug interactions, will be locality-independent, and could be developed globally. Other aspects, such as rules on recommended treatments, may be institution specific and would then need local development 298 and ownership. There is therefore a need for processes to maintain the rulebase, carry out assessments of evidence and provide rule legitimacy. Similarly, there is a policy need for the specification of EHRs to take account of the needs of current and future ePrescribing systems, which will require coordination with the emergent rule-base. 10.5.6 Temporal issues regarding the evaluation of ePrescribing implementation This overview of the evidence provides very useful insights into the implementation of ePrescribing systems in a variety of contexts and settings. However, systematic reviews can also obfuscate in some respects. This is seen most clearly in the difficulty of addressing the dynamics of the emergence of a new technology. Almost without exception the reviews are atemporal, giving all papers within the review period equal weight. However, Mollon et al. noted that all of the studies they identified showing positive impacts on patients were from after 2005.13 This suggests, unsurprisingly, that there may have been a change in the technology through time. It can tentatively be proposed that there is evidence in the reviews that this is due to the benefits increasing as the systems take on more advanced decision support functions and become more integrated with other record systems. However, it may also partly be that as experience of these systems increases there is an underlying process of social learning about how these systems can be implemented and used effectively. Similarly, the studies generally ignore the short-term dynamics of system implementation, taking the implicit assumption that the changes observed shortly after implementation represent the steady-state system performance. In the summarising of most cases it is unclear how the time period of system observation related to the period of implementation, but the pressure to carry out a controlled before-and-after assessment implies that the evidence in most studies is from a period shortly following implementation. The potential danger of drawing conclusions about the long-term impacts based on these snap-shot evaluations is clearest in comparing the coverage in the reviews of Ammenwerth et al.25 and van Rosse et al.50 of the studies in Han et al. 80 and 299 Del Beccarro et al. 81 which studied the same ePrescribing implementation. It is reported that where Han et al. in a shorter study found an increase in mortality, Del Becarro et al. in their longer study did not find the same effect.81 One interpretation of this is that the mortality increase may have been a transient artefact of the organisational change process and that organisational learning and adaptation removed the effect. This is an important insight as it questions the interpretation of the effects identified as significant in many of the papers covered by the systematic reviews in this synthesis. 10.5.7 Synthesised research methods for the studies of eHealth It is important to address the limitations of a “systematic review” approach for the presentation of solid evidence. While the systematic reviews considered provide valuable insights into the impacts of ePrescribing systems, there is danger that the importance of factors influencing the impacts of ePrescribing systems will be underestimated. There is a trend in the area of health care study that synthesises qualitative and quantitative health evidence.116 117 Greenhalgh’s paper on meta-narrative approach towards systematic literature review, ‘Tensions and paradoxes in electronic patient record research: a systematic literature review using the meta-narrative method’ questions the meaning of ‘rigorous’ research engaging with philosophical debates.118 Lilford’s paper, ‘Evaluating eHealth: How to make evaluation more methodologically robust’ argues that a mixed research methods approach to evaluating IT systems in health care is needed, questioning the validity of evidence obtained by combining formative assessments with summative ones.119 The central distinction here is between treating the ePrescribing system as a work-in-progress where it is being recursively shaped by the studies or whether it is treated as a stable “black-box” with the focus of the study being to assess its impacts. Possible areas which are relatively neglected by the use of “systematic review” and “critical appraisal” methods are: • Communication amongst different groups of practitioners (e.g. clinicians-pharmacists; clinicians-nurses; pharmacists-nurses, patientspharmacists, or multi-groups of people of the above) 300 • Capture of complex changes in work-organisation /workflows • The impact of institutional differences between national healthcare systems on the shaping of ePrescribing systems and on the outcomes of ePrescribing trials. Systematic review tries to treat eHealth technologies as scientific objects not as social artefacts which are complex and organic in nature and can potentially lead to unexpected outcomes (see “blackboxing” arguments in the studies of science and technologies120). The systematic reviews examined implicitly assume that the results of small-scale trials can be scaled up, despite evidence that scaling IT systems leads to increasing problems of accommodating wider practice diversity and less identification of users with the systems. The use of short-term studies of recent implementations may overlook the impacts, both positive and negative, on longer term organisational learning. Finally, while the aggregation of short-term studies provides evidence on the impact of the systems on operational risks, it is harder to assess their impact on the risks of rare, but major systemic failure. 10.5.8 Areas for further research For future research, more sufficiently powered RCTs are needed.16 31 49 122 Such trials are however difficult to mount in this field, and the alternative of time-series based designs, preferably with contemporaneous control groups, should also therefore be considered. 106 Furthermore, research on functionality specific effects or technical specification effects31 33 34 46 are urgently needed, in particular evaluating the implementation of commercial systems.46 51 In order to make the evidence drawn from such studies, standardised reporting for healthcare IT evaluations is essential.25 32 33 40 46 The evaluation of the risk of MEs and ADEs will also be more reliable with the use of standardised metrics and reporting.44 45 50 51 Studying healthcare technology is a complicated task and the evaluation of ePrescribing systems is not an exception. In order to capture a more holistic 301 picture, multi-disciplinary methods are indispensable.25 33 34 42 There are a number of studies adopting before-after and time-series designs, but more evaluation of long-term effects is required. Also, evaluations immediately after implementation need to pay more attention to organisational learning processes with the focus on learning curve.50 This allows healthcare professionals to foresee the generalisability of the obtained evidence in their particular organisational settings. Evaluations in long-term care setting will also useful to assess the long-term impacts.51 In employing a more multi-disciplinary approach to the evaluation of ePrescribing systems, the study of how human factors and socio-technical issues influence the degree of implementation success becomes central.33 34 39 43 44 Also, the analysis of macro effects on collaborative work-flow and organisational efficiency is important.42 43 Finally, comprehensive economic evaluation of immediate and long-term effects is also urgently needed.37-40 References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. Bates DW, Leape LL, Petrycki S. Incidence and preventability of adverse drug events in hospitalized adults. Journal of General Internal Medicine 1993;8(6):289-94. Lesar TS, Briceland LL, Delcoure K, Parmalee JC, Mastagornic V, Pohl H. Medication prescribing errors in a teaching hospital. Jama-Journal of the American Medical Association 1990;263(17):2329-34. Pirmohamed M, James S, Meakin S, Green C, Scott AK, Walley TJ, et al. Adverse drug reactions as cause of admission to hospital: prospective analysis of 18,820 patients. British Medical Journal 2004;329(7456):15-19. Davies EC, Green CF, Taylor S, Williamson PR, Mottram DR, Pirmohamed M. Adverse Drug Reactions in Hospital In-Patients: A Prospective Analysis of 3695 Patient-Episodes. Plos One 2009;4(2). Howard RL, Avery AJ, Slavenburg S, Royal S, Pipe G, Lucassen P, et al. Which drugs cause preventable admissions to hospital? A systematic review. British Journal of Clinical Pharmacology 2007;63(2). Cresswell KM, Fernando B, McKinstry B, Sheikh A. Adverse drug events in the elderly. British Medical Bulletin 2007;83(1):259-74. Dobrev A, Stroetmann, K.A., Stroetmann, V.N., Artmann, J.N., Jones, T., and Hammerschmidt, R. The conceptual framework of interoperable electronic health record and ePrescribing systems, 2008. NHS CfH. Patient safety through e-prescribing: E-Health Insider and British Computer Society Health Informatics Forum, 2008. Teich JM, Marchibroda JM et. al., Electronic prescribing: toward maximum value and rapid adoption. Washington: eHealth Initiative, 2004 (170 pp). Avery AJ, Sheikh A, Hurwitz B, Smeaton L, Chen YF, Howard R, et al. Safer medicines management in primary care. British Journal of General Practice 2002;52:S17-S22. 302 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. Bell DS, Cretin S, Marken RS, Landman AB. A conceptual framework for evaluating outpatient electronic prescribing systems based on their functional capabilities. Journal of the American Medical Informatics Association 2004;11(1):60-70. Cornford T, Fraklin BD, Svage I, Barber N, Jani Y, Electronic Prescribing in Hospitals: Challenges and Lessons Learning, NHS Connecting for Health (NHS CFH) 2009;71. Mollon B, Chong JJR, Holbrook AM, Sung M, Thabane L, Foster G. Features predicting the success of computerized decision support for prescribing: a systematic review of randomized controlled trials. Bmc Medical Informatics and Decision Making 2009;9. Dobrev A, Jones, T., Stroetmann, V., Stroetmann, K., Vatter, Y., Peng, K. Interoperable eHealth is Worth it: Securing Benefits from Electronic Health Records and ePrescribing, 2010. Pollock N, Williams R, D'Adderio L. Global software and its provenance: Generification work in the production of organizational software packages. Social Studies of Science 2007;37(2):254-80. Rothschild J. Computerized physician order entry in the critical care and general inpatient setting: A narrative review. Journal of Critical Care 2004;19(4):271-78. McDonald CJ, Overhage JM, Mamlin BW, Dexter PD, Tierney WM. Physicians, information technology, and health care systems: A journey, not a destination. Journal of the American Medical Informatics Association 2004;11(2):121-24. Bates DW, Cullen DJ, Laird N, Petersen LA, Small SD, Servi D, et al. Incidence of adverse drug events and potential adverse drug events implications for prevention. Jama-Journal of the American Medical Association 1995;274(1):29-34. NHS CfH. ePrescribing: Overview and Benefits. NHS NPSA. IT requirement Specification Safety Alert: Oral Methotrexate 2.5mg and 10mg Tablets. 2006. Bates DW, Evans RS, Murff H, Stetson PD, Pizziferri L, Hripcsak G. Detecting adverse events using information technology. Journal of the American Medical Informatics Association 2003;10(2):115-28. Wolinsky H. Resolution due in medical software regulation. Annals of Internal Medicine 1997;127(10):953-54. Caudill-Slosberg M, Weeks WB. Case study: Identifying potential problems at the human/technical interface in complex clinical systems. American Journal of Medical Quality 2005;20(6):353-57. Campbell EM, Sittig DF, Ash JS, Guappone KP, Dykstra RH. Types of unintended consequences related to computerized provider order entry. Journal of the American Medical Informatics Association 2006;13(5):547-56. Ammenwerth E, Schnell-Inderst P, Machan C, Siebert U. The effect of electronic prescribing on medication errors and adverse drug events: A systematic review. Journal of the American Medical Informatics Association 2008;15(5):585-600. Ammenwerth E, Schnell-Inderst P, Siebert U. Vision and challenges of Evidence-Based Health Informatics: A case study of a CPOE meta-analysis. International Journal of Medical Informatics 2010;79(4):E83-E88. Bassi J, Lau F, Bardal S. Use of Information Technology in Medication Reconciliation: A Scoping Review. Annals of Pharmacotherapy 2010;44(5):88597. Chatellier G, Colombet I, Degoulet P. An overview of the effect of computerassisted management of anticoagulant therapy on the quality of anticoagulation. International Journal of Medical Informatics 1998;49(3):311-20. 303 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. Chaudhry B, Wang J, Wu SY, Maglione M, Mojica W, Roth E, et al. Systematic review: Impact of health information technology on quality, efficiency, and costs of medical care. Annals of Internal Medicine 2006;144(10):742-52. Clamp S, Keen J. The Value of Electronic Health Records: A Literature Review, 2005. Durieux P, Trinquart L, Colombet I, Nies J, Walton R, Rajeswaran A, et al. Computerized advice on drug dosage to improve prescribing practice. Cochrane Database Syst Rev 2008(3):CD002894. Eslami S, Abu-Hanna A, de Jonge E, de Keizer NF. Tight glycemic control and computerized decision-support systems: a systematic review. Intensive Care Medicine 2009;35(9):1505-17. Eslami S, Abu-Hanna A, De Keizer NF. Evaluation of outpatient computerized physician medication order entry systems: A systematic review. Journal of the American Medical Informatics Association 2007;14(4):400-06. Eslami S, de Keizer NF, Abu-Hanna A. The impact of computerized physician medication order entry in hospitalized patients - A systematic review. International Journal of Medical Informatics 2008;77(6):365-76. Fitzmaurice DA, Hobbs FDR, Delaney BC, Wilson S, McManus R. Review of computerized decision support systems for oral anticoagulation management. British Journal of Haematology 1998;102(4):907-09. Garg AX, Adhikari NKJ, McDonald H, Rosas-Arellano MP, Devereaux PJ, Beyene J, et al. Effects of computerized clinical decision support systems on practitioner performance and patient outcomes - A systematic review. JamaJournal of the American Medical Association 2005;293(10):1223-38. Georgiou A, Williamson M, Westbrook JI, Ray S. The impact of computerised physician order entry systems on pathology services: A systematic review. International Journal of Medical Informatics 2007;76(7):514-29. Hayward GL, Parnes AJ, Simon SR. Using health information technology to improve drug monitoring: a systematic review. Pharmacoepidemiology and Drug Safety 2009;18(12):1232-37. Hider P. Electronic Prescribing: A Critical Appraisal of the Literature, 2002. Jamal A, McKenzie K, Clark M. The impact of health information technology on the quality of medical and health care: a systematic review. Health Information Management Journal 2009;38(3):26-37. Khajouei R, Jaspers MWM. The impact of CPOE medication systems' design aspects on usability, workflow and medication orders: a systematic review. Methods of Information in Medicine 2010;49(1):3-19. Niazkhani Z, Pirnejad H, Berg M, Aarts J. The impact of computerized provider order entry systems on inpatient clinical workflow: a literature review. Journal of the American Medical Informmatics Association 2009;16(4):539-49. Poissant L, Pereira J, Tamblyn R, Kawasumi Y. The impact of electronic health records on time efficiency of physicians and nurses: A systematic review. Journal of the American Medical Informatics Association 2005;12(5):505-16. Schedlbauer A, Prasad V, Mulvaney C, Phansalkar S, Stanton W, Bates DW, et al. What evidence supports the use of computerized alerts and prompts to improve clinicians' prescribing behavior? Journal of the American Medical Informatics Association 2009;16(4):531-8. Shamliyan TA, Duval S, Du J, Kane RL. Just what the doctor ordered. Review of the evidence of the impact of computerized physician order entry system on medication errors. Health Services Research 2008;43(1):32-53. Shekelle PG, Morton SC, Keeler EB. Costs and benefits of health information technology. Evidence Report/ Technology Assessment (Full Report) 2006(132):1-71. 304 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. Shiffman RN, Liaw Y, Brandt CA, Corb GJ. Computer-based guideline implementation systems: A systematic review of functionality and effectiveness. Journal of the American Medical Informatics Association 1999;6(2):104-14. Sintchenko V, Magrabi F, Tipper S. Are we measuring the right end-points? Variables that affect the impact of computerised decision support on patient outcomes: A systematic review. Medical Informatics and the Internet in Medicine 2007;32:225-40. Tan K, Dear PRF, Newell SJ. Clinical decision support systems for neonatal care. Cochrane Database of Systematic Reviews 2005(2):CD004211. van Rosse F, Maat B, Rademaker CMA, van Vught AJ, Egberts ACG, Bollen CW. The Effect of Computerized Physician Order Entry on Medication Prescription Errors and Clinical Outcome in Pediatric and Intensive Care: A Systematic Review. Pediatrics 2009;123(4):1184-90. Wolfstadt JI, Gurwitz JH, Field TS, Lee M, Kalkar S, Wu W, et al. The effect of computerized physician order entry with clinical decision support on the rates of adverse drug events: A systematic review. Journal of General Internal Medicine 2008;23(4):451-58. Yourman L, Concato J, Agostini JV. Use of computer decision support interventions to improve medication prescribing in older adults: A systematic review. American Journal of Geriatric Pharmacotherapy 2008;6(2):119-29. Classen DC, Metzger J. Improving medication safety: the measurement conundrum and where to start. International Journal for Quality in Health Care 2003;15:I41-I47. Nebeker JR, Hurdle JF, Hoffman JM, Roth B, Weir CR, Samore MH. Developing a taxonomy for research in adverse drug events: Potholes and signposts. Journal of the American Medical Informatics Association 2002;9(6):S80-S85. Bates DW, Leape LL, Cullen DJ, Laird N, Petersen LA, Teich JM, et al. Effect of computerized physician order entry and a team intervention on prevention of serious medication errors. Jama-Journal of the American Medical Association 1998;280(15):1311-16. Potts AL, Barr FE, Gregory DF, Wright L, Patel NR. Computerized physician order entry and medication errors in a pediatric critical care unit. Pediatrics 2004;113(1):59-63. Bates DW, Teich JM, Lee J, Seger D, Kuperman GJ, Ma'Luf N, et al. The impact of computerized physician order entry on medication error prevention. Journal of the American Medical Informatics Association 1999;6(4):313-21. Evans RS, Pestotnik SL, Classen DC, Clemmer TP, Weaver LK, Orme JF, et al. A computer-assisted management program for antibiotics and other antiinfective agents. New England Journal of Medicine 1998;338(4):232-38. Mullett CJ, Evans RS, Christenson JC, Dean JM. Development and impact of a computerized pediatric antiinfective decision support program. Pediatrics 2001;108(4):art. no.-e75. Shulman R, Singer M, Goldstone J, Bellingan G. Medication errors: a prospective cohort study of hand-written and computerised physician order entry in the intensive care unit. Critical Care 2005;9(5):R516-R21. Nies J, Colombet I, Degoulet P, Durieux P. Determinants of success for computerized clinical decision support systems integrated in CPOE systems: a systematic review. AMIA Annual Symposium Proceedings 2006:594-8. Balaguer Santamaria JA, Fernandez Ballart JD, Escribano Subias J. Usefulness of a software package to reduce medication errors in neonatal care. Anales Españoles de Pediatría 2001;55(6):541-5. The Audit Commission. A Spoonful of Sugar: Medicines Management in NHS Hospitals, 2001. 305 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. Abu-Zayed L, Farrar K, Mottram DR. Time spent on drug supply activities in United Kingdom hospitals. American Journal of Health-System Pharmacy 2000;57(21):2006-07. Murray MD, Loos B, Tu WZ, Eckert GJ, Zhou XH, Tierney WM. Effects of computer-based prescribing on pharmacist work patterns. Journal of the American Medical Informatics Association 1998;5(6):546-53. Ross SM, Papshev D, Murphy EL, Sternberg DJ, Taylor J, Barg R. Effects of electronic prescribing on formulary compliance and generic drug utilization in the ambulatory care setting: a retrospective analysis of administrative claims data. Journal of Managed Care Pharmacy 2005;11(5):410-5. Beer J, Dobish R, Chambers C. Physician order entry: a mixed blessing to pharmacy? Journal of Oncology Pharmacy Practice 2002;8:119-26. Wogen SE, Fulop, G., Heller, J. Electronic prescribing: improving the efficiency of the prescription process and promoting plan adherence. Drug Benefit Trends 2003;15:35-40. Mekhjian HS, Kumar RR, Kuehn L, Bentley TD, Teater P. Immediate benefits realized following implementation of physician order entry at an academic medical center. Journal of the American Medical Informatics Association 2002;9(5):529-39. Lehman ML, Brill JH, Skarulis PC, Keller D, Lee C. Physician order entry impact on drug turn-around times. Journal of the American Medical Informatics Association 2001:359-63. Bernstein SL, Whitaker D, Winograd J, Brennan JA. An electronic chart prompt to decrease proprietary antibiotic prescription to self-pay patients. Academic Emergency Medicine 2005;12(3):225-31. Siegel C, Alexander MJ, Dlugacz YD, Fischer S. Evaluation of a computerized drug review system - impact, attitudes, and interactions. Computers and Biomedical Research 1984;17(5):419-35. Walton RT, Gierl C, Yudkin P, Mistry H, Vessey MP, Fox J. Evaluation of computer support for prescribing (CAPSULE) using simulated cases. British Medical Journal 1997;315(7111):791-95. Christakis DA, Zimmerman FJ, Wright JA, Garrison MM, Rivara FP, Davis RL. A randomized controlled trial of point-of-care evidence to improve the antibiotic prescribing practices for otitis media in children. Pediatrics 2001;107(2):art. no.-e15. Rivkin S. Opportunities and challenges of electronic physician prescribing technology. Medical Interface 1997;10(8):77-8, 83. Chin HL, Wallace P. Embedding guidelines into direct physician order entry: Simple methods, powerful results. Journal of the American Medical Informatics Association 1999:221-25. Overhage JM, Tierney WM, Zhou XH, McDonald CJ. A randomized trial of ''corollary orders'' to prevent errors of omission. Journal of the American Medical Informatics Association 1997;4(5):364-75. Smith BJ, McNeely MDD. The influence of an expert system for test ordering and interpretation on laboratory investigations. Clinical Chemistry 1999;45(8):1168-75. Kuperman GJ, Teich JM, Tanasijevic MJ, Ma'Luf N, Rittenberg E, Jha A, et al. Improving response to critical laboratory results with automation: Results of a randomized controlled trial. Journal of the American Medical Informatics Association 1999;6(6):512-22. Han YY, Carcillo JA, Venkataraman ST, Clark RSB, Watson RS, Nguyen TC, et al. Unexpected increased mortality after implementation of a commercially sold computerized physician order entry system. Pediatrics 2005;116(6):1506-12. 306 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. Del Beccaro MA, Jeffries HE, Eisenberg MA. Computerized provider order entry implementation: No association with increased mortality rates in an intensive care unit. Pediatrics 2006;118(1):290-95. Keene A, Ashton L, Shure D, Napoleone D, Katyal C, Bellin E. Mortality before and after initiation of a computerized physician order entry system in a critically ill pediatric population. Pediatric Critical Care Medicine 2007;8(3):268-71. Ammenwerth E, Talmon J, Ash JS, Bates DW, Beuscart-Zephir MC, Duhamel A, et al. Impact of CPOE on mortality rates - Contradictory findings, important messages. Methods of Information in Medicine 2006;45(6):586-93. Rosenbloom ST, Harrell FE, Lehmann CU, Schneider JH, Spooner SA, Johnson KB. Perceived increase in mortality after process and policy changes implemented with computerized physician order entry. Pediatrics 2006;117(4):1452-55. Sittig DF, Ash JS, Zhang JJ, Osheroff JA, Shabot MM. Lessons from "Unexpected increased mortality after implementation of a commercially sold computerized physician order entry system". Pediatrics 2006;118(2):797-801. Bradley VM, Steltenkamp CL, Hite KB. Evaluation of reported medication errors before and after implementation of computerized practitioner order entry. Journal of Healthcare Information Management 2006;20(4):46-53. Koppel R, Metlay JP, Cohen A, Abaluck B, Localio AR, Kimmel SE, et al. Role of computerized physician order entry systems in facilitating medication errors. Jama-Journal of the American Medical Association 2005;293(10):1197-203. Bates DW. Computerized physician order entry and medication errors: Finding a balance. Journal of Biomedical Informatics 2005;38(4):259-61. Morris CJ, Savelyich BSP, Avery AJ, Cantrill JA, Sheikh A. Patient safety features of clinical computer systems: questionnaire survey of GP views. Quality & Safety in Health Care 2005;14(3):164-68. Rotman BL, Sullivan AN, McDonald TW, Brown BW, DeSmedt P, Goodnature D, et al. A randomized controlled trial of a computer-based physician workstation in an outpatient setting: Implementation barriers to outcome evaluation. Journal of the American Medical Informatics Association 1996;3(5):340-48. Schectman JM, Schorling JB, Nadkarni MM, Voss JD. Determinants of physician use of an ambulatory prescription expert system. International Journal of Medical Informatics 2005;74(9):711-17. Tamblyn R, Huang A, Kawasumi Y, Bartlett G, Grad R, Jacques A, et al. The development and evaluation of an integrated electronic prescribing and drug management system for primary care. Journal of the American Medical Informatics Association 2006;13(2):148-59. Spencer DC, Leininger A, Daniels R, Granko RP, Coeytaux RR. Effect of a computerized prescriber-order-entry system on reported medication errors. American Journal of Health-System Pharmacy 2005;62(4):416-19. Overhage JM, Perkins S, Tierney WM, McDonald CJ. Research paper Controlled trial of direct physician order entry: Effects on physicians time utilization in ambulatory primary care internal medicine practices. Journal of the American Medical Informatics Association 2001;8(4):361-71. Tierney WM, Miller ME, Overhage JM, McDonald CJ. Physician inpatient order writing on microcomputer workstations - effects on resource utilization. JamaJournal of the American Medical Association 1993;269(3):379-83. Bates DW, Boyle DL, Teich JM. Impact of computerized physician order entry on physician time. Proceedings of Annual Symposium on Computer Application in Medical Care 1994:996. Shu K, Boyle D, Spurr C, Horsky J, Heiman H, O'Connor P, et al. Comparison of time spent writing orders on paper with computerized physician order entry. 307 98. 99. 100. 101. 102. 103. 104. 105. 106. 107. 108. 109. 110. 111. 112. 113. Medinfo 2001: Proceedings of the 10th World Congress on Medical Informatics, Pts 1 and 2 2001;84:1207-11. van der Meijden MJ, Tange H, Troost J, Hasman A. Development and implementation of an EPR: how to encourage the user. International Journal of Medical Informatics 2001;64(2-3):173-85. Cordero L, Kuehn L, Kumar RR, Mekhjian HS. Impact of computerized physician order entry on clinical practice in a newborn intensive care unit. Journal of Perinatology 2004;24(2):88-93. Fernando B, Savelyich BSP, Avery AJ, Sheikh A, Bainbridge M, Horsfield P, et al. Prescribing safety features of general practice computer systems: evaluation using simulated test cases. British Medical Journal 2004;328(7449):1171-72. Teich JM, Merchia PR, Schmiz JL, Kuperman GJ, Spurr CD, Bates DW. Effects of computerized physician order entry on prescribing practices. Archives of Internal Medicine 2000;160(18):2741-47. Evans RS, Pestotnik SL, Classen DC, Bass SB, Burke JP. Prevention of adverse drug events through computerized surveillance. Proceedings of Annual Symposium on Computer Application in Medical Care 1992:437-41. Dortch MJ, Mowery NT, Ozdas A, Dossett L, Cao H, Collier B, et al. A computerized insulin infusion titration protocol improves glucose control with less hypoglycemia compared to a manual titration protocol in a trauma intensive care unit. Journal of Parenteral and Enteral Nutrition 2008;32(1):18-27. Boord JB, Sharifi M, Greevy RA, Griffin MR, Lee VK, Webb TA, et al. Computer-based insulin infusion protocol improves glycemia control over manual protocol. Journal of the American Medical Informatics Association 2007;14(3):278-87. Fortescue EB, Kaushal R, Landrigan CP, McKenna KJ, Clapp MD, Federico F, et al. Prioritizing strategies for preventing medication errors and adverse drug events in pediatric inpatients. Pediatrics 2003;111(4):722-29. Weir CR, Staggers N, Phansalkar S. The state of the evidence for computerized provider order entry: A systematic review and analysis of the quality of the literature. International Journal of Medical Informatics 2009;78(6):365-74. Cresswell K, Bates, D.W., Phansalkar, S. and Sheikh, A. KOpportunities and challenges in creating an international centralised knowledge base for clinical decision support systems (CDSS) in prescribing. Quality and Safety in Health Care Forthcoming. British TS. British Guideline on the Management of Asthma: Quick Reference Guide, June 2009. Aronson JK. A prescription for better prescribing. British Journal of Clinical Pharmacology 2006;61(5):487-91. Strom BL, Schinnar R, Aberra F, Bilker W, Hennessy S, Leonard CE, et al. Unintended Effects of a Computerized Physician Order Entry Nearly Hard-Stop Alert to Prevent a Drug Interaction A Randomized Controlled Trial. Archives of Internal Medicine 2010;170(17):1578-83. Van der Sijs H, Aarts J, Vulto A, Berg M. Overriding of drug safety alerts in computerized physician order entry. Journal of the American Medical Informatics Association 2006;13(2):138-47. Procter R, Pappas, Y., Car J., Sheikhh A. and Majeed A. From human factors to human actors: taking user involvement in eHealth systems seriously. Forthcoming. Cafazzo JA, Trbovich PL, Cassano-Piche A, Chagpar A, Rossos PG, Vicente KJ, et al. Human factors perspectives on a systemic approach to ensuring a safer medication delivery process. Healthcare Quarterly 2009;12 Spec No Patient:70-4. 308 114. Vicente KJ, editor. The Human Factor. New York: Routledge, 2004. 115. Saathoff A. Human factors considerations relevant to CPOE implementations. Journal of Healthcare Information Management 2005;19(3):71-8. 116. Pearson M. Synthesizing qualitative and quantitative health evidence: A guide to methods. Sociology of Health & Illness 2008;30(2):330-31. 117. Pope C, Mays, Nicholas, Popay, Jennie, editor. Synthesizing Qualitative and Quantitative Health Evidence: A Guide to Methods. New York: Open University Press, 2007. 118. Greenhalgh T, Potts HWW, Wong G, Bark P, Swinglehurst D. Tensions and Paradoxes in Electronic Patient Record Research: A Systematic Literature Review Using the Meta-narrative Method. Milbank Quarterly 2009;87(4):729-88. 119. Lilford RJ, Foster J, Pringle M. Evaluating eHealth: how to make evaluation more methodologically robust. PLoS Medicine 2009;6(11):e1000186. 120. Williams R, Edge D. The social shaping of technology. Research Policy 1996;25(6):865-99. 121. Clamp, S, Keen J. Electronic health records: Is the evidence base any use? Medical Informatics and the Internet in Medicine 2007;32(1):5-10. 122. Durieux P, Trinquart L, Colombet I, Nies J, Walton R, Rajeswaran A, et al. Computerized advice on drug dosage to improve prescribing practice. Cochrane Database of Systematic Reviews 2010(10):CD002894. 309 Chapter 11 Case study: ePrescribing for oral anticoagulation therapy in primary care with warfarin Summary • Anticoagulants are highly efficacious treatments for a range of conditions; they are, however, also associated with considerable risk of iatrogenic harm if monitoring of treatment and dosage adjustments are poorly managed. Monitoring has thus historically taken place in hospital clinics. • Widening indications for the use of oral anticoagulation therapy have, however, in recent years led to overstretch of many hospital-based anticoagulation services. Coupled with the political imperative to provide more accessible, patient-centred, models of care, this has catalysed the recent move of anticoagulation services to primary care. • These drivers need to be balanced against concerns about the safety of prescribing and monitoring of warfarin therapy in primary care. • ePrescribing for oral anticoagulation therapy, specifically the use of computerised decision support systems, has the potential to provide prescribers with real-time advice on prescribing and monitoring decisions. • There is strong and consistent evidence that these theoretical benefits in relation to computerised clinical decision support for oral anticoagulation therapy with warfarin can be realised in primary care resulting in greater therapeutic control than might otherwise be possible. • The automation of dosage calculations and determination of time until the next appointment is acceptable to primary care prescribers. • There are strong arguments for the rolling out and integration of this decision support tool into future upgrades of ePrescribing systems in English primary care. • Understanding the reasons underpinning the success of computerised decision support for oral anticoagulation therapy—which to a large extent relate to the fact that this meets a genuine clinical need rather than a technologically driven “solution”—should provide insights into the contexts in which it is best to prioritise development of other decision support tools 310 11.1 Introduction The National Institute for Health and Clinical Excellence (NICE) has estimated that approximately 1.4% of the population now needs oral anticoagulation therapy (OAT).1 This high population estimate reflects the fact that indications for OAT have broadened in recent years, in particular following publication of the evidence for the effectiveness of warfarin in stroke prevention in patients with atrial fibrillation.2 3 This substantial increase in the absolute number of patients treated with OAT led to overburdened anticoagulation services, the policy response to which has been the development of new models of care, these centring on an outsourcing of care from specialist to generalist settings. 4 5 When the transition was first being made in the early 1990s, general practitioners (GPs) were on the whole resistant to the idea of running their own anticoagulation clinics, this hesitation stemming from: perceived insufficient time, knowledge and training; lack of appropriate facilities and equipment; and a lack of financial incentives.6 Regardless of GPs initial apprehension, where OAT was once delivered exclusively in specialised secondary care clinics, it is now regularly delivered in primary care in England.2;7 Considering the high numbers of people now taking OAT, this is in many senses an appropriate development.8 This reorganisation of the provision of anticoagulation services, however, remains controversial, with criticism centring on the quality of OAT when administered in primary care compared with its delivery in specialist care settings.9 The benefit to risk ratio of anticoagulants depends heavily on keeping anticoagulation control within narrow limits (i.e. within the therapeutic range, which is typically measured using the International Normalised Ratio or INR) in order to minimise the risk of adverse events due either to under- or overtreatment.10 Successful control depends on skilled dosing, effective laboratory quality control, and regular monitoring. Of particular note in the context of this chapter is that in its influential report, Improving Medication Safety, the Department of Health (DH) has highlighted concerns with the prescribing of anticoagulants in primary care, noting that it is now one of the classes of medication most commonly associated with fatal medication errors. 11 Also 311 noteworthy are the findings from a systematic review (SR) and meta-analysis of 45 studies conducted in both primary and secondary care anticoagulation clinics, which found that improved therapeutic control could reduce the risk of anticoagulation-related adverse events by almost half.12 In response to a risk assessment of anticoagulation therapy conducted by the National Patient Safety Agency (NPSA), a report was issued recommending that all healthcare organisations take a range of actions (Box 11.1) under the auspices of Patient Safety Alert 18 to make anticoagulant therapy safer.13 Box 11.1: Actions that can make anticoagulant therapy safer • Ensure all staff caring for patients on anticoagulant therapy have the necessary work competences. Any gaps in competence must be addressed through training to ensure that all staff undertake their duties safely. • Review and, where necessary, update written procedures and clinical protocols for anticoagulant services to ensure they reflect safe practice and that staff are trained in these procedures. • Audit anticoagulant services using British Society for Haematology and NPSA safety indicators as part of the annual medicines management audit programme. The audit results should inform local actions to improve the safe use of anticoagulants, and should be communicated to clinical governance, and drugs and therapeutics committees (or equivalent). Commissioners and external organisations should use this information as part of the commissioning and performance management process. • Ensure that patients prescribed anticoagulants receive appropriate verbal and written information at the start of therapy, at hospital discharge, on the first anticoagulation clinic appointment and when necessary throughout the course of their treatment. • Promote safe practice amongst prescribers and pharmacists to check that patients’ INR is being monitored regularly and that the INR level is safe before issuing or dispensing repeat prescriptions for oral anticoagulants. • Promote safe practice for prescribers co-prescribing one or more clinically significant interacting medicines for patients already on oral anticoagulants to make arrangements for additional INR blood tests and to inform the anticoagulant service that an interacting medicine has been prescribed. Ensure that those 312 dispensing clinically significant interacting medicines for these patients check that these additional safety precautions have been taken. • Ensure that dental practitioners manage patients on anticoagulants according to evidence-based therapeutic guidelines. In most cases, dental treatment should proceed as normal and oral anticoagulant treatment should not be stopped or the dosage decreased inappropriately. • Amend local policies to standardise the range of anticoagulant products used, incorporating characteristics identified by patients as promoting safer use. • Promote the use of written safe practice procedures for the use of anticoagulants in care homes. It is safe practice for all dose changes to be confirmed in writing by the prescriber. A risk assessment should be undertaken on the use of Monitored Dosage Systems for anticoagulants for individual patients. The general use of Monitored Dosage Systems should be minimised as dosage changes using these systems are more difficult. Ensure all staff caring for patients on anticoagulant therapy have the necessary work competences. Any gaps in competence must be addressed through training to ensure that all staff undertake their duties safely. Source: NPSA Patient Safety Alert 18 (2007) 13 Reprinted with permission from the National Patient Safety Agency The NPSA and the British Committee for Standards in Haematology (BCSH) have identified safety indicators (see Box 11.2) for outpatient OAT.14 Monitoring these indicators should help to identify risks and promote appropriate action to minimise risk. The safety indicators can also be used to audit the implementation of the recommendations made in the NPSA’s Patient Safety Alert 18.13 313 Box 11.2: Patient safety indicators for outpatient oral anticoagulation therapy • Proportion of patient time in range (if this is not measurable because of inadequate decision support software then a secondary measure of percentage of INRs in range should be used). • Percentage of INRs > 5·0. • Percentage of INRs > 8·0. • Percentage of INRs > 1·0 INR unit below target (e.g. percentage of INRs < 1·5 for patients with target INR of 2·5). • Percentage of patients suffering adverse outcomes, categorised by type, e.g. major bleed. • Percentage of patients lost to follow up (and risk assessment of process for identifying patients lost to follow up). • Percentage of patients with unknown diagnosis, target INR or stop date. • Percentage of patients with inappropriate target INR for diagnosis, high and low. • Percentage of patients without written patient educational information. • Percentage of patients without appropriate written clinical information, e.g. diagnosis, target INR, last dosing record. 14 Adapted from BCSH (2007). Reprinted with permission from the National Patient Safety Agency. Employing an ePrescribing application (see Chapter 10)–with computerised decision support system (CDSS) that provides support for determining dosage and time until next appointment–in primary care has been shown to be effective in improving therapeutic control.7 support systems could furthermore 15 16 17 18 facilitate The use of decision adherence to NPSA recommendations and aid in the monitoring of BCSH safety indicators to support the provision of anticoagulation services in primary care. 11.2 Theoretical considerations 11.2.1 Potential benefits Decision support systems for OAT can utilise input which takes the form of a target INR, the therapeutic range for a specific condition, history of bleeding problems, number of prior visits to the clinic, variability of prothrombin over time, cost of complications and the estimated costs of a visit.19 20 Systems employ a number of inference mechanisms (see Chapter 10), which draw on 314 pharmacological models of anticoagulation control, to arrive at the recommended output. The use of ePrescribing for OAT has the potential to benefit patients, providers and health services. For example, patients may benefit from improved therapeutic control, spending more time within the therapeutic range, thereby reducing the number of INR tests and visits required to maintain therapeutic control in comparison to therapy delivered without a CDSS. Depending on whether near patient testing (NPT; also sometimes known as “point-of-care” or POC testing) with coagulometers is employed in primary care, the need for laboratory monitoring could be almost entirely removed with the potential benefit of increased patient convenience. This is because use of NPT also virtually eliminates turnaround time for results compared with centralised lab-based testing, making the process of anticoagulation management both faster and more straightforward. Providing anticoagulation services in primary care should furthermore enable clinicians to provide more comprehensive and continuous healthcare; and this more complete clinical knowledge about patients should enable clinicians to make more informed clinical decisions both in relation to anticoagulation decisions and care more generally. As the use of CDSSs allows for the management of patients by non-clinicians, some workload may be shifted away from clinical staff, thus creating new roles and responsibilities for other (less expensive) healthcare providers. Use of these technologies also offer the inherent advantage of opening up the database of patients to a variety of secondary uses, these including audit, significant event analysis and research.21 11.2.2 Potential risks The quality and/or efficiency of healthcare often initially suffers as new systems are introduced and take time to embed (see Chapter 18); there are 315 also resource and opportunity costs associated with system implementation. Cognisance of the disruptive effect on clinical workflow associated with implementation is therefore important. Training of staff is important–as is, for example, highlighted by The National Centre for Anticoagulation Training, which cautions that the use of a CDSS should be restricted to those staff that have completed appropriate training and recommend a training record log is kept for all staff members working in such clinics–but this is also likely to disrupt clinical workflows by taking busy professionals out of other core activities.7 Although there have been no reports of CDSSs providing care inferior to that delivered without CDSSs, this might occur if the support provided by the CDSS is predicated on an invalid knowledge base or algorithm. Even though it is in the manufacturers’ best interests to produce systems that have been rigorously evaluated, there is as noted in earlier chapters no third party assurance of validity and safety. As such, patients might actually suffer from a worsening in therapeutic control, with the associated increased risks of adverse events. The provision of less effective OAT might not be detected initially and if detected, could result in abandonment of the application and subsequent financial loss for primary care. Careful monitoring and evaluation of the impact of the application on the quality and safety of healthcare is thus imperative in order to reduce risks and attend to problems expeditiously. 11.3 Empirically demonstrated impact 11.3.1 Benefits The SR by Garg et al. included a number of randomised (and other) trials of warfarin dosing systems.22 Seven of twelve trials for warfarin dosing improved practitioner performance as measured in the main by time to achieve a therapeutic INR, the proportion of time within the therapeutic range, the proportion of patients with therapeutic INR, the number of days between INR testing and the number of test measurements. This review, however, included both initiation and maintenance phases of OAT conducted in inpatient and 316 outpatient settings. This represents a major challenge to interpreting the findings from this review as the underlying algorithms differ between the initiation and maintenance phases.23 An earlier SR and meta-analysis of nine randomised controlled trials (RCTs) with a total of 1336 patients—the majority of whom were also included in the review by Garg et al.22—focusing largely on systems for OAT with warfarin, found that the use of computer programmes for anticoagulation optimisation increased the proportion of visits where patients were within the therapeutic range by 29 per cent (pooled odds ratio 1.29; 95%CI 1.12, 1.49).24 There was, however, significant heterogeneity (P=0.02) between trials making this summary analysis open to question; to their credit, the authors subsequently conducted two additional analyses to investigate the source of the heterogeneity after excluding: • The only study for which the medication in question was heparin, the results remained unchanged from that of the overall analysis • One of the smallest studies with the largest effect the pooled odds ratio decreased slightly to 1.25 (95%CI 1.08, 1.45), but this however reduced the heterogeneity such that it was no longer significant (P=0.12); this additional analysis thus uncovered the source of the heterogeneity and, importantly, demonstrated that even after removing this trial that CDSS for OAT resulted in significant improvements in INR control.24 The unit of assessment in this meta-analysis was not the patient, but the anticoagulation test and the end point of the proportion of tests within the target range; using this outcome, the sample size was 3416 dosages (carried out in 1327 patients).24 Further, it should be noted that this SR and metaanalysis also included studies of both initiation and maintenance phases and both inpatients and outpatients; it therefore poses many of the same challenges to interpretation as Garg et al’s review.22 317 Another SR on the use of CDSSs for OAT, focusing on its use in primary care with NPT, was published in 1998.24 Of the seven included studies (with little overlap of studies included by Garg et al.22 and Chatellier et al.24), only one was judged to be of high quality. The authors concluded from the one high quality study that there was evidence that a CDSS could achieve improved therapeutic control in terms of INR, when compared with human performance alone.24 The important series of studies by Fitzmaurice and colleagues from the University of Birmingham, England are worth reflecting on in greater detail. The first study, published in 1996, assessed 49 patients treated with warfarin for 12 months using a RCT design.15 There were significant improvements noted in the proportion of patients with INR control within therapeutic limits, from 23 per cent to 86 per cent (P<0.001) in the practice where all patients received dosages with the aid of a CDSS. In the practice where patients were randomised to either CDSS or hospital dosage, analysis showed a significant improvement in the CDSS group, which was not apparent in the patients who received dosage in hospital (P<0.001). Mean recall times were significantly extended in patients who received dosage instructions with the aid of the CDSS, from 24 to 36 days (P=0.03). Patient satisfaction with the practice clinics was also high. This study, however, did not use NPT and specimens were sent to laboratory with results returned usually on the same day.15 An extension of this study–with the addition of NPT–was published two years later.16 This was conducted outside trial conditions with data collected over the course of 12 months from a dedicated nurse-led OAT clinic within primary care. The cumulative results of this longitudinal study were that the overall mean percentage of patients within the therapeutic range was 71 per cent and overall the proportion of INRs within the therapeutic range was 53 per cent. Importantly, no adverse events were reported and no patients had to be referred back to secondary care.16 The same group published a second RCT in 2000.17 Of the 248 practices within Birmingham, England, 12 were randomly selected from a list of 21 318 practices that had expressed interest in the study. The design was somewhat unusual in that there were two sets of controls: those from the intervention practices (described as the intra-practice controls) and all patients in the control practices (described as the inter-practice controls). The intervention was a nurse-led practice-based anticoagulation clinic using CDSS and NPT, this being compared to the standard hospital-based assessment. The primary outcome for this trial was therapeutic control of the INR and this showed significant improvements for the intervention patients (P<0.01).17 An extension of this study published in 2001 reported the degree of OAT control for patients from the same practices for an 18-month period after the completion of the original study.18 the authors found that there were no significant differences between the practice- and hospital-based populations in terms of the percentage of time in range (69 per cent practice-based, 64 per cent hospital-based). The proportion of tests in range was, however, significantly higher in the practice-based group at 61 per cent vs 57 per cent for hospital-based (P=0.02). Mean recall time was virtually identical in both groups at 36 days. There were no significant differences between groups for the number of clinical outcomes per patient. Overall, the authors found comparable quality of control when compared to the original study, this demonstrating that primary care-based OAT supported by CDSS and NPT was of at least equivalent quality to the hospital-based care.18 In their 1996 study, using average hospital review rates, Fitzmaurice et al. estimated 148 additional appointments would have been offered to 26 patients in an intervention practice during the 12 months if CDSS had not been used.15 The authors assessed the total cost savings to be non-existent in the first year of use (-£476), but estimated a savings of £2,604 for each subsequent year for the intervention practice with 26 patients. The authors also concluded that the greater the number of patients seen in a high cost provider environment, the more economically efficient a CDSS becomes.15 Extending their 1996 study15 outside trial conditions with the addition of NPT, Fitzmaurice et al. estimated a total cost savings of £539, from £2290 to £1751 319 from the use of their dedicated, nurse-led primary care anticoagulation clinic using a CDSS and NPT for the 12-month study period.16 In an analysis of patients’ costs in primary versus secondary care, a patient’s cost per visit was found to be significantly higher when being reviewed in secondary care: average patient cost per visit in primary care of £6.78 vs £14.58 in secondary care. These cost differences were to a large extent as a result of longer journey time and the use of more expensive means of transport when attending for hospital-based assessments, and shorter clinic visits in primary care.25 11.3.2 Risks There was no evidence of an increased risk of adverse events reported in these studies of primary care delivered CDSS supported OAT. Furthermore, there were no significant new risks introduced indicating that the underlying algorithms are on the whole probably well-constructed. The overall costs to healthcare systems are however likely to be increased with primary care-led models of care.17 25 26 11.4 Conclusions: Implications for practice, policy and research The principal outcome measures for any anticoagulation service are the prevention of thrombotic and avoidance of haemorrhagic events.23 Unfortunately, none of the SRs investigating the role of CDSSs found clear evidence for improved outcomes for these clinically important endpoints. There are, more generally, limited data regarding the absolute and relative risks of OAT as most studies lack sufficient power to detect significant changes in these relatively uncommon outcomes.17 22 24 This issue is further complicated by the lack of agreement on definitions and criteria for major and minor adverse events. It is perhaps for these reasons that evaluations have tended to focus on time spent on proxy measures such as the proportion of patients in and time spent within the therapeutic range. As most of the research pertains to warfarin, further research into the prescribing of other oral anticoagulants such as phenindione and 320 acenocoumarol should also be considered, as the findings might not be directly transferable to these other oral anticoagulants.26 There are also outstanding questions about the positioning of newer agents–which do not require monitoring–such as dabigatran and rivaroxaban, which may in due course replace use of warfarin.27 28 Finally, both the systematic reviews by Garg et al.22 and Chatellier et al.24 did not mention whether NPT was part of the intervention in the included primary studies. This is important as the degree to which an additional technology such as NPT contributes to the overall effect of CDSSs for anticoagulation is not yet well understood. However, the automation of dosage calculation and determination of time until next appointment has proved to be readily adoptable. As outlined in the chapters on CDSSs, ePrescribing human factors and organisational issues in design, development and deployment (see Chapters 8, 10, 16 and 17) the success of an application is dependent on a number of variables. It seems likely that in this case of CDSS-supported OAT this is most likely due to the relative advantage of using the technology being driven by a recognised need to improve the quality and safety of OAT and reduce the inconvenience to patients (rather than primarily a desire to improve technology and application validity). That said, there is still the need for further research, specifically to compare in a head-to-head fashion, the different algorithms that are currently being employed in CDSS applications for OAT using warfarin, to establish which, if any, is best at supporting practitioner performance. A final point worth noting about CDSSs for OAT is that the applications are for the most part stand-alone systems and this poses potential problems with regards to integration with existing records (see Chapter 5 for a further discussion on the matter). Concerns over interoperability are likely to serve as a deterrent to implementation and use of the application in primary care and may also potentially compromise safety as a consequence of fragmenting 321 the record of care. This concern should be mitigated by ensuring compatibility with primary care computing systems or incorporating the functionality within ePrescribing systems. Increased provision of OAT opens the potential door to a future move to patient self-testing and -monitoring based models, however, available data suggest that whilst effective, these models are not particularly cost-effective.29 In summary, given that the technical infrastructure exists within primary care in which to readily incorporate this particular ePrescribing application, the fact that there are also now clear financial incentives to provide OAT services in primary care and the clearly demonstrated beneficial impact on the quality of anticoagulation care, it is important that this eHealth application is now used much more widely in English primary care.30 References 1. National Institute for Health and Clinical Excellence (NICE). Assumptions used in estimating a population benchmark. Available from: http://www.nice.org.uk/usingguidance/commissioningguides/anticoagulationthera pyservice/popbench.jsp (last accessed 30/12/10) 2. Ezekowitz MD, Bridgers SL, James KE, Carliner NH, Colling CL, Gornick CC, for the Veterans Affairs Stroke Prevention in Nonrheumatic Atrial Fibrillation Investigators. Warfarin in the prevention of stroke associated with nonrheumatic atrial fibrillation. N Engl J Med. 1992;327:1406–12 3. Aguilar MI, Hart R. Oral anticoagulants for preventing stroke in patients with nonvalvular atrial fibrillation and no previous history of stroke or transient ischemic attacks. Cochrane Database of Systematic Reviews 2005, 3. Art. No.: CD001927. DOI: 10.1002/14651858.CD001927.pub2 4. Sudlow CM, Rodgers H, Kenny RA, Thomson RG. Service provision and use of anticoagulants in atrial fibrillation. BMJ 1995; 311: 558-60. 5. Sweeney KG, Gray DP, Steele R, Evans P. Use of warfarin in non-rheumatic atrial fibrillation: a commentary from general practice. Br J Gen Pract 1995; 45: 153-58. 6. Taylor F, Ramsay M, Voke J, Cohen H. Anticoagulation in patients with atrial fibrillation. GPs not prepared for monitoring anticoagulation. BMJ 1993; 307: 1493. 7. National Centre for Anticoagulation Training. Available from: http://www.anticoagulation.org.uk/courses.php (last accessed 30/12/10) 8. National Institute for Health and Clinical Excellence. Anticoagulative therapy service. Commissioning Guides - Supporting Clinical Service Redesign. Available from: http://www.nice.org.uk/page.aspx?o=370987 (last accessed 30/12/10). 322 9. Wilson SJ, Wells PS, Kovacs MJ, Lewis GM, Martin J, Burton E et al. Comparing the quality of oral anticoagulant management by anticoagulation clinics and by family physicians: a randomized controlled trial. CMAJ 2003; 169: 293-98. 10. Oake N, Jennings A, Forster AJ, Fergusson D, Doucette S, van Walraven C. Anticoagulation intensity and outcomes among patients prescribed oral anticoagulant therapy: a systematic review and meta-analysis. CMAJ 2008; 179: 235-44. 11. Department of Health. Building a Safer NHS for Patients: Improving Medication Safety. London: DH, 2004. 12. Oake N, Fergusson DA, Forster AJ, van Walraven C. Frequency of adverse events in patients with poor anticoagulation: a meta-analysis. CMAJ 2007; 176: 1589-94. 13. National Patient Safety Agency. Patient Safety Alert 18 - Actions that can make anticoagulant therapy safer. Available from: http://www.nrls.npsa.nhs.uk/resources/?entryid45=59814 (last accessed 30/12/10) 14. Baglin TP, Cousins D, Keeling DM, Perry DJ, Watson HG. Safety indicators for inpatient and outpatient oral anticoagulant care: recommendations from the British Committee for Standards in Haematology and National Patient Safety Agency. Br J Haematol 2007; 136: 26-29. 15. Fitzmaurice DA, Hobbs FD, Murray ET, Bradley CP, Holder R. Evaluation of computerized decision support for oral anticoagulation management based in primary care. Br J Gen Pract 1996; 46: 533-35. 16. Fitzmaurice DA, Hobbs FD, Murray ET. Primary care anticoagulant clinic management using computerized decision support and near patient International Normalized Ratio (INR) testing: routine data from a practice nurse-led clinic. Fam Pract 1998; 15: 144-46. 17. Fitzmaurice DA, Hobbs FD, Murray ET, Holder RL, Allan TF, Rose PE. Oral anticoagulation management in primary care with the use of computerized decision support and near-patient testing: a randomized, controlled trial. Arch Intern Med 2000; 160: 2343-48. 18. Fitzmaurice DA, Murray ET, Gee KM, Allan TF. Does the Birmingham model of oral anticoagulation management in primary care work outside trial conditions? Br J Gen Pract 2001; 51: 828-29. 19. Ryan PJ, Gilbert M, Rose PE. Computer control of anticoagulant dose for therapeutic management. BMJ 1989; 299: 1207-09. 20. Poller L, Shiach CR, MacCallum PK, Johansen AM, Munster AM, Magalhaes A et al. Multicentre randomised study of computerised anticoagulant dosage. European Concerted Action on Anticoagulation. Lancet 1998; 352: 1505-09. 21. Jones RT, Sullivan M, Barrett D. INRstar: computerised decision support software for anticoagulation management in primary care. Inform Prim Care 2005; 13: 215-21. 22. Garg AX, Adhikari NK, McDonald H, Rosas-Arellano MP, Devereaux PJ, Beyene J, et al. Effects of computerized clinical decision support systems on practitioner performance and patient outcomes: a systematic review. JAMA 2005; 293: 122338. 23. Fitzmaurice DA, Hobbs FD, Delaney BC, Wilson S, McManus R. Review of computerized decision support systems for oral anticoagulation management. Br J Haematol 1998; 102: 907-09. 24. Chatellier G, Colombet I, Degoulet P. An overview of the effect of computerassisted management of anticoagulant therapy on the quality of anticoagulation. Int J Med Inform 1998; 49: 311-20. 25. Parry D, Bryan S, Gee K, Murray E, Fitzmaurice D. Patient costs in anticoagulation management: a comparison of primary and secondary care. Br J Gen Pract 2001; 51: 972-76. 323 26. NHS Economic Evaluation Database. Available from: http://www.crd.york.ac.uk/crdweb/ShowRecord.asp?LinkFrom=OAI&ID=2200100 0113 (last accessed 31/12/10) 27. Connolly SJ, Ezekowitz MD, Yusuf S, Eikelboom J, Oldgren J, Parekh A, et al. Dabigatran versus warfarin in patients with atrial fibrillation. N Engl J Med 2009; 361: 1139-51. 28. EINSTEIN Investigators, Bauersachs R, Berkowitz SD, Brenner B, Buller HR, Decousus H, et al. Oral rivaroxaban for symptomatic venous thromboembolism. N Engl J Med 2010; 363: 2499-510. 29. Connock M, Stevens C, Fry-Smith A, Jowett S, Fitzmaurice D, Moore D, et al. Clinical effectiveness and cost-effectiveness of different models of managing long-term oral anticoagulation therapy: a systematic review and economic modelling. Health Technol Assess 2007; 11: 1-66 30. Refreshed indicators, December 2010. Available from: http://www.ic.nhs.uk/webfiles/Work%20with%20us/consultations/CQI/IQI%20Wha ts%20New%20docs/IQI_Update_Dec_2010.pdf (last accessed 31/12/10). 324 Chapter 12 Case study: eHealth-based approaches to reducing repeat drug exposure in patients with known drug allergies Summary • There is increasing interest internationally in ways of reducing the high disease burden resulting from errors in medicine management. Repeat exposure to drugs to which patients have a known allergy has been a repeatedly identified error, often with disastrous consequences. • Drug allergies are immunologically-mediated reactions that are characterised by specificity and recurrence on re-exposure. These repeat reactions should therefore be completely preventable. • There is at present insufficient attention being paid to studying and implementing system-based approaches to reducing the risk of such accidental re-exposure. • A number of eHealth-based interventions may in this context be used to reduce risk of recurrent exposure, these including: o ePrescribing o bar-coding o radio-frequency identification o biometric technologies. • The most consistent evidence in this respect comes from interventions which provide prescribers with real-time clinical decision support; whilst promising, there is however as yet less clear evidence in support of interventions aiming to enhance patient recognition such as bar-coding, radio-frequency identification and biometric technologies. • ePrescribing support for drug allergies is now routine in UK general practice. Its use to hospital prescribing now needs to be considered. • An important underlying limiting factor, however, is the need to improve the accuracy of drug allergy coding in electronic records; this is because ePrescribing systems are fundamentally dependant on the accuracy of these coded patient data. 325 12.1 Introduction 12.1.1 Key concepts and definitions An adverse drug reaction (ADR) is an undesired outcome attributable to a drug. There are two main classes of ADRs recognised: Type A ADRs are predictable and therefore potentially preventable, while Type B ADRs are unpredictable and therefore far more difficult to prevent.1 Allergic reactions to drugs are a class of Type A ADRs, which are immunologically-mediated, this reaction being directed either at the drug in question or its breakdown product.2 The still widely used Gell and Coombs classification system can be used to categorise allergic reactions on the basis of the underlying mechanisms involved.3 4 These immunological mechanisms share the characteristic of specificity, transferability and, crucially, a high risk of recurrence on re-exposure. Repeat reactions in individuals with a previous reaction should therefore be preventable. There are however often a number of important practical challenges to establishing a diagnosis of drug allergy, these including the need to obtain a detailed clinical history, the relatively few validated tests available and the issue of possible cross-sensitivity, all of which often result in considerable uncertainty in relation to securing a diagnosis of drug allergy. Despite these difficulties, it is however often possible for clinicians to arrive at a working diagnosis of drug allergy and it is these individuals–that is, those in whom a working diagnosis of drug allergy has been made and documented in the clinical records–that represent the focus of our deliberations in this case study and who are thereafter described as having a known drug allergy. 12.1.2 Scale of the problem Obtaining epidemiological data on the frequency of re-exposure to drugs in those with known drug allergies is difficult for a number of reasons. Firstly, there is a problem with under-reporting in schemes that collect information on adverse drug reactions, such as the Yellow Card system run by the UK Medicines and Healthcare products Regulatory Agency (MHRA),5 the US 326 Food and Drug Administration’s MedWatch Program (http://www.fda.gov/medwatch/index.html), or the Reporting and Learning System (RLS)–the first national database of patient safety incidents.6 Secondly, retrospective reviews may not provide sufficient information to clearly establish the background and context of the event raising the possibility of significant misclassification errors. Thirdly, there is still uncertainty among many patients and healthcare professionals as to which reactions are truly allergic and which are, for example, due to intolerances and other non-allergic mechanisms. Despite these difficulties, data from studies conducted in US secondary care suggest that 6-12 per cent of medication errors are due to patients receiving a medication to which they were already known to be allergic.7 8 9 This translates into between 90,000 and 180,000 episodes per year in the US (based on the assumption that medication errors harm about 1.5 million people per year). Research from US ambulatory care suggests similarly high figures.10 Crucially, most instances of these repeat exposures have been judged across studies as being preventable.9 10 More recent epidemiological data comes from an analysis of around 60,000 patient safety incidents reported to the RLS between 2005 and 2006.11 This found that the administration of medication–despite a known drug allergy–was the cause of 3.2% of medication-related patient safety incidents in hospitals and of 2.6% of incidents in family practice settings. Almost one-third of these resulted in harmful consequences to patients, including death in some cases. Drugs most frequently involved in these incidents were found to be antibiotics (in particular, penicillins), opiods and non-steroidal anti-inflammatory drugs (NSAIDS). Moreover, the Department of Health’s (DH) report of medical insurance claims found that 11 of 193 (5.7%) Medical Protection Society claims (for general practice) and 25 of 234 (10.7%) Medical Defence Union claims (for hospital care) related to allergic reactions to medication.12 Many of these concerned antibiotics that had been prescribed to patients with known allergies.11 Other drugs and agents that are commonly responsible for 327 triggering allergic reactions include antihypertensive agents, vaccines, anaesthetic agents, anti-epileptic and anti-arrhythmia medications. The conclusions drawn from these studies do however have to be treated with some caution as data derived from retrospective studies of adverse drug events may not be directly comparable to studies of global medication errors or studies of medication administration. Nevertheless, a growing body of international evidence suggests that accidental re-exposure in patients with known drug allergies is one of the most common preventable medication errors, sometimes with disastrous consequences for the patient (see Box 12.1). This appears to be true for both hospital and community care. Of concern is that despite the frequency of accidental re-exposure and the likely consequences, systems-based approaches to reducing the risk of repeat exposure in individuals with known drug allergies are not being given the attention they warrant, as evidenced, for example, by their failures to be considered in key relevant international allergy guidelines.13 14 Box 12.1 Specimen cases reported to the Reporting and Learning System 6 “GP [general practitioner] on call administered amoxicillin to the patient. A short time later the patient collapsed in cardiac arrest. An ambulance was called and a CPR [cardiopulmonary resuscitation] protocol followed. Whilst CPR was being completed, it was ascertained that the patient had been prescribed an antibiotic ... which a relative stated the patient had told the GP that they were allergic to. (Reported as resulting in death)” “A patient was prescribed and given a dose of amoxicillin. The patient developed itching and felt unwell, and later collapsed requiring antihistamine and adrenaline injection to treat the reaction. The patient was known to be penicillin allergic but this had not been transcribed onto the new medicine chart.” 328 “A patient with a penicillin allergy documented on their medicine chart and who was wearing a red allergy alert wristband, was prescribed and given a dose of piperacillin.” 12.2 The theoretical and empirical evidence for eHealth-based approaches to reducing repeat drug allergy As discussed in the preceding chapter, eHealth-based solutions are particularly promising in reducing risk of re-exposure as they can help in overcoming some of the underlying systemic failings, particularly in relation to managing, processing, retaining and making accessible large amounts of disparate data to multiple end-users. Below we consider theoretical, and where available, empirical evidence of a number of key eHealth-based approaches currently under investigation with the aim of reducing the risk of re-exposure in individuals with known drug allergies. In so doing, where possible, we focus on findings from studies employing rigorous designs, but as this is very much an area still under development–particularly in relation to some of the newer technologies–where relevant experimental studies have yet to be conducted, we also draw on evidence from formative work and/or research in other disease areas which may be generalisable to the context of managing patients with known drug allergies. A detailed summary of the studies on which we have drawn to inform this case study are included in the Online Repository accompanying the publication on which this chapter is based.15 12.2.1 Computer systems with warning messages There is some encouraging evidence for the effectiveness of using computer systems incorporating hazard messages that alert healthcare professionals to patients’ allergies (see Figure 12.1). For example, Bates and colleagues evaluated the effectiveness of ePrescribing alone and in combination with a 329 team intervention in preventing medication errors in a tertiary care hospital over a six month period.16 In this study, clinicians used a computerised physician order entry (CPOE) system, which involved electronically selecting and ordering prescription requests that were chosen from a drop-down menu. The system also had a component that checked for the most common drug allergies. It was found that the ePrescribing system reduced known drug allergy errors by 56% (P=0.009). In a similar study in an intensive care unit, Evans and colleagues found that a computerised decision support system (CDSS) programme, which was linked to patient records and issued recommendations as well as warnings, significantly reduced (from 146 to 35, P<0.01) the number of instances in which drugs were prescribed to which the patient had a recorded allergy.17 Studies of ePrescribing systems in general practice have pointed in a similar direction.10 Figure 12.1: Example of a drug allergy alert Despite these encouraging results there are several issues with computerised prescribing support tools that often prevent them for realising their potential. 330 Firstly, they require relevant and accurate drug allergy information to be entered into the record in order to be effective. Secondly, the production of hazard alerts of known drug allergies depends on the level of system sophistication.16 Thirdly, studies have shown that there is often a lack of training amongst healthcare professionals as to how best to use these systems, sometimes resulting in a false sense of security if data on allergies are, for example, entered into an inappropriate section of the records or as a free text entry rather than as a code that will be recognised by the system.18 Fourthly, they often lack specificity, which can result in spurious hazard messages being generated. This in turn may lead to frustration amongst users with the consequent danger that hazard messages are accidentally overridden.19 Conversely, an over-reliance on warnings may contribute to errors with clinicians “blindly” trusting system alerts.18 Fifthly, there are still sometimes problems with the underlying design of the prescribing software, as demonstrated in a study by Fernando and colleagues in which they tested the safety features of four widely used family practice computer systems in the UK and found that only three out of four systems produced alerts when penicillin was prescribed to a dummy patient with a penicillin allergy previously recorded in the system.20 Thus efforts still need to concentrate on improving the safety features of such prescribing support systems in order to further enhance their effectiveness. And finally, as discussed in earlier chapters, it is still unknown whether this evidence of effectiveness from “home-grown” systems studied in “bench-mark” institutions is generalisable to routine care settings, which will in most cases be using commercial systems; it is also unknown whether these positive impacts on errors in prescribing will translate into actual improved outcomes for patients i.e. a reduced incidence of reactions. Kuperman and colleagues have recommended several approaches to ensure more effective checking in relation to drug allergies.21 Included in these is the suggestion that computer systems should store allergy information centrally in order to ensure it is up-to-date and readily available in different settings (e.g. family practice and hospital). The authors also suggest assigning levels of importance to allergy warnings so that the likelihood of truly important alerts 331 being over-ridden will be minimised. Creating a common language for groups of allergies and strategies to prompt healthcare professionals to enter all allergic reactions into the system are additional helpful suggestions. Investigations into the effectiveness and potential improvement of existing computer systems incorporating hazard messages alerting healthcare professionals to patients’ allergies are ongoing. 12.2.2 Bar-codes in wristbands Another stream of efforts has focused on investigating the use of bar-coded wristbands in hospital settings. While colour-coded wristbands can help to identify a particular risk in a patient (such as a drug allergy), bar-coded wristbands can provide more comprehensive patient-specific information on the class of drug to which the patient is allergic. Here, the system includes a bar-code reader that is connected to a host computer that retrieves the desired patient information when the wristband is scanned. The scanned information can then be checked against the medication packaging, which may also have a bar-code. Although there is a broader literature surrounding bar-codes on medication packaging, we focus here on bar-codes in wristbands as this is most relevant for preventing repeat exposure in patients with drug allergies. A number of different wristband bar-coding systems have now been developed and are at various stages of implementation in hospitals across the world. For example, Franklin and colleagues have recently investigated the effectiveness of what is described as a “closed-loop electronic prescribing and administration system” in a surgical ward of a UK teaching hospital.22 This involves giving patients bar-coded wristbands, which are connected to a reader on a drug trolley. After the wristband is scanned, the trolley releases medication through a drawer; details of items dispensed in this way are stored on computer. This has been shown to significantly reduce prescribing and medication administration errors; it also resulted in increased identity checking of the patient by hospital staff. 332 The majority of studies investigating the effectiveness of bar-coded wristbands come from the area of transfusion medicine. For example, an investigation by the Department of Haematology at the Oxford Radcliffe Hospital in the UK found that the introduction of patients wearing a bar-coded identification wristband that could be scanned into a handheld computer for compatibility resulted in a significant improvement in patient identification.23 When the same model was tested in two other hospitals, one in Italy and the other in the US,24 similar results were obtained, thereby demonstrating the potential generalisability of these findings. Whilst promising, there is however as yet a paucity of high quality evidence from randomised controlled trials (RCTs) for the effectiveness of bar-coded wristbands. Systems have also so far not been tested specifically with regard to reducing known drug allergies. In addition, some have questioned the practicality of such devices. The National Patient Safety Agency (NPSA), for example, refers to healthcare staff’s lack of compliance with using bar-coded wristbands (see Box 12.1) and a lack of standardisation across institutions.25 There is furthermore a danger of patients inadvertently being given a wristband belonging to another patient and the associated risk that may ensue from this misidentification.26 Other issues include the problem of access as wristbands need to be scanned with a handheld device from relatively close proximity. This might under some circumstance prove problematic, for example if the position of the patient does not permit scanning (e.g. they are sleeping or unconscious) or if the handheld scanner is impractical to use (e.g. in the operating room).27 Bar-code technology may also be used for Medic-Alert identification jewellery. These are simple bracelets or necklaces with the medical condition of the patient (e.g. allergy information) is recorded along with an identification number and a phone number for someone who can be contacted in an emergency. Although Medic-Alert bracelets are commonly recommended to allergic patients and highly valued by allergists (especially in patients with anaphylaxis), we are not aware of any studies evaluating their effectiveness. A likely benefit of these bracelets is that they can be carried around by 333 patients and therefore used outside hospital settings. Nevertheless, it has to be kept in mind that in addition to problems mentioned above, there may be issues with compliance of wearing Medic-Alert bracelets.28 12.2.3. Radio-frequency identification A more recent development still is radio-frequency identification (RFID). These are chipped tags that are connected to a transceiver via radio-waves, which allow the storage of information (such as drug allergies) remotely. Radio-frequency identification chips allow more information to be stored than bar-codes, are more user-friendly (as they can be scanned through fabric) and are commonly regarded as more fail-safe (as they can work independently of any user input). They are also relatively cheap at a cost of between 5-10 pence each. Pilot studies in hospitals support the feasibility of such systems, but as with bar-codes, more rigorous evidence of effectiveness is still needed. Most activity in the developing and implementing RFID chips can currently be found in the US. For example, the Jacobi Medical Center in New York has already implemented a system using RFID tags in wristbands to improve patient identification and reduce medication errors in two acute care departments.29 Patients are issued with a so-called “Smart Band” on admission, which can be scanned by a handheld reader that brings up all necessary patient information on a screen at the bedside (see Figure 12.2). This information can then be updated and modified as necessary. The Jacobi Medical Center has reported a reduction of medication errors as well as staff time associated with data entry as a result of introducing the system. 334 Figure 12.2: Handheld scanner attached to a screen and RFID tagged wristband used at the Jacobi Medical Center in New York (Adopted from http://www.ti.com/rfid/docs/news/eNews/enewsvol40.htm) Again, there are several pilot studies using RFID chips to try to reduce errors in blood transfusion. The San Raffaele Hospital in Milan (Italy) is currently piloting a system, where RFID tags are incorporated into wristbands given to donors.30 These are matched with a tag on the blood, which is then scanned at the time of transfusion (see Figure 12.3). Similar systems are also currently being piloted in hospitals in the German Saarbrucken Clinic Winterberg, in Washington and Boston in the US, in Taiwan and in England.31 32 33 34 Figure 12.3: Blood handling process using RFID technology at the San Raffaele Hospital in Italy 335 (Adapted from Cisco systems at http://www.cisco.com/web/IT/local_offices/case_history/rfid_in_blood_transfus ions_final.pdf) In the UK, the Birmingham Heartlands Hospital has been implementing RFID wristbands in surgical wards after successful piloting. Here the whole operating team has wireless personal digital assistants that can be used to scan a patient’s wristband giving instant access to operating schedules and patient records.35 Taking the use of RFID chips one step further, the Hackensack University Medical Center in the US is currently piloting a patient identification system using implantable RFID chips for patients with chronic conditions, which provides detailed emergencies. 36 and instantly accessible patient information in A similar pilot has been started in the Alzheimer's Community Care headquarters in the US.37 The advantages of these implants include the comfort of wearing them and their potential application in the community. Here the family physician will merely have to scan the patient, gaining instant and up-to-date access to medical records. These implants do however raise important ethical concerns and their effectiveness remains to be evaluated in high quality clinical trials. 12.2.4. Biometric technologies Another possible way to facilitate patient identification (and therefore potential drug allergies) is the use of biometric technologies such as fingerprint, face or iris scanning. These systems are not as invasive as implanted RFID chips and have the distinct advantage of being easily and instantly accessible. Obviating the need for patients to carry or wear an external object there are also no issues with compliance. Such systems have already been piloted around the world, again mainly in the US. The majority of applications have been implemented in outpatient clinics helping to identify patients coming in to collect medication. For example, the Netherlands are using fingerprint scanners to identify heroin addicts who come to methadone distribution 336 services.38 Also, an increasing number of healthcare providers in the US are using fingerprint scanning to prevent recipient and provider fraud.39 Moreover, South Africa has launched a project allowing for patient identification using fingerprints, and the “Methadose” system in Australia is scanning the irises of patients who collect their methadone at St Vincent's Hospital in Sydney.40 In England, the Patient Access to Electronic Records System (PAERS) gives patients electronic access to their medical records in general practices via fingerprint scanning. The system is now implemented in some first-wave practices across England.41 However, again robust scientific evidence of the effectiveness of such systems is lacking. These remain as yet untested with regard to reducing repeat exposure in patients with known drug allergies–an area in which their application might have significant potential. 12.2.5 Patient-managed electronic health records The approaches outlined above all involve some type of IT application. They also rely, to a lesser or greater extent, on patient involvement and it is important in this respect that parallel attempts are made to improve providerpatient communication and more actively involve patients (and carers) in their care with a view to reducing such reactions.42 Patients, after all, have the most to lose through accidental re-exposure and the most to be gained through getting involved with care delivery in general. Patient-generated and managed electronic health records will thus be particularly important in this respect, with the onus on patients taking responsibility for ensuring that their records are accurate and up-to-date and that they are not given treatments that place them at unnecessary risk.43 Examples of the successful implementation of personal health records include HealthConnectOnline in the US, where patients can access and update their medical records and manage treatments online (http://www.healthconnectonline.com/). A similar service has been introduced in some parts of Europe (including Germany, Switzerland, Austria and Bulgaria) under the name LifeSensor 337 (https://www.lifesensor.com/us/us/), which is a web-based personal health record accessible for both health professionals and patients. Patients in England can use HealthSpace (https://www.healthspace.nhs.uk/visitor/default.aspx; see Chapter 3), a web-based health organiser. Although some concerns have been voiced regarding the accuracy of personal health records and the limited usage to-date (in England),44 their potential to help improve delivery of care is increasingly being recognised, particularly if their introduction can be aligned better with the needs of patients and care pathways adopted by professionals. In the context of patients with known drug allergies, they can provide a valuable additional safety net for use in combination with other established and emerging technologies discussed above. Which of the approaches or combinations of approaches discussed will ultimately prove most effective in reducing instances of repeat exposure in patients with known drug allergies remains to be determined. Given the (relatively) strong empirical evidence of their impact on relevant surrogate markers, it is important that ePrescribing systems are considered for implementation, particularly in hospital settings to help reduce accidental repeat exposure. Similarly, it is anticipated that patient-managed personal electronic records will in due course occupy a more central position in the delivery of care.45 Amongst the other emerging technologies discussed, RFID chips and biometrics hold potential for the integration of allergy information across care settings. But the lack of formal evaluations of the effectiveness of such devices with regard to patients with known drug allergies, their high capital costs and the considerable training needs of healthcare staff and patients may slow developments in this area. Also, ethical issues, including the accessibility of confidential information that patients carry with them at all times may prove problematic. 338 12.3 Implications for practice, policy and research There is an increasing body of epidemiological work indicating that adverse drug reactions due to known drug allergies are relatively common and potentially preventable. Alongside the need for increased research into investigating the underlying mechanisms and processes involved in allergic reactions to drugs, and the accompanying need to develop improved screening and diagnostic tests,4 there is, we believe, the pressing need to investigate and, where found to be effective, implement systems-based approaches to reducing these events into routine care. We have discussed existing and emerging information technology-based interventions, which are proven and have the potential to prove successful in minimising the risk of reexposure. ePrescribing systems with carefully-refined alerts are likely to continue to play an important role in general practice and the wider use of this technology should also be considered in hospital practice, albeit within an evaluative context. Newer technological developments in this area as bar-codes in wristbands and Medic-Alert bracelets, RFID chips, biometrics and patientmanaged electronic records all have potential, but the evidence-base is as yet too immature to contemplate their widespread deployment. There is thus an urgent need to evaluate the effectiveness of these newer technologies in the context of managing patients with a history of drug allergy.46 Whilst these developments are important, they do not detract from the onus of responsibility on the prescribing and administering clinician, and the patient who agrees to take the drug. Acknowledgements: This chapter is based on our publication: Cresswell KM, Sheikh A. Information technology-based approaches to reducing repeat drug exposure in patients with known drug allergies. J Allergy Clin Immunol 2008; 121: 1112-17. Permission to reproduce this material has been applied for. 339 References 1. Adkinson NF, Jr., Friedmann PS, Pongracic JA. Drug allergy. In: Holgate ST, Church MK, Lichtenstein LM, editors. Allergy. 3rd ed. London: Mosby Elsevier; 2006: 157-166. 2. Vervloet D, Durham S. ABC of allergies: Adverse reactions to drugs. BMJ 1998; 316: 1511-14. 3. Coombs RRA, Gell PGH. Classification of allergic reactions responsible for drug hypersensitivity reactions. In: Coombs RRA, Gell PGH, editors. Clinical Aspects of Immunology. Philadelphia, PA: Davis; 1968: 575-96. 4. Adkinson NF, Jr., Essayan D, Gruchalla R, Haggerty H, Kawabata T, Sandler JD et al. Task force report: future research needs for the prevention and management of immune-mediated drug hypersensitivity reactions. J Allergy Clin Immunol 2002; 109: S461-78. 5. Monitoring the safety and quality of medicines: the Yellow Card Scheme. London: Medicines and Healthcare Product Regulatory Agency. Available from: http://www.mhra.gov.uk/index.htm (last accessed 30/12/10) 6. The National Reporting and Learning Service. Available from http://www.nrls.npsa.nhs.uk/report-a-patient-safety-incident/ (last accessed 30/12/10). 7. Lesar TS, Briceland LL, Delcoure K, Parmalee JC, Masta-Gornic V, Pohl H. Medication prescribing errors in a teaching hospital. JAMA 1990; 263: 2329-34. 8. Lesar TS, Briceland L, Stein DS. Factors related to errors in medication prescribing. JAMA 1997; 277: 312-17. 9. Leape LL, Bates DW, Cullen DJ, Cooper J, Demonaco HJ, Gallivan T, et al. Systems analysis of adverse drug events. ADE Prevention Study Group. JAMA 1995; 274: 35-43. 10. Gandhi TK, Weingart SN, Borus J, Seger AC, Peterson J, Burdick E. Adverse drug events in ambulatory care. N Engl J Med 2003; 348: 1556-64. 11. National Patient Safety Agency. Safety in Doses: Medication Safety Incidents in the NHS. London: National Patient Safety Agency, 2007. 12. Chief Pharmaceutical Officer. Building a Safer NHS for Patients - Improving Medication Safety. London: Department of Health, 2004. 13. Bernstein I, Gruchalla RS, Lee R, Nicklas R, Dykewicz M. Executive summary of disease management of drug hypersensitivity: a practice parameter. Joint Task Force on Practice Parameters, the American Academy of Allergy, Asthma and Immunology, the American Academy of Allergy, Asthma and Immunology, and the Joint Council of Allergy, Asthma and Immunology. Ann Allergy Asthma Immunol 1999; 83: 665-700. 14. Gomes RE, Pichler WJ, Demoly P, Aberer W, Frew AJ, de Weck A. The Drug Ambassador Project - the diversity of diagnostic procedures for drug allergy around Europe. Allergy Clin Immunol Int - J World Allergy Org 2005; 17: 9-18. 15. Cresswell KM, Sheikh A. Information technology-based approaches to reducing repeat drug exposure in patients with known drug allergies. J Allergy Clin Immunol 2008; 121: 1112-17. 16. Bates DW, Leape LL, Cullen DJ, Laird N, Petersen LA, Teich JM et al. Effect of computerized physician order entry and a team intervention on prevention of serious medication errors. JAMA 1998; 280: 1311-16. 17. Evans RS, Pestotnik SL, Classen DC, Clemmer TP, Weaver LK, Orme JF et al. A computer-assisted management program for antibiotics and other anti-infective agents. N Engl J Med 1998; 338: 232-38. 340 18. Morris CJ, Savelyich BSP, Avery AJ, Cantrill JA, Sheikh A. Patient safety features of clinical computer systems: questionnaire survey of GP views. Qual Saf Health Care 2005; 14: 164-68. 19. Hsieh TC, Kuperman GJ, Jaggi T, Hojnowski-Diaz P, Fiskio J, Williams DH et al. Characteristics and consequences of drug allergy alert overrides in a computerized physician order entry system. J Am Med Inform Assoc 2004; 11: 482-91. 20. Fernando B, Savelyich BSP, Avery AJ, Sheikh A, Bainbridge M, Horsfield P, et al. Prescribing safety features of general practice computer systems: evaluation using simulated test cases. BMJ 2004; 328: 1171-72. 21. Kuperman GJ, Gandhi TK, Bates DW. Effective drug-allergy checking: methodological and operational issues. J Biomed Inform 2003; 36: 70-79. 22. Franklin BD, O'Grady K, Donyai P, Jacklin A, Barber N. The impact of a closedloop electronic prescribing and administration system on prescribing errors, administration errors and staff time: a before-and-after study. Qual Saf Health Care 2007; 16: 279-84. 23. Turner CL, Casbard AC, Murphy MF. Barcode technology: its role in increasing the safety of blood transfusion. Transfusion 2003; 43: 1200-09. 24. Marconi M, Langeberg AF, Sirchia G, Sandler SG. Improving transfusion safety by electronic identification of patients, blood samples, and blood units. Immunohematol 2000; 16: 82-85. 25. National Patient Safety Agency. Standardising Wristbands Improves Patient Safety. 2007. London: NPSA, 2007. 26. McDonald CJ. computerization can create safety hazards: a bar-coding near miss. Ann Intern Med 2006; 144: 510-16. 27. Patterson ES, Cook RI, Render ML. Improving patient safety by identifying side effects from introducing bar coding in medication administration. J Am Med Inform Assoc 2002; 9: 540-53. 28. Siminerio L, Clougherty M, Gilliland A, Kelly K. Evaluating children with diabetes and their parent's preparedness and involvement in diabetes care for the school. Diabetes 2000; 49: A173. 29. PDC. RFID System: Jacobi Medical Center. Available from http://www.pdcorp.com/en-us/healthcare/case-study-jacobi-medical-center.html (last accessed 31/12/10) 30. Dalton J, Ippolito C, Poncet I, Rossini S; 2005. Using RFID Technologies to Reduce Blood Transfusion Errors. White Paper by Intel Corporation, Autentica, Cisco Systems, and San Raffaele Hospital. Available from http://www.cisco.com/web/IT/local_offices/case_history/rfid_in_blood_transfusion s_final.pdf (last accessed 28/12/10) 31. Wessel R. German Clinic Uses RFID to Track Blood. Available from http://www.rfidjournal.com/article/articleview/2169/1/1/. (last accessed 30/12/10) 32. Dzik S. Radio frequency identification for prevention of bedside errors. Transfusion 2007; 47: 125S-29S. 33. Precision Dynamics Corporation. RFID Technology Integration - 100% Accuracy in Patient Identification; 2005. Available from http://www.pdcorp.com/enus/healthcare/case-study-chang-gung-hospital.html (last accessed 30/12/10) 34. Goddard K, Shabestari O, Kay JD, Roudsari A. Formatively evaluating the importance of different aspects of an electronic blood transfusion system from the end users' point of view: a questionnaire study. Stud Health Technol Inform 2010;160: 1211-15. 35. Bacheldor B. Tags Track Surgical Patients at Birmingham Heartlands Hospital; 2007. Available from http://www.rfidjournal.com/article/articleview/3222/2/1/ (last accessed 30/12/10). 341 36. Medical News Today. Horizon Blue Cross Blue Shield Of New Jersey To Sponsor Two-Year Pilot With Hackensack University Medical Center; 2006. Available from http://www.medicalnewstoday.com/articles/47480.php (last accessed 30/12/10) 37. More RFID. VeriChip Corporation and Alzheimer's Community Care Officially Launch VeriMed Patient Identification Project on August 28; 2007. Available from http://www.morerfid.com/details.php?subdetail=Report&action=details&report_id= 3442 (last accessed 30/12/10). 38. Schmitz PE. Biometrics in Europe. Available from: http://www.libertysecurity.org/IMG/pdf/Trend_Report_2006.pdf (last accessed 31/12/10) 39. Kohl G. As records go digital, biometrics integrates into healthcare sector. Available from: http://www.securityinfowatch.com/online/Biometrics/As-RecordsGo-Digital--Biometrics-Integrates-into-Healthcare-Sector/8706SIW417 (last accessed 30/12/10). 40. Adwoa Serwaa-Bonsu,1 Abraham J. Herbst,2* Georges Reniers,3,4 Wilfred Ijaa,5 Benjamin Clark,3 Chodziwadziwa Kabudula,3 and Osman Sankoh6. First experiences in the implementation of biometric technology to link data from Health and Demographic Surveillance Systems with health facility data. Glob Health Action 2010; 3: 10.3402/gha.v3i0.2120 41. Head B. The eyes have it - Identification using biometrics is finding a range of applications. Available from: http://cio.co.nz/cio.nsf/depth/52FF4B475BD4DAFACC2573540082A2C7 (last accessed 30/12/10) 42. E Health Insider. Patient information - a powerful prescription; 2006. Available from: http://www.e-health-insider.com/Features/item.cfm?docId=86 (last accessed 30/12/10) 43. Cohen MR. Medication Errors: Causes, Prevention and Risk Management. Sudbury, MA: Jones and Bartlett Publishers, 2000. 44. Greenhalgh T, Hinder S, Stramer K, Bratan T, Russell J. Adoption, non-adoption, and abandonment of a personal electronic health record: case study of HealthSpace. BMJ 2010; 341: c5814. 45. Pagliari C, Detmer D, Singleton P. Potential of electronic personal health records. BMJ 2007; 335: 330-33. 46. Catwell L, Sheikh A. Evaluating eHealth interventions: the need for continuous systemic evaluation. PLoS Med 2009; 6: e1000126. 342 Chapter 13 Telehealthcare Summary • Current systems of healthcare are inadequate to deal with rapidly increasing populations, many of whom have one or more long-term conditions. • Telehealthcare can be defined as “personalised healthcare delivered over a distance.” It involves transfer of data from the patient to the professional using information and communication technologies and personalised feedback from the professional to the patient. • Whilst potentially attractive, these new models of care can however prove very costly and may also cause a considerable change to consultation dynamics and workflow processes; these should therefore be introduced only after a careful assessment of effectiveness, costeffectiveness and safety considerations. • Most patients have in general a positive view of telehealthcare, so long as it is offered in addition to, rather than as a substitution for, the faceto-face consultation. Qualitative studies have highlighted that the needs of the carer should also be taken in account when designing telehealthcare systems. • Overall, for people with long-term conditions such as asthma, COPD and diabetes, there is now encouraging evidence that telehealthcare can reduce hospital admissions in those with more severe disease without increasing mortality; the evidence in relation to improving quality of life in those with milder disease is less encouraging. • There is a need for more cost-effectiveness studies which take into account the full range of perspectives i.e. that of the patient, healthcare services and society. • Potential pitfalls of telehealthcare include user interface issues such as older people not understanding how to operate systems, technical breakdown, and safety concerns such as data loss and breach of confidentiality. 343 Summary contdinued… • The use of telehealthcare often results in a reshaping of the doctorpatient relationship and attention should be paid to the opportunities for humanising this interaction. Workflows also need to be considered to minimise the risk of unintended disruptions to normal working routines. 13.1 Introduction Telehealthcare has been used by healthcare professionals since the invention of the telephone over 100 years ago, in Canada. Whenever practitioner and patient are not physically present in the same place and some form of telecommunications technology is used to enable the consultation, an action of “telehealthcare” may be considered to have taken place. What is new is the plurality of devices that now facilitate telehealthcare consultations. Also new, is the increasing interest on the part of policymakers who now tend to see telehealthcare implementations as central to their drive to modernise health systems.1 This is because ageing populations and the related inexorable increase in the numbers of people living for extended periods of time with long-term conditions is placing a major financial and staffing burden on health systems. Older people, have need of more medical and social support to ensure that they are functioning to their maximum potential. This change in demography has increased pressure on health and social care services to manage more patients with fewer resources. Health systems internationally are looking to solutions such as telehealthcare in order to save time and money and improve efficiency. Telehealthcare providers often promise that their products will do exactly this:2 however, it is necessary to examine the evidence of the effectiveness of telehealthcare, its risks and benefits and what is known regarding its implementation, before drawing such conclusions. Telehealthcare is anticipated to have other benefits as well, such as increasing patient satisfaction and even improving health outcomes. 344 This chapter aims to provide an overview of telehealthcare, and the studies undertaken so far which demonstrate the contexts and settings in which it is likely to represent a useful development. There then follows a discussion of policy, practice and research considerations in relation to telehealthcare. The important question of how best to implement telehealthcare is discussed in more detail in Chapter 17. The case study on telehealthcare in the context of asthma in the following chapter (Chapter 14), provides an opportunity for a much more detailed discussion of the opportunities and risks associated with use of telehealthcare in the context of long-term conditions, which is a key area of policy focus. 13.2 Definition, description and scope for use 13.2.1 Definition There are a multiplicity of definitions of telehealthcare and related terms. Here we discuss what we mean by the term “telehealthcare” and what is meant by related terms, telehealth, telemedicine, telecare etc. Term Origin Definition Telecare Barlow et al ‘The use of communications technology to 20073 provide health and social care directly to the user (patient). This excludes the exchange of information solely between professionals, generally for diagnosis or referral.’ Telehealth Glueckauf et ‘The al 20024 use of telecommunications and information technologies to provide access to health information and services across a geographical limited to) distance, including consultation, (but not assessment, intervention, and health maintenance.’ Telemedicine Commission (1) of European ‘The provision of health care services, through the the use of information and communications technology (ICT), in situations where the health Communities professional and the patient (or two health 345 COM professionals) are not in the same location. It (2008)6895 involves secure transmission of medical data and information through text sound, images or other forms needed for the prevention, diagnosis, treatment and follow-up of patients.’ Telemedicine Sood et al ‘Telemedicine being a subset of telehealth, (2) 20076 uses communications networks for delivery of healthcare services and medical education from one geographical location to another, primarily to address challenges like uneven distribution and shortage of infrastructural and human resources.’ Home Paré et al7 ‘An automated process for the transmission of telemonitoring 2007 data on a patient’s health status from home to the respective health care setting.’ Table 13.1 Some examples of definitions of telehealthcare and related terms We prefer to use the term “telehealthcare”, rather than “telehealth” or “telecare”, “telemedicine” or “home telemonitoring” as this incorporates both the health or medical element of the interaction and the caring element whether nursing care or social care. The term “telehealthcare” allows the interaction to contain many elements without role-based delineations. One might expect “telemedicine” to be delivered by a doctor, telecare by a carer, a nurse or a social worker and telenursing by a nurse. Telehealthcare is a much more flexible and encompassing term and so is preferred. In this sense, “telehealthcare” is a pragmatic, rather than an academic term. It synthesises across the health and social care boundary and implies the use of distance communication technologies for supporting the care of people with long-term conditions in an interactive manner. 346 Telehealthcare does, however, bracket some things outwith its definition. Telehealthcare is not the use of distance technologies for inter-professional communication, (sometimes described as business-to-business or B2B,communication–in contrast to business-to-consumer or B2C) or passive information provision (as in traditional online health education or marketing tools). Used in this sense, “telehealthcare” thus refers to “the provision of personalised healthcare over a distance”.8 It has the three following essential components:9 1. Information obtained from the individual, whether voice, video, electrocardiography, oxygen saturation or other similar data, that provides information on their condition. 2. Electronic transfer of such information–over a distance–to a healthcare professional. 3. Personalised feedback tailored to the individual from a healthcare professional who exercises their clinical skills and judgement. “Telehealthcare” is the provision of “healthcare”, “over a distance.” “Over a distance” implies the delivery of healthcare by means of the use of some tool of distance communication, which operates without the simultaneous physical, geographical presence of the two or more participants in the healthcare interaction. The postal service, although a means of communication at a distance does not use information technology (IT), but relies on another human acting as a relay messenger and therefore it is conventional to exclude the postal service from discussions on telehealthcare. The telephone, sometimes referred to in the literature as the “Plain Old Telephone System (POTS)”, to emphasise that telehealthcare is a novel employment of what is now an established technology; provides an often reliable means of providing telehealthcare. In addition, the Internet; email; mobile networks; radio transmission and other wired and wireless networks; 347 short message service (SMS), i.e. text messaging by phone and instant messaging over the Internet are all included in telehealthcare interventions. Video-conferencing allows the addition of the visual element to the healthcare interaction. 13.2.2 Description of use The most common model of telehealthcare is shown in the diagram below: data flows firstly from the patient to the healthcare professional at a distance by means of IT facilitating communication across a distance; the patient then receives tailored advice. Data Patient: At home, Work, Care home, In a car, • • • • • • • Telephone Mobile phone Email Internet Video-conferencing Text Electronic monitoring devices such as blood glucose monitor, peak expiratory flow meter which relay data via a modem • Daily questionnaires Healthcare professional in distant office or hospital Tailored advice Figure 13.1 Diagrammatic representation of telehealthcare with flow of information from the patient to professional across a distance with personalised feedback Telehealthcare can be delivered by both synchronous and asynchronous (i.e. store-and-forward) technologies (see Figure 13.1). For example, telephone and video-conferencing enable consultations in real time, whereas 348 asynchronous communication includes, for example, storing two weeks of spirometry results in a batch and then forwarding these on, via a modem, to a healthcare provider who responds by email or telephone. 13.2.3 Scope for use Telehealthcare has been piloted in many different situations, however, it would seem that its greatest potential is in the management of chronic conditions away from the settings of traditional healthcare such as general practice or the hospital or some other clinic. It thus enables patients to be seen and receive clinical input without incurring the time and cost of travel. Even more important, the inconvenience of travel is avoided–this is especially pertinent when considering future population demographics with an increased proportion of frail individuals suffering from more that one long-term condition. Such individuals may have several appointments a week with different hospital departments for different healthcare needs. They may have mobility and memory issues, which render them vulnerable when travelling and increase the time it takes them to complete their normal activities of daily living even before one takes into account the time required for healthcare, physiotherapy, podiatry etc. Telehealthcare to assist such people to receive their healthcare in their home or a familiar local clinic setting is potentially very welcome from their perspective. Similarly, people living in remote and rural areas distant from the nearest setting of healthcare provision can use telehealthcare to enable consultations from their own home.10 And where there is limited specialist coverage for rare diseases, for example in paediatrics, expertise can be more easily accessed by patients. In terms of the definition above we are concentrating on telehealthcare as represented in the systematic review (SR) literature, which focuses on telehealthcare in relation to long-term conditions. Although access to emergency services such as NHS Direct and 999 calls fall within our definition of telehealthcare, studies on these issues have not as yet been summarised in the SR literature, hence they will not be considered further in 349 this chapter. Before moving on, it is important to note that this represents an important gap in the evidence base for health systems. We do however include the use of the telephone for delivery of telehealthcare in non-triage situations. Our focus on long-term conditions is also justified in terms of the international policy context. For example, the 2008 Darzi report1 emphasised the need to concentrate resources on planning for an increasingly elderly population with increased frailty and multiple long-term conditions. Darzi was very clear that there may well be untapped potential in telehealthcare with regard to the more efficient and cost-effective management of limited resources to supply health and social care for this population. With the change to the coalition government, policies,11 such as decentralisation, increased emphasis on selfcare and driving-up efficiency are to an extent predicated on telehealthcare delivering on some of its potential benefits. There may however be a danger of adopting technology for its own sake, or because it is well-marketed. 13.3 Theoretical benefits and risks 13.3.1 Benefits In theory, telehealthcare can have both preventive and supportive roles. An individual with a long-term condition may be well enough to manage their activities of daily living with minimal support, potentially delivered via telehealthcare. In the event of an acute deterioration in their condition, telehealthcare enables swift intervention to prevent a prolonged or deeper deterioration and dependence. This rapid intervention is enabled by telehealthcare’s potential to deliver an early warning to the healthcare professionals to enable them to take the necessary action. Other potential uses of telehealthcare fall into the category of “upstream” interventions or, in medical terminology, “primary prevention” including for example improved monitoring of glucose and of blood pressure to prevent heart attacks and strokes. Secondary prevention is improved by ensuring medical management is taken as prescribed. 350 The potential supportive role of telehealthcare is demonstrated by the help people can get in interpreting their day-to-day symptoms (e.g. pain), readings (such as blood pressure) and test results (e.g. blood glucose), all of which may be improved and the patient’s anxiety alleviated when such problems are discussed with healthcare professionals. In the absence of regular clear guidance from health professionals regular blood glucose monitoring in Type II diabetes leads to feelings of guilt and self-blame especially in women. Patients attribute the monitoring readings to whether they have been “bad” or “good” with their diet and ignore the contribution of factors such as exercise and medication. Without supportive, ideally tailored, education “monitoring fatigue” sets in and patients abandon monitoring. Patients question the point of monitoring if they are not seeing positive changes as a result of it.12 In addition, in the event of a deterioration in symptoms or readings, the need for an increase in the intensity or frequency of clinical contact may be enabled without the upheaval and stresses generated by a hospital admission. The risks of cross-infection or loss of function which may also be associated with a bed-bound hospital admission are also avoided. This type of management at home may be more in keeping with the patient and their family’s own wishes for privacy and dignity at a time of vulnerability. It is also hoped that telehealthcare will enable care to be tailored to the individual. There is a policy push towards more personalised care and the use of telehealthcare may enable patients to have their individual needs assessed and then met in a more personalised manner.13 13.3.2 Risks In theory, there is a risk with telehealthcare that the level of care will be underestimated or over-estimated. Over-estimating the level of care that is required is more likely during a preventive phase of management. There are risks that patients find the care obtrusive, imposing on their independence with reminders regarding what to eat, what to drink, what exercises to do and when to take medication. This may undermine autonomy to vulnerable groups of people.14 Smart homes which detect people’s movements from room to room might be said to be looking out for elderly patients in case of falling, 351 however, some people feel as though their privacy is compromised and they are “under surveillance”.15 16 Or, alternatively, there is a risk of medicalising the spectrum of normality and reducing an individual to a “patient” in a “sick role” when they were coping quite well alone and independently or with traditional support. Under-estimating the level of care is also possible, for example, when a patient with an acute exacerbation of illness is managed at home where traditionally they would be managed in hospital. If the illness is more unstable and less predictable than at first thought and the patient lives a long way from more comprehensive medical care, then there are potential risks to patient safety. However, competent patients should be encouraged to weigh up these benefits and risks and decide for themselves whether they would rather be at home or in hospital according to their individual priorities. Staff may have to spend additional time learning to use the telehealthcare technology. Staff users may then find that the time required to operate the new system on a day-to-day basis may be greater than the systems and processes that were replaced. Also during replacement staff will have to duplicate their workload as the two systems run in parallel. There is a risk that staff may become demoralised and overburdened when trying to implement an innovation that is not “fit-for-purpose”. Similarly, if the telehealthcare operates as planned and allows increased access to staff by patients staff may become overwhelmed with the increased demands on their time as without further resources it may be difficult for them to provide an enhanced service. There is also the risk that the systems and processes of care will need to be redesigned to incorporate the technology which can lead to realignment of staff roles and political problems. There are technical risks: of breakdown, of unreliable telephone/network connection and of poor interoperability.13 352 Vast amounts of data will be collected to enable telehealthcare. On the one hand, if granted access, researchers may be able to interpret these data and increase understanding of diseases to the benefit of society. On the other hand, other groups may request access to these data for example in the United States (US) there is a request to enable targeted marketing by commercial companies for monetary gain, without the explicit consent of the patients. Specifically, insurance and healthcare marketing may be anticipated to be interested in such databases.17 Healthcare is a safety critical infrastructure industry, meaning that its smooth running is imperative to the integrity of the nation. If vast numbers of vulnerable people are living at home dependent on telehealthcare and there is a sudden problem with the telephone or mobile network which links them, then services will no longer be at a scale that can cope in looking after such people. Similarly if the Internet server has a problem, there is a power failure or a cyber-attack many people may be left without services they rely on. This possibility demands careful contingency planning, including planning for more than one problem occurring at a single point in time.17 13.4 Empirically demonstrated benefits and risks We identified 41 relevant SRs3 7 18-56 investigating use of a variety of telehealthcare interventions in different contexts and settings. These ranged from connections between hospital specialist services and patients’ homes to Internet discussion groups which remotely connected several peer patients and a professional facilitator who may be based in the community or in a specialist service, and included primary studies employing different designs. A fuller description of these SRs and our assessment of their quality is available in Appendices 4 and 5. Below, we provide a synopsis of the main findings to emerge from this body of work. 13.4.1 Benefits Satisfaction There is evidence that patients are largely satisfied with telehealthcare. For example, patients reported high levels of satisfaction and empowerment in 353 two overlapping systematic reviews of telehealthcare by the same author.7 57 Satisfaction tended to be reported on a Likert (or similar) scale.36 37 46 However, it may be that qualitative evidence gives greater insights regarding how to achieve and maintain satisfaction. For example, participants in a focus group discussed telehealthcare and agreed that it remained acceptable as long as it was not compulsory. They wanted to still be able to have the option of face-to-face care.9 A way forward may be to have the first visit/assessment face-to-face with the option of telehealthcare follow-up. Effectiveness The effectiveness of telehealthcare is often assumed by industry and policymakers. However, when scrutinised, research on effectiveness depends very much on the context of the introduction of a specific telehealthcare system. It is thus crucial that the aim of introducing the telehealthcare is clear from the start.58 For example, is the aim to widen access, improve clinical endpoints, aid the early detection of disease exacerbations and reduce the risk of hospital admissions, reduce mortality, or is it to reduce the degree of dependency in old age? These aims are not necessarily mutually exclusive, but agreement on why telehealthcare has been introduced, followed by focused implementation, and then evaluation to see if the specific need has been addressed should help to determine if these aims have been realised. Improving access Access to specialist care in rural locations was enabled by telehealthcare which improved the quality of care, especially for rare diseases.38 Teleradiology improved access to radiological services, saving transfer time, cost and inconvenience.38 Likewise access to renal dialysis services was improved with telehealthcare, especially as patients require several treatments a week. There was also improved access to cardiology, colposcopy, neuropsychology, minor injuries, dermatology and nutrition services.38 Another review found that using the web as the medium of delivery improved access to the benefits of tailored therapies for mental health problems regardless of geographical and temporal barriers.43 354 Similarly, using the telephone improved accessibility to primary care asthma reviews in one study, with the authors concluding that it was possible to review almost double the number of patients via telephone in comparison to face-to-face assessments.59 This telephone-based approach, with limited start-up costs, was also found to be cost-effective.60 Further supporting evidence comes from a review, which reported that 20% of included studies had found an improvement in access.31 And so overall it appears that telehealthcare can, in carefully selected contexts, improve access to care. There is some suggestive evidence that access can also be enhanced for carers. For example, peer support and relief of social isolation have been reported, both for people with mental health difficulties and for carers. Improving clinical endpoints The evidence is less clear cut regarding the improvement of clinical endpoints. Firstly, attention should be drawn to the fact that many of the clinical endpoints reported in reviews are surrogate markers rather than “hard” clinical outcomes such as mortality. For example, telehealthcare may be shown to increase adherence to therapy and although it is assumed that this is beneficial, studies rarely mentioned adverse events and side-effects. Similarly, although slight improvements in blood pressure (BP),49 glycated haemoglobin (HbA1c)50 and lipid profiles have been demonstrated, studies were neither powerful, nor long enough to demonstrate a translation of these results into fewer heart attacks, strokes or other relevant cardiovascular morbidity. Despite this difficulty some improvements were demonstrated, but not consistently. One difficulty is in deciding whether the comparisons were valid. Many studies used a before-and-after design in which it is possible that only regression to the mean is being detected or the natural fluctuations of the illness. Where comparisons were made across telehealthcare and control groups, the groups were often not significantly different at the end of the 355 study, which may be taken to mean that patients were not significantly worse off for receiving telehealthcare, however, not significantly better either. In terms of considering whether a comparison is valid it is worth pondering on what is the best control. For example, for depression, one study compared usual care with cognitive behavioural therapy via telephone. However, it was not stipulated what the control consisted of. Did they receive any cognitive behavioural therapy, or perhaps a lesser form of counselling? How often? From whom? Did they receive medication? Were they given a leaflet? These variables all need to be considered in deciding whether telehealthcare has helped or hindered a patient’s recovery or deterioration. A wide range of disease specific clinical endpoints were studied, however, they each need to be considered in context given the caveats above. Even then, the literature is somewhat inconsistent. It may be anticipated that increased frequency and intensity of therapy delivered by whatever means whether via telehealthcare or face-to-face would correlate with improvement. However, one then needs to factor in side effects, quality of life, dependence and loss of self-efficacy to complete the complex picture. In this review, overall answers remained elusive. Reducing mortality There was very little data on mortality. A meta-analysis of telehealthcare interventions in coronary heart disease found a non-significant lower all cause mortality compared with controls (RR 0.70, 95%CI 0.45, 1.1).49 A review of studies in patients with chronic obstructive pulmonary disease (COPD) found increased mortality in the telehealthcare arm; however, we believe this is due to a data entry error and this is discussed in our forthcoming Cochrane review. Reducing risk of hospital admission Our systematic review (see Chapter 14) identified a number of telehealthcare trials in asthma.45 It showed that hospital admissions can be reduced in 356 carefully selected patients who have severe asthma and/or have recently been admitted to hospital and are consequently at high risk of re-admission.45 A reduction in the number of patients who were hospitalised, the number of hospitalisations and the length of stay associated with these admissions was also found with telehealthcare in diabetes.50 Similarly, monitoring of chronic heart failure and pacemaker patients reduced hospital readmissions and emergency department visits at a similar cost to conventional interventions.38 Therefore in selected high risk subgroups with appropriate conditions it is possible to reduce hospital admissions. Cost-effectiveness Studies which considered cost savings generally had methodological flaws in that they looked only at list prices and did not calculate opportunity costs. In general, cost savings were analysed from the perspective of the healthcare body, with the greatest saving coming from the prevention of hospital admission.55 There were very few cost-effectiveness studies reported in terms of Quality Adjusted Life Years(QALYs).56 Interventions which may have proved cost-effective over the long term were generally not followed up for long enough to derive any reliable estimates.45 Overall, in the absence of a convincing impact on mortality or morbidity, at this moment in time, it appears that telehealthcare may only prove cost-effective in carefully considered clinical situations in which there are substantial opportunities for reducing risk of premature death or hospitalisation. The relative costs of the telehealthcare intervention adopted is also relevant. 13.4.2 Risks Confidentiality There has been much publicity recently concerning the loss of confidential data from medical and other record systems.61 62 Such events become more likely if people’s sensitive medical data are available on networks and mobile storage devices which are not protected with encryption facilities. In addition, with the increased use of mobile phones, people often phone the doctor from 357 their place of work, or when in a public place, this can lead to problems with private conversations being overheard. There is some evidence to show that in situations in which doctors call patients unexpectedly, then there may be situations in which their privacy may at times be compromised through being overheard and that patients may be inclined to answer sensitive questions in “yes/no” terms.63 With telephone technology older patients with problems with vision and hearing may need a younger person to help to make a telephone call, this leads to problems of third party confidentiality, which are often ignored by professionals.63 The use of online technology has also been associated with significant breaches of confidentiality, for example, through the Kaiser Permanente Internet Patient Portal, when details of e-mail conversations between patients and their doctors were accidentally sent to multiple other patients.64 There needs to be reporting of such events so that organisational processes can respond to minimise repeat events and a “safety culture” can be encouraged in healthcare organisations. Human contact Contact with machines which look after people in place of human and family carers can result in loneliness, especially acute in those who have lost a spouse where there may be a risk of over-reliance on devices. Rather than empowering patients to become more active in looking after themselves, telehealthcare may result in an increase in the patient’s passivity.65 In addition, older people with long-term conditions, reduced dexterity, poor eyesight and deteriorating cognition sometimes have difficulty operating modern technology interfaces without help from a younger person.65 13.5 Implications for policy, practice and research 13.5.1 Policy It used to be that physical co-location was a prerequisite for the definition of a healthcare consultation according to various insurance companies within the US and the European Union (EU). This has changed and now the European Commission provides some guidance regarding the quality of the remote 358 healthcare encounter and the safety nets required.5 66 Within the UK, the Medical Defence Union however reminds doctors that they are contractually obliged to enable a physical examination where necessary.67 Until GP contracts catch up with the implications of telehealthcare, it seems unlikely that professional bodies such as the British Medical Association (BMA) will support the introduction of telehealthcare. Doctors and nurses need to ensure that they are professionally insured for any increased risk in caring for patients via telehealthcare. The General Medical Council (GMC) in the UK describes how best to carry out and document a remote prescription. Although this guidance was written with regard to telephone consultations, it can also be applied to email, Internet and video-consultations.68 Early signs are that the UK’s coalition government is enthusiastic regarding telehealthcare and the White paper ‘Equity and excellence: Liberating the NHS’11 and the consultation document ‘An Information Revolution’69 discuss the potential of telehealthcare in very optimistic terms. Its themes include the following: • Consortia of General Practitioners will be able to commission new ways of delivering care, including telehealthcare, in order to improve services and manage the increasing need for healthcare more efficiently. • Patients will have the opportunity to choose the providers of their health and geographical social care. This boundaries so will include flexibility regarding that, especially in the case of telehealthcare, there will be fewer restrictions. • Health and social care delivery will overlap to some degree depending on local needs, including a single telephone number for every kind of urgent health and social care. • Patients will have access to their own health records and be participants in the forming of the health record. They may also choose to share their health records with third parties such as charities, 359 advocacy organisations and information management service providers and interpreters. • Determination of central standards by the NHS Commissioning Board regarding record keeping, data sharing capabilities and efficiency of data transfer. This should enable locally commissioned systems to be interoperable. • An emphasis on new types of information such as quality standards for patients to help them make choices, for clinicians as feedback on their performance and for the public to hold the government to account regarding improved service delivery. • The use of personal health budgets and an emphasis on responsibility for increased self-care for long-term conditions. These over-arching themes will have an important impact on how quickly telehealthcare provision is scaled up. There is a strong policy drive but many aspects remain to be thought through. For example, if GP consortia working in partnership with local authorities are able to commission telehealthcare services according to their patient population, this will result in many different providers and technology packages. The government anticipates that competition between providers will drive innovation and lower prices for commissioners. However, the need for interoperability is crucial in order to enable many of the presumed benefits of telehealthcare (see Chapter 5) and this can only be delivered with careful technical specification during procurement procedures. The implications of our research on policy appear to be that using the telephone to deliver telehealthcare is the most mature technology. It is certainly the best understood in terms of the way in which it works and its potential shortcomings and challenges. Scaling up of different telehealthcare projects should be done with the uncertainties and expectancies of new technology, firmly in mind. The use of the telephones should be encouraged and the use of more complex telehealthcare equipment delayed until further large scale trial results are available. 360 13.5.2 Practice Introducing telehealthcare into practice involves a number of important considerations. Firstly, this review has found that the most robust evidence exists for telephone consultations and follow-up. This medium of telehealthcare has been around the longest and is the most well studied. It is cheap to provide and access for both patients and clinicians. It does not necessitate expensive new technology, but only uses reliable and widely available technology. For example, when telephones are used to provide asthma review appointments a greater proportion of the eligible patient population are reviewed and the intervention is cheaper than a face-to-face review and overall more cost-effective.70 Telephone follow-up after hospital discharge, initiated by a hospital team, has the potential to replace follow-up face-to-face visits and help address post-discharge problems.46 However, when telephone consultations were used as an alternative to acute general practice face-to-face consultation the evidence is that telephone consultations were shorter, presented fewer problems and included less data gathering, counselling and rapport-building than traditional face-to-face consultations.71 This comparison also found that doctors conducting telephone consultations were less likely to have gathered sufficient information to exclude important serious illnesses.71 A further study found that rather than replacing face-toface consultation the encounter was postponed as patients managed by telephone were 50% more likely to re-consult in the following two weeks and so an inefficiency is introduced into the workflow, although the precise reasons for this repeat consultation remain to be clarified.72 Such evidence casts doubt on whether to use telephone consulting for triage of acute emergencies. Patients have had little say in the widespread adoption of telephone consulting. There are differing contextual reasons for its use, including access targets, which, when combined with high demand for appointments and inflexible working practices, resulted in an increased use of telephone appointments in urban practices. Whereas, in rural practices, the telephone is used to overcome distances and maintain continuity of care.73 361 Alternatives to telephone appointments include workflows designed around a patient with a telehealthcare device which accesses directly to secondary care. These have important implications for practice. Consider, for example, a programme in which patients monitor their own weight, blood pressure and record their own electrocardiogram to transmit to a hospital-based specialist nurse. The nurse, in consultation with a cardiologist, provides tailored advice to the patient over the telephone on managing their heart failure.74 • There is a change in the role and responsibility of the nurse; they are involved increasingly in decision-making and so there is a need for training to reflect this. • The GP is somewhat marginalised as patients with increasingly complex management regimens, default to a trusted specialist relationship. • There is a need to maintain accountability, protocols and guidelines proliferate for case managers, describing how much they can undertake themselves and when to involve a senior medical practitioner. • There is an issue regarding multi-morbidity, when patients have more than one long term condition. Patients may find that they receive conflicting and complex advice from multiple specialists, however, their GPs are no longer well placed to guide them holistically when continuity of care is ruptured and they also provide predominantly telehealthcare care. 13.5.3 Research As can be seen from the summary of SRs on telehealthcare, there is much about this technology which is promising. However, many aspects of the technology remain poorly researched. This is accompanied by a pressure for implementation for the demographic reasons given at the beginning of this chapter. There is also pressure from commercial companies to implement their telehealthcare solutions while regulation remains minimal in order to maximise their opportunity to profit financially. 362 A recent independent industry report, ‘Healthcare without walls’75 proclaims that ‘telehealth can enable an incredible 40% reduction or more in the level of unplanned hospital admissions’. There is no citation for this figure and it does not mention whether this is an absolute or relative reduction. The brief summaries of evidence cited in the appendices are the only quantitative data included in this report. These data are presented out of context and without discussion of the limitations and accuracy of the figures. No randomised control trial evidence is cited. The report recommends a rapid broad strategic “push” for telehealthcare by addressing a business case to primary care providers before they become consortia of general practitioners. It is proposed that telehealthcare be adopted at scale by the NHS in order to share best practice and ensure efficient procurement and economies of scale. We feel that the note this paper strikes is very optimistic and somewhat premature. We do not know enough about telehealthcare models of care yet to anticipate that adoption at scale would not be an expensive catastrophe. The report from the Whole System Demonstrator Project76 is eagerly awaited this year (2011) to help to fill this knowledge gap. However, it is clear that telehealthcare should be studied in projects of greater size than the large number of published small studies e.g. 500+ participants before such widespread adoption goes ahead. In future, research needs to concentrate on large comparative studies between telehealthcare options and usual care controls. The studies need to fully describe and report both controls and interventions. The controls need to be real-world, usual-care controls in order to allow meaningful comparisons – to see if the introduction of telehealthcare would result in a significant improvement to care. Understandably, the validity of many studies is compromised by enhancements in the control arm; this does however make interpreting this evidence difficult. Such studies should be randomised controlled trials with follow-up of at least 6-12 months, preferably longer. It is also important that studies include meaningful clinical endpoints rather than surrogate and process measures, wherever possible. For example, the coronary heart disease trials need longer follow-up and comparison of the 363 incidence rates of cardiac and cardiovascular morbidity and readmission rates. Such clinical endpoints are necessary to allow clinicians to decide if telehealthcare does indeed improve outcomes for patients or whether there is a risk that the reduction in the intensity of care of telehealthcare causes more problems than it solves. One recent large study which fulfilled these criteria of large study population and meaningful clinical endpoints was a study of 1653 patients with heart failure.77 This followed a pilot study where a highly skilled and motivated specialist nurse case manager had help to reduce readmission rate by 44%.78 The authors were very concerned regarding the generalisability of this pilot. Their misgivings were confirmed in this large multicentre heart failure trial when they found no reduction in the risk of readmission or death from any cause with telehealthcare as compared with usual care. This study77 had overcome the methodological challenges listed above and provides a high quality contradiction to the systematic review on heart failure by Inglis et al., which pooled a number of lower quality studies and found a positive effect of telehealthcare.36 There is furthermore a need to find effective ways of safely and reliably integrating the vast amounts of data generated by telehealthcare initiatives into electronic patient records, rather than as stand-alone datasets. In addition, there needs to be improved study of the cost-effectiveness of telehealthcare. Accurate costs from different perspectives, including that of the healthcare provider, the patient and society should be combined with cost per QALY analysis to inform the incremental adoption of telehealthcare techniques into existing and adapted workflows. When resources are tight it is important that the costs of every new technology are substantially justified in terms of proven clinical benefit. 13.6 Conclusions Telehealthcare is widely anticipated as the future of healthcare internationally in order to relieve demographic pressure on healthcare services. There are 364 many potential benefits from “personalised healthcare, over a distance”, including tailoring advice, and reducing time and cost of travel. Access can be enabled for those in remote and rural areas and for frail and vulnerable populations, in particular, those suffering from long-term conditions. Support in managing long-term conditions should prevent deterioration and reduce the risk of hospital admission, which may in turn result in cost savings. However, will this lead to dependency on cold technology in place of warm human interaction? Or to underestimating the seriousness of an acute illness? Will systems be easy for staff and patients to operate and if so will access to healthcare be so easy as to overwhelm the ability of staff to respond? What of the data generated by the system and its related security and privacy issues? These are some of the more pressing questions raised. The SRs we identified have demonstrated that telehealthcare can have a measurable impact in certain specific contexts including long term conditions, patient assessment and social outcomes. Cost-effectiveness has not yet been demonstrated and further research in this area is urgently required. There is a very focused policy push for telehealthcare including modern digital technologies and the Internet, but the best evidence is for the telephone. The widespread introduction of telehealthcare into practice will engender major changes to the roles, responsibilities and workflows of healthcare organisations and some of these are discussed further in Chapter 17. Finally, more, longer term, larger research studies are required in order to address these uncertainties so that the potential benefits will be realised if telehealthcare is implemented. Without such studies, implementation of telehealthcare risks being expensive, inefficient, and even dangerous to vulnerable patients. Acknowledgments: This chapter is partially based on: McLean S, Protti D, Sheikh A. Telehealthcare for long-term conditions. BMJ (in press). Permission to reproduce this material has been granted. 365 References 1. Darzi A. High Quality Care for All: NHS Next Stage Review Final Report. Department of Health 2008 2. Cisco. United Healthgroup, Cisco Launch Connected Care, First National Telehealth Network. http://newsroom.cisco.com/dlls/2009/prod_071509b.html 2009 3. Barlow J, Singh D, Bayer S, Curry R. A systematic review of the benefits of home telecare for frail elderly people and those with long-term conditions. J Telemed Telecare 2007;13(4):172-9 4. Glueckauf R, Fritz S, Ecklund-Johnson E, Liss H, Dages P, Carney P. Videocounseling for rural teens with seizure disorders: Phase I findings. Rehabilitation Psychology 2002;47(49-72) 5. European-Commision. Communication from the commission to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the regions on telemedicine for the benefit of patients , healthcare systems and society. 2008:1-13 6. Sood S, Mbarika V, Jugoo S, Dookhy R, Doarn CR, Prakash N, et al. What is telemedicine? A collection of 104 peer-reviewed perspectives and theoretical underpinnings. Telemed J E Health 2007;13(5):573-90 7. Pare G, Jaana M, Sicotte C. Systematic Review of Home Telemonitoring for Chronic diseases: the evidence base. Journal of the American Medical Informatics Association 2007;14:269-277 8. McLean S, Sheikh A. Does telehealthcare offer a patient-centred way forward for the community based management of long-term respiratory disease? Primary Care Respiratory Journal 2009;18(3):125-126 9. Miller E. Solving the disjuncture between research and practice, Telehealth trends in the 21st Century. Health Policy 2007;82:133-41 10. Rygh EM, Hjortdahl P. Continuous and integrated health care services in rural areas. A literature study. Rural Remote Health 2007;7(3):766 11. DoH. Equity and excellence: Liberating the NS: Department of Health, 2010. 12. Peel E, Douglas M, Lawton J. Self monitoring of blood glucose in type 2 diabetes: longitudinal qualitative study of patients' perspectives. BMJ 2007;335:493-8 13. Goodwin N. The State of Telehealth and Telecare in the UK: Prospects for Integrated Care. Journal of Integrated Care 2010;18(6):3-10 14. Hensel BK, Demiris G, Courtney KL. Defining obtrusiveness in home telehealth technologies: a conceptual framework. J Am Med Inform Assoc 2006;13(4):428-31 15. Dinesen B, Nohr C, Andersen SK, Sejersen H, Toft E. Under surveillance, yet looked after: telehomecare as viewed by patients and their spouse/partners. Eur J Cardiovasc Nurs 2008;7(3):239-46 16. Hughes JC, Louw SJ. Electronic tagging of people with dementia who wander. BMJ 2002;325:847-8 17. Johnson CW. Politics and Patient Safety Don't Mix; On the Interaction between Patient Safety , Politics and Software Engineering. 18. Akesson KM, Saveman B-I, Nilsson G. Health care consumers' experiences of information communication technology--a summary of literature. International Journal of Medical Informatics 2007;76(9):633-45 19. Bailey Julia V, Murray E, Rait G, Mercer Catherine H, Morris Richard W, Peacock R, et al. Interactive computer-based interventions for sexual health 366 promotion. Cochrane Database of Systematic Reviews. Chichester, UK: John Wiley & Sons, Ltd, 2010. 20. Balas EA, Jaffrey F, Kuperman GJ, Boren SA, Brown GD, Pinciroli F, et al. Electronic communication with patients. Evaluation of distance medicine technology. JAMA 1997;278(2):152-9 21. Bartoli L, Zanaboni P, Masella C, Ursini N, x00F. Systematic review of telemedicine services for patients affected by chronic obstructive pulmonary disease (COPD). Telemedicine Journal & E-Health 2009;15(9):877-83 22. Bessell TL, McDonald S, Silagy CA, Anderson JN, Hiller JE, Sansom LN. Do Internet interventions for consumers cause more harm than good? A systematic review. Health Expectations 2002;5(1):28-37 23. Biem HJ, Turnell RW, D'Arcy C. Computer telephony: automated calls for medical care. Clinical & Investigative Medicine - Medecine Clinique et Experimentale 2003;26(5):259-68 24. Boren SA, Puchbauer AM, Williams F. Computerized prompting and feedback of diabetes care: a review of the literature. Journal of Diabetes Science & Technology 2009;3(4):944-50 25. Bowles KH, Baugh AC. Applying research evidence to optimize telehomecare. Journal of Cardiovascular Nursing 2007;22(1):5-15 26. Brebner JA, Brebner EM, Ruddick-Bracken H. Accident and emergency teleconsultation for primary care--a systematic review of technical feasibility, clinical effectiveness, cost effectiveness and level of local management. Journal of Telemedicine & Telecare 2006;12 Suppl 1:5-8 27. Bunn F, Byrne G, Kendall S. Telephone consultation and triage: effects on health care use and patient satisfaction. Cochrane Database of Systematic Reviews 2004(4):CD004180 28. Currell R, Urquhart C, Wainwright P, Lewis R. Telemedicine versus face-toface patient care:effects on professional practice and health care outcomes. . Cochrane Database of Systematic Reviews 2000(2) 29. Dellifraine J, Dansky K. Home-based telehealth: a review and meta-analysis. Journal of Telemedicine and Telecare 2008;14:62-66 30. Demiris G, Hensel BK. Technologies for an aging society: a systematic review of "smart home" applications. Yearbook of Medical Informatics 2008:33-40 31. Durrani H, Khoja S. A systematic review of the use of telehealth in Asian countries (Structured abstract). Journal of Telemedicine and Telecare, 2009:175-181. 32. Duvvuri V, Jianhhong W. Information and communication technology developments in asthma management: a systematic review. Indian Journal of Medical Science 2007;61(4):221-241 33. Eysenbach G, Powell J, Englesakis M, Rizo C, Stern A. Health related virtual communities and electronic support groups: systematic review of the effects of online peer to peer interactions. BMJ 2004;328(7449):1166 34. Hersh WR, Helfand M, Wallace JA, Kraemer DF, Patterson P, Shapiro S, et al. Clinical outcomes resulting from telemedicine interventions: a systematic review. BMC Medical Informatics & Decision Making 2001;1(5) 35. Hersh WR, Wallace JA, Paterson PK, Shapiro SE, Kraemer DF, Eilers GM. Telemedicine for the medicare population: Paediatric, obstetric and clinician indirect home interventions. Evid Rep Technol Assess (Summ) 2001;24 Suppl:1-32 36. Inglis Sally C, Clark Robyn A, McAlister Finlay A, Ball J, Lewinter C, Cullington D, et al. Structured telephone support or telemonitoring programmes for patients with chronic heart failure. Cochrane Database of Systematic Reviews. Chichester, UK: John Wiley & Sons, Ltd, 2010. 37. Jackson CL, Bolen S, Brancati FL, Batts-Turner ML, Gary TL. A systematic review of interactive computer-assisted technology in diabetes care. 367 Interactive information technology in diabetes care. Journal of General Internal Medicine 2006;21(2):105-10 38. Jennett P, Hall L, Hailey D, Ohinmaa A, Anderson C, Thomas R, et al. The socio-economic impact of telehealth: a systematic review. Journal of Telemedicine and Telecare 2003;9:311-320 39. Jones JF, Brennan PF. Telehealth interventions to improve clinical nursing of elders. Annual Review of Nursing Research 2002;20:293-322 40. Kairy D, Lehoux P, Vincent C, Visintin M. A systematic review of clinical outcomes, clinical process, healthcare utilization and costs associated with telerehabilitation (Structured abstract). Disability and Rehabilitation, 2009:427-447. 41. Leach L, Christensen H. A systematic review of telephone-based interventions for mental disorders. Journal of Telemedicine and Telecare 2006;12:122-129 42. Lehmann S, Domdey A, Bramesfeld A. [Telephone case management: is it beneficial for the care of depression patients in Germany? A systematic literature survey]. Gesundheitswesen;72(5):e33-7 43. Lustria MLA, Cortese J, Noar SM, Glueckauf RL. Computer-tailored health interventions delivered over the Web: review and analysis of key components. Patient Education & Counseling 2009;74(2):156-73 44. Martin S, Kelly G, Kernohan W, McCreight B, Nugent C. Smart home technologies for health and social care support. cochrane Database of Systematic Reviews 2008(4):Art. No. CD006412. DOI: 10.1002/14651858.CD006412.pub2 45. McLean S, Nurmatov U, Chandler D, Liu J, Pagliari C, Sheikh A. Telehealthcare for Asthma. cochrane Database of Systematic Reviews 2009(2):Art. No.: CD007717.DOI:10.1002/14651858.CD007717 46. Mistiaen P, Poot E. Telephone follow-up, initiated by a hospital-based health professional, for postdischarge problems in patients discharged from hospital to home. Cochrane Database Syst Rev 2006(4):CD004510 47. Montori VM, Helgemoe PK, Guyatt GH, Dean DS, Leung TW, Smith SA, et al. Telecare for patients with type 1 diabetes and inadequate glycemic control: a randomized controlled trial and meta-analysis (Provisional abstract). Diabetes Care, 2004:1088-1094. 48. Murray E, Burns J, See TS, Lai R, Nazareth I. Interactive Health Communication Applications for people with chronic disease. Cochrane Database of Systematic Reviews 2005(4):CD004274 49. Neubeck L, Redfern J, Fernandez R, Briffa T, Bauman A, Freedman S. Telehealth interventions for the secondary prevention of coronary heart disease: a systematic review. European Journal of Cardiovascular Prevention and Rehabilitation 2009;16(3):281-289 50. Polisena J, Tran K, Cimon K, Hutton B, McGill S, Palmer K. Home telehealth for diabetes management: a systematic review and meta analysis. Diabetes Obes Metab 2009;11(10):913-30 51. Polisena J, Tran K, Cimon K, Hutton B, McGill S, Palmer K, et al. Home telemonitoring for congestive heart failure a systematic review and metaanalysis. Journal of Telemedicine and Telecare 2010;16:68-76 52. Revere D, Dunbar PJ. Review of computer-generated outpatient health behavior interventions: clinical encounters "in absentia". Journal of the American Medical Informatics Association 2001;8(1):62-79 53. Roine R, Ohinmaa A, Hailey D. Assessing telemedicine: a systematic review of the literature. CMAJ Canadian Medical Association Journal 2001;165(6):765-71 54. Samoocha D, Bruinvels DJ, Elbers NA, Anema JR, van der Beek AJ. Effectiveness of web-based interventions on patient empowerment: a 368 systematic review and meta-analysis. Journal of Medical Internet Research;12(2):e23 55. Stinson J, Wilson R, Gill N, Yamada J, Holt J. A Systematic Review of Internet-based Self Management Intervetions for Youth with Health Conditions. Journal of Pediatric Psychology 2008;34(5):495-510 56. Whitten P, Mair F, Haycox A, May C, Williams T, Hellmich S. Systematic review of cost effectiveness studies of telemedicine interventions. BMJ 2002;324:1434-7 57. Pare G, Moqadem K, Pineau G, St-Hilaire C. Clinical effects of home telemonitoring in the context of diabetes, asthma, heart failure and hypertension: a systematic review. Journal of Medical Internet Research;12(2):e21 58. Catwell L, Sheikh A. Evaluating eHealth Interventions: The Need for Continuous Systemic Evaluation. PLoS Medicine 2009;6(8):e1000126 59. Pinnock H, Bawden R, Proctor S, Wolfe S, Scullion J, Price D, et al. Accessibility, acceptability, and effectiveness in primary care of routine telephone review of asthma: pragmatic, randomised controlled trial. BMJ 2003;326(7387):477-9 60. Pinnock H, McKenzie L, Price D, Sheikh A. Cost-effectiveness of telephone or surgery asthma reviews: economic analysis of a randomised controlled trial. Br J Gen Pract 2005;55(511):119-24 61. Mail. Data blunder. Mail Online 2009:http://www.dailymail.co.uk/news/article495188/Now-MORE-discs-containing-personal-data-missing-bunglingMinistry-Mayhem.html 62. BBC. UK families put on fraud alert.http://news.bbc.co.uk/1/hi/7103566.stm 63. McKinstry B, Watson P, Pinnock H, Heaney D, Sheikh A. Confidentiality and the telephone in family practice: a qualitative study of the views of patients, clinicians and administrative staff. Family Practice 2009;26:344-350 64. Collmann J, Cooper T. Breaching the Security of the Kaiser Permanente Internet Patient Portal: The Organizational Foundations of Information Security. Journal of the American Medical Informatics Association 2007;14(2):239-243 65. Mort M, Finch T, May C. Making and Unmaking Telepatients: Identity and Governance in New Health Technologies. Science, Technology and Human Values 2008;34(9):8-33 66. EU. Opinion of the European Economic and Social Committee on the "Communication from the Commission to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the Regions on telemedicine for the benefit of patients, healthcare systems and society". Official Journal of the European Union 2009;COM (2008) 689 final:C 317/84 67. Cuzner E. Medico-legal dilemmas posed by advances in computer technology. Good Practice 2010;1(1):4-6 68. GMC. Guidance on remote prescribing. In: http://www.gmcuk.org/guidance/ethical_guidance/prescriptions_faqs.asp, editor. Good Medical Practice: General Medical Council, 2006:Paras 39-43. 69. DoH. Liberating the NHS: An Information Revolution: Department of Health, 2010. 70. Pinnock H, Adlem L, Gaskin S, Harris J, Snellgrove C, Sheikh A. Accessibility, clinical effectiveness, and practice costs of providing a telephone option for routine asthma reviews: phase IV controlled implementation study. Br J Gen Pract 2007;57(542):714-22 71. McKinstry B, Hammersley V, Burton C, Pinnock H, Elton R, Dowell J, et al. The quality, safety and content of telephone and face-to-face consultations: a comparative study. Qual Saf Health Care 2010;19:298-303 369 72. McKinstry B, Walker J, Campbell C, Heaney D, Wyke S. Telephone consultations to manage requests for same-day appointments: a randomised controlled trial in two practices. Br J Gen Pract 2002;477:306-10 73. McKinstry B, Watson P, Pinnock H, Heaney D, Sheikh A. Telephone consulting in primary care: a triangulated qualitative study of patients and providers. British Journal of General Practice 2009;June:e209-e218 74. Nicolini D. Stretching out and expanding work practices in time and space: the case of telemedicine. Human Relations 2007;60:689 75. Cruickshank J, Beer G, Winpenny E, Manning J. Healthcare without walls, A framework for delivering telehealth at scale: 2020health.org, 2010. 76. DoH. Whole system demonstrator project., 2008. 77. Chaudhry SI, Mattera JA, Curtis J, Spertus Ja, Herrin J, Lin Z, et al. Telemonitoring in patients with heart failure. N Eng J Med 2010;363(24):23019 78. Krumholz H, Amatruda J, Smilth G, etal. Randomized trial of an education and support intervenion to prevent readmission of patients with heart failure. J Am Coll Cardiol 2002;39:83-9 370 Chapter 14 Case study: Telehealthcare for asthma Summary • Asthma is now one of the commonest long-term conditions in the UK and other economically-developed countries. • A range of telehealthcare interventions have been developed and trialled, but the effectiveness and cost-effectiveness of these remain unclear. • We conducted a Cochrane systematic review, which identified 21 eligible randomised controlled trials. • These trials included study of the telephone, video-conferencing, the Internet and other networked communications; and text messaging. • Meta-analysis showed that these interventions did not result in clinically important improvements in asthma quality of life (minimum clinically important difference=0.5): mean difference in Juniper’s Asthma Quality of Life Questionnaire (AQLQ) 0.08 (95%CI 0.01, 0.16). • Telehealthcare resulted in a non-significant increase in the odds of emergency department visits over a 12-month period: OR 1.16 (95%CI 0.52, 2.58). • Meta-analysis however showed a significant reduction in hospitalisations over a 12-month period: OR 0.21 (95%CI 0.07, 0.61), the effect being most marked in people with more severe asthma managed predominantly in secondary care settings. • There were only limited data on cost-effectiveness of telehealthcare interventions. • Overall, it seems that telehealthcare interventions are unlikely to result in clinically relevant improvements in health outcomes in those with relatively mild asthma, but they may have a role in those with more severe disease who are at high risk of hospital admission. • Further trials evaluating the effectiveness and cost-effectiveness of a range of telehealthcare interventions are needed. 371 14.1 Introduction There is no gold standard objective definition of asthma; its diagnosis is clinical, based on the presence of characteristic symptoms (wheeze, breathlessness, chest tightness and nocturnal or exercise induced cough) and of variable airflow obstruction (BTS/SIGN 2008)1. The features of asthma are so heterogeneous that, in both children and adults, it seems that what is currently termed ’asthma’ is unlikely in the future to be regarded as a single disease entity2. Much research is still needed to answer the following three fundamental questions: 1. What is asthma? 2. Who gets asthma and why? 3. Which factors predict exacerbations and treatment response?3 The Global Initiative for Asthma (GINA),4 run in collaboration with the World Health Organization and the U.S. National Heart Lung and Blood Institute (NHLBI) and National Institute for Health (NIH), estimates that 300 million people have asthma.4 Asthma is thus now a very common long-term condition and there has been an increase in prevalence in recent decades.5-7 The highest prevalence rates, as high as 30%, are amongst certain age groups in economically developed English-speaking countries.8-10 However, there has also been an increase in asthma prevalence in many economically developing countries11 12 ISAAC 1998;13 This increase affects both children and adults. Worldwide asthma presents substantial challenges. The high disease burden demands improvements in the development of and access to treatments.10 14 15 Patterns of help-seeking behaviour are also relevant, as delayed reporting is associated with greater morbidity and the need for costly emergency care. There is also a significant indirect cost burden associated with asthma through school and work absences. We summarise in this chapter the main findings from our recently completed systematic review; the methods are detailed in our Cochrane review. 372 14.2 Theoretical considerations 14.2.1 Potential benefits Telehealthcare interventions may help to address some of the above challenges by enabling remote delivery of patient-centred care, facilitating timely access to health advice and medications, prompting self-monitoring and medication compliance, and educating patients on trigger avoidance.16-19 Telehealthcare is a complex intervention and, as such, it is quite difficult to specify exactly why it works or does not work, i.e. what is/are the ’active ingredient(s)’ within the intervention.20 Some potential mechanisms through which the use of telehealthcare may enhance the quality of care and achieve cost savings include (adapted from Finkelstein 200021): • Providing patient education and counselling for primary prevention and early detection of disease; • Replacing face-to-face nursing/doctor visits; • Improving adherence to medications and other treatment regimens; • Monitoring patients’ health parameters remotely; • Enabling early detection of incipient disease exacerbation and timely intervention for early symptom management; • Reducing unscheduled/unnecessary visits to the physician and emergency room; • Preventing repeat hospitalisations. 14.2.2 Potential risks There are now in many parts of the world an increasing array of electronic tools for remotely helping people with asthma and often many are now beginning to be implemented in the absence of an explicit evidence base.19 22 A recent Cochrane Review of generic tele-consultations compared with faceto-face consultations found little evidence of clinical benefit.23 There were also a lack of analysable data for assessing the cost-effectiveness of these interventions. The authors concluded that further research is required.23 Another systematic review24 of studies of patient satisfaction with telehealthcare raised a number of important questions, these included: • What types of consultation are suitable for remote consulting? 373 • What are the effects of this mode of healthcare delivery on the clinician-patient relationship? • How do communication issues affect the delivery of healthcare via telehealthcare? • What are the possible limitations of telehealthcare in clinical practice? Answers to such questions are urgently needed or we risk blindly implementing a non-proven way of working which may have a negative effect on patients and professionals. One commonly used argument for telehealthcare is that long-term running costs will be lower than in conventional care because disease will be detected and treated early, preventing ensuing morbidity and hospitalisations and allowing patients to be cared for in their own home. However, the initial start-up costs of telehealthcare may be substantial.25 The cost-effectiveness of telehealthcare interventions therefore also needs to be established. In asthma, patients often have a high level of responsibility for their own health. In some people it can also be a life-limiting and challenging disease to manage. Telehealthcare interventions may make this easier for patients by providing timely professionally guided feedback on their condition. Such interventions may help patients to identify and address triggers and to optimise their medication regimens to address the fluctuations in their illness - and at low cost. This is the ideal situation, however such results cannot be presumed and a robust critique of the evidence base is overdue. 14.3 Empirically demonstrated main findings 14.3.1 Description of studies Our searches found 525 titles and abstracts and following review we considered 76 to be relevant. After detailed examination of the 76 full texts, 21 trials satisfied our inclusion criteria. In addition, we found 14 ongoing trials that have reported only as abstracts and a further 21 trials that have yet to report in any format. Two studies had to be translated from Japanese and one from Italian. It was only possible to obtain partial translations of the Japanese 374 reports and so the information is taken from the studies’ figures which were published largely in English. Two studies used pharmacists as the main deliverer of the telehealthcare intervention26 27 and the rest used a combination of doctors (both general practitioners and specialists) and nurses, including specialist nurses. The most common model for intervention was to have an initial face-to-face introductory session and then follow-up using telephone, telephone and web, web/other networked system or text message. This approach featured in the following studies: 27 28 29 30 31 32 33. Pinnock34 and Donald29 published follow-up papers dealing with the costs and cost-effectiveness of the interventions (35 and 36 ). Willems37 evaluation, 38 38 and 33 refer to only one trial of which publishes cost-effectiveness data and Kokubu39 was expanded on in 40 33 37 is a process is the main report. with the addition of more data and a section on costs; however, this was hard to interpret given the incomplete translations. In terms of the major telecommunication devices used in the studies, overall nine studies used the telephone (27 41 28 29 42 43 34 44 and video26 46 . In 26 45 ). Two studies used video-conferencing was used to deliver education on inhaler technique and in 46 participants submitted repeated videos for checks of their inhaler technique via modem. One study used text messaging32. Two studies used the Internet47 48. Other networked systems were used by six trials;49 30 31 39 40 33 . Van der Meer50 used text or internet. There were a number of reasons for excluding studies. Most often studies were excluded because they did not fulfil our definition of telehealthcare, i.e. there was not a two-way exchange of information between patient and healthcare professional. If the intervention involved only education without feedback or if feedback was only mechanical in nature, e.g. from a peak flow meter and not involving a professional, then the study was excluded. In addition, we excluded studies if they were found not to employ a randomised controlled design or if they were not studying an asthma population. 375 14.3.2 Risk of bias in included studies Allocation Fifteen trials used appropriate randomisation27 26 46 28 47 49 42 31 43 40 32 34 44 50 33 . For the remaining studies the methods of allocation were unclear 41 29 30 39 45 48 .The most common method of randomisation was to use a random number table, often computer-generated; however, other acceptable methods were also used, including tossing a coin. In most studies which included details of concealment, sealed envelopes were used. However, some studies appeared to use a centralised randomisation hub, but this was often unclear. Blinding Four studies 27 28 29 43 made some attempt to blind researchers as to the group allocation of their participants. The remainder did not and this may have introduced bias. Guendelman30 used self-reporting of outcomes to the nurse co-ordinator so that the same person was both involved with delivering the intervention and assessing outcomes, thus substantially increasing the risk of bias. In Pinnock34 where blinding was not feasible due to the pragmatic nature of the trial, an independent researcher validated a 20% sample of the results. Incomplete outcome data Several studies had high drop-out rates 26 29 42 31 43 . In Pinnock 44 the uptake rate and patient population dynamics were being studied as a primary outcome measure because it was a pragmatic phase IV implementation trial (as per Medical Research Council 2008).20 Selective reporting Research protocols were not sought for any of the studies. There was nonetheless some evidence of selective reporting of results. In most studies, all outcomes specified in the methods section were reported in the results, however there were some exceptions, for example: 31 the data on satisfaction were not reported and 47 did not report quality of life data. These last points do not seem to relate to selective reporting, but rather other problems. Other potential sources of bias 376 Variable efforts to recruit from ethnically diverse and marginalised populations may have impacted on the external validity of the findings. There was variable consideration of smoking. Patients with paper diaries filled in more than one day’s entry at a single time point thereby opening the data to recall bias. Some studies recruited from academic centres rather than from primary care which may limit the generalisability of findings. 14.3.3 Effectiveness of interventions Primary outcomes – clinical: Asthma quality of life The impact of telehealthcare interventions on disease-specific quality of life was assessed in 14 trials46 28 49 29 42 31 43 40 34 44 48 50 45 33 . The effect of intervention is shown in the forest plot Figure 14.1. Figure 14.1 Asthma quality of life questionnaires AQLQ Juniper mean scores Five of these studies 28 42 34 44 45 used Juniper’s validated Mini-AQLQ.51 This instrument contains 15 items which are scored from 7 (no impairment) to 1 (maximum impairment), so high scores indicate better quality of life. Three studies 48 50 33 used Juniper’s validated full 32-item Adult-AQLQ in which similarly high scores indicate better quality of life. Two studies used Juniper’s validated 23-item Paediatric Asthma Quality of Life Questionnaire (PAQLQ) 46 33 , high scores again indicate better quality of life. Jan31 used the Paediatric Asthma Caregiver’s Quality of Life Questionnaire (PACQLQ), which is filled in by the patients’ parents as did 49 and 43 . There are 13 items in this instrument and, in keeping with the other instruments, higher scores represent better 377 quality of life and less impairment by asthma52 51 . As Juniper’s quality of life instruments are similarly structured with each question answered on a Likert scale with a minimum value of one and a maximum value of seven, we considered it appropriate to perform meta-analysis of the data derived from these instruments. We performed a meta-analysis of 46 28 49 42 53 44 50 45 33 . This meta-analysis of nine studies, yielding a total of 1566 intervention and 1553 control patients, revealed a mean difference of 0.08 point improvement on this scale (95%CI 0.01, 0.16) in those randomised to intervention compared with controls. This is lower than the minimal clinically important difference of 0.5 points on the Juniper scale. It was not possible to include other studies in the meta-analysis because: • 48 had used the AQLQ, but could not be included because the data were not normally distributed and so median scores were supplied by the author on request. AQLQ score was 6.42 (IQR 3.62 to 7.00) in the Internet group, 6.31 (IQR 3.98 to 7.00) in the GP group and 6.17 (IQR 1.41 to 7.00) in the specialist group. • 35 used the Modified Marks Asthma Quality of Life Questionnaire, a validated scale in which a higher score indicated a less detrimental impact on quality of life. A clinically important difference was seen in the intervention group from recruitment to 12 months. There was no clinically important difference seen in the control group in this time. • 40 did not use a validated instrument for the measurement of quality of life. However, they showed a greater improvement in the intervention group than in the control group (P=0.04). These results, however, need to be interpreted with caution because of the unvalidated nature of the scale and, furthermore, the inability to determine what constituted a minimal clinically important difference. • 31 used the Juniper’s PACQLQ, however insufficient summary statistics were reported for meta-analysis. Only the caregivers of asthmatic children randomised to the intervention group showed significant improvement after the study compared to before the study. • 43 used the Juniper’s validated PACQLQ. Small increases in median score after the study were not statistically significant for either the 378 control (P=0.11) or the intervention group (P=0.6). These data were not normally distributed and so the mean was not used for comparison. Overall, across all these quality of life studies, these results suggested that telehealthcare did not result in clinically important improvements in quality of life. The mean difference (fixed effect) was 0.08 (95%CI 0.001, 0.16). Emergency department visits Ten studies reported data on emergency department visits 46 28 29 30 43 40 34 48 45 33 . The effect of telehealthcare interventions on emergency department visits over 12 months is shown in Figure 14.2. This meta-analysis included five trials 46 29 40 48 33 representing 619 patients in total. It revealed a non- significant increase in the odds of emergency department attendance: OR 1.16 (95%CI 0.52, 2.58). Figure 14.2 One or more emergency dept visit in 12 months In 40 the authors reported a greater reduction of night and daytime emergency room visits per patient in the control group than in the intervention group. It appears from their data that all patients in both arms were admitted to the emergency department at least once at some point during the study; it was therefore not possible to include these data in the meta-analysis. The absolute numbers of visits (presumably more often than one per patient) are not given in the study’s English tables. Of the other studies not included in the meta-analysis: 379 • 43 reported emergency department visits over a six-month interval and so could not be included in the meta-analysis, but again numbers were very small with only one or two patients attending from each arm over the study period. • 28 • There were no emergency department consultations in 34. • 45 reported only within group analyses. did not distinguish between emergency department care and full hospitalisation, therefore it was not possible to include these results in the meta-analysis. • 49 reported survival analysis by means of a Kaplan-Meier curve of time to first prednisolone course, emergency visit or hospitalisation or to whichever came first. There were a total of 31 events in these categories, but the detailed breakdown of data was not reported and so the data could not be included in meta-analysis. However, the time to the first emergency department visit was considered comparable across the two arms of the study (P=0.13). • The remaining trials27 26 41 47 35 42 39 31 32 44 50 did not include data on emergency department visits. Hospitalisations Six studies 46 35 30 40 32 48 presented data on hospitalisations. For two studies30 32 these hospitalisations occurred over a three-month period. For the remaining four studies the hospitalisations were recorded as occurring over a 12-month period46 35 40 48 . Meta-analysis of the two studies30 32 that reported data from hospitalisations over three months of study duration is shown in the forest plot in Figure 14.3. This includes data from 138 patients. Overall, there was no significant difference in the odds of hospitalisation in the intervention groups when compared to the control group (OR 0.47, 95%CI 0.010, 36.46). The confidence intervals were very wide. 380 Figure 14.3 Hospitalised once or more in 3 months of study Meta-analysis of the four studies 46 35 40 48 which reported on hospitalisation within 12 months of randomisation is shown in Figure 14.4. This includes data from 499 patients. This gave an OR of 0.21 (95%CI 0.07, 0.61) indicating that telehealthcare reduces the risk of hospitalisation. 46 35 and 48 , cross the line of no difference. These studies are large and of reasonable quality. In contrast, Kokubu40, which contributes a weight of 55.6%, shows a clearly beneficial effect. This trial focused on patients with severe asthma. Patients who had visited the night emergency department room three times or more within a year in spite of corticosteroid therapy were selected and so these patients were not representative of a typical asthma population. In the Kokubu39,40 study, 2/32 intervention patients were hospitalised and 11/34 of the control patients were hospitalised. Overall, it therefore seems as though telehealthcare interventions may prevent hospitalisations in selected high-risk populations studied over long timescales. Figure 14.4 Hospitalised once or more over 12 months of study There were no hospitalisations during the period of follow-up in 34 Numbers emergency of hospitalisations department visits in 45 . But 41 were indistinguishable from or in 43 . reported hospitalisations over the three-month period of its trial, and it was unclear as to whether only data for the control 381 group were presented and so the study was not included in the meta-analysis. One study49 reported survival analysis (Kaplan-Meier curves) of time to first hospitalisation for both groups with P=0.13 for intention-to-treat analysis. From the curve, approximately 5% of the intervention group were hospitalised once or more and 15% of the control group, however as these figures are rough visual estimates they were not used in the meta-analysis. Another42 did not clearly report on the number of hospitalisations, referring instead to mean length of inpatient stay. Also 47 31 and 44 did not report on hospitalisations. Overall, hospitalisation was an infrequent outcome. In the forest plot for hospitalisations in studies over a three-month period it can be seen that telehealthcare is not associated with a reduction in hospital admissions. However, there is a reduction over 12 months. This may point to a role for telehealthcare to reduce hospitalisation in high-risk individuals. Other primary outcomes Symptoms The following 17 studies reported on symptoms as an outcome measure: 41 28 49 42 30 31 43 39 32 34 44 48 50 45 33 27 46 In Barbanel27 symptom scores improved in the intervention group when compared to the control group over the three months of the study; the difference between groups at three months adjusted for baseline scores was 7.0 (95%CI 4.4, 9.5; P<0.001), on a scale of 10 to 40. This difference remained significant when adjustments were made for missing data. Another study46 reported the number of symptom-free days recorded by each group, but the difference between groups was not significant (P=0.13, our calculation).In Clark28 study, there was a small drop in the average number of nights with night-time symptoms per month experienced by the women following the intervention (from 5.1 to 3.7, i.e. -1.4 nights). However, across groups there was no difference in average number of nights with nighttime symptoms per month: control group 3.8 nights, intervention group 3.7 nights. Wheezing-related sleep disturbances were studied in 49 not use a validated questionnaire. Study 41 , but they did calculated the within-group percentage of symptom-free days. This was found to improve significantly with P<0.001 in both groups. The authors speculate that this may have been 382 due to frequent monitoring and telephone contacts in both arms, which could not be further improved by adding the nitric oxide telemonitoring. There was no difference across groups, the baseline-adjusted difference in mean percentage of symptom-free days was 0.3%, (SD -10% to 11%, P=0.95). 30 reported asthma control problems between groups as not significantly different at 12 weeks (P=0.07). Study43 reported that there was no significant difference between groups in their primary outcome of wheezing in the last three months and study42 reported the mean of individual changes in the Asthma Control Questionnaire (ACQ), a validated questionnaire, over 12 months. The clinic group changed by -0.11 (-0.32 to 0.11) and the telephone group by -0.18 (0.38 to 0.02), this representing a non-significant improvement in asthma control (P=0.35 when adjusted for baseline differences) (a negative change in ACQ is an improvement). Between group differences were significant in Jan 31 for the Paediatric Asthma Control Test scores change from baseline at 12 weeks: the intervention group had a significant decrease of night-time (P=0.028) and daytime (P=0.009) symptoms compared with the children in the control group. There were no between group comparisons in this study. Kokubu40 method for analysing the symptom score remains obscured by the lack of a full translation of the Japanese paper. Ostojic32 used a bespoke symptom score which produced significant results across scores for the study group and control group demonstrating that the control group had more symptoms during the study. Scores for cough were 1.42 (SD=0.28) for the study group and 1.85 (SD=0.43), P=0.028, and scores for sleep quality were: study group 0.85 (SD=0.32) and control group 1.22 (SD=0.23), P=0.021. However, it is not reported whether this symptom scoring system had been validated. 383 Pinnock34 measured symptom scores using the ShortQ asthma morbidity score three months after randomisation and found these to be similar in the two groups. Face-to-face consultations had a mean score of 1.96 (SD=1.96) and telephone consultations had a mean score of 2.10 (SD=2.16), difference 0.41 (95%CI -0.41, 0.68) (P=0.62), i.e. a non-significant difference. In 2007, study44 used the validated Asthma Control Questionnaire and found a non-significant mean difference of 0.12 (95%CI -0.06, 0.31) (P=0.19) for the telephone option versus face-to-face only. Rasmussen48 reported an improvement in symptoms in the Internet versus specialist groups: OR 2.64 (95%CI 1.43, 4.88, P=0.002); and also in the Internet versus GP groups: OR 3.26 (95%CI 1.71, 6.19, P<0.001).Here the Internet group showed the greatest likelihood of improvement over six months. Another study50 used the validated ACQ to compare groups. The Internet group showed greater improvement of asthma control than did the usual care group: change from baseline -0.54 versus -0.06; adjusted difference -0.47 (CI -0.64 to -0.3) which was just slightly smaller than a clinically relevant difference of -0.5 (where negative changes represent improvements). Vollmer45 assessed several markers of symptom status including the Asthma Therapy Assessment Questionnaire (ATAQ), asthma impact score, selfreported health status, self-assessed severity of asthma in the past four weeks and nocturnal waking in the past four weeks. No significant difference between the telephone and usual care groups were found for any of these outcomes (ATAQ P=0.56, asthma impact score P=0.46, self-reported health status P=0.08, self-assessed severity of asthma in the past four weeks P= 0.22 and nocturnal waking in past four weeks P=0.84). Willems33 recorded the changes in asthma symptoms of coughing, production of sputum and shortness of breath/wheezing, however no statistically significant differences in improvement in any of the symptoms were observed. The above results suggest that symptom scores may be improved with 384 telehealthcare. However, in many cases there was no difference between groups. Future research should use validated scoring systems which can then be pooled in meta-analysis to give a more accurate picture of the extent to which telehealthcare interventions improve symptom scores. Improved access Improved access was clearly demonstrated in Pinnock34. Of patients randomised to receive the telephone review, 74% of patients were reviewed, whereas only 48% of patients in the surgery-only group were reviewed. There was therefore a significant improvement of 26% (95%CI 14, 37; P<0.001). Asthma-related morbidity at three months (in terms of acute exacerbations of asthma) P=0.68; emergency broncho-dilation P=0.97; and steroid courses for asthma P=0.64) were not significantly different across groups, and neither was quality of life (Juniper scores P=0.69). And so in this study access was improved with no loss in quality and at no additional cost. Access in 44 was also improved: the proportion reviewed was 66.4% of patients in the telephone option group compared with 53.8% in the face-toface only group. Chan46 treated the control group with an ambulatory asthma clinical pathway with six visits scheduled over the period of 12 months after the initial visit. The intervention group had three in-person visits at 0, 26 and 52 weeks and the rest as virtual visits using video-conferencing technology. In Gruffydd-Jones’ study, 35% more asthma patients received their annual review in the telephone group than in the clinic group and the cost of the telephone group was less. In Vollmer45 there was evidence of a negative reaction to automated computerised calling and the intervention was more successfully delivered in the live caller arm (P<0.001). Secondary outcomes Study withdrawal Barbanel 27 lost one patient when (s)he moved away. 385 Bynum26: 49 students were randomised. Three did not attend any visit, eight had never used a metered dose inhaler (MDI) before and so did not meet the inclusion criteria (the study does not explain why they were not screened out before randomisation) and two students did not attend the follow-up visit. Chan46: 127 children were assessed for eligibility: one did not meet the inclusion criteria, five refused to participate and one was not able to participate. The study report does not give further details of why not. One hundred and twenty children were randomised, 11 were lost to follow-up, seven discontinued because they moved, and so 55 were included in the analysis for the control group and 47 remained for analysis in the intervention group. Chatkin41: 293 patients were screened for participation, four were excluded for not fulfilling the inclusion criteria, eight for not responding to the telephone calls and 10 for not returning their inhaler disk to the study office. In the final analysis control group n = 131 and intervention group n = 140. Clark28: 808 women provided baseline data and mailed a consent form back following the mailing of 2336 invitation letters. Four hundred and twenty-four were randomised to the intervention group and 384 to the control group. There was then attrition of 133 participants from the intervention group and 87 participants from the control group. The reasons for this were not reported. Cruz-Correia47: 21 patients were included from 28 patients assessed in the outpatient clinic. Patients were only included if they used the Internet. Sixteen patients provided their opinions on the monitoring instruments. Four patients did not use the monitoring tools because of technical problems with the instruments. de Jongste49: 151 children were randomised and four children were excluded, two due to non-compliance, one because he or she had been inappropriately included and did not have allergy and the final child had moved abroad and was unable to transfer data. All others completed the study. 386 Donald29: 660 patients were assessed for eligibility, 385 could not be contacted, 154 did not want to take part in the study 31 were excluded and 19 did not attend the first meeting. Following randomisation there were 36 in the intervention group with 31 left at the end of the study for analysis and 35 controls, of which 29 were left in the final analysis. The study did not discuss why participants withdrew following randomisation. Gruffydd-Jones42: 97 patients were randomised to the control group, only 82 attended their first visit and 62 completed the study, of which 28 completed all visits and 34 completed two visits only. Ninety-seven patients were similarly randomised to the telephone group, 90 attended their baseline visit, three patients left the area, two patients died and 84 completed all three visits. An explanation for the final patient was missing from the patient flow diagram. Guendelman30: Reasons for withdrawing from the study included moving out of the area (n = 3) or life crisis (n = 4). Five families who dropped out were unavailable for contact by the study authors to find out their reasons. Jan31: 184 families were eligible with access to the Internet. Five families declined to participate as they were “too busy”, “not interested” or found it “too complex to perform the diary card”. Fifteen participants were excluded because of either a request from the parents or a lack of data due to Internet failure. Khan43: 310 children were enrolled in this study, 266 completed the follow-up questionnaires. The reasons for non-response were not explored although in two cases a search of the telephone directory enabled the discovery of a new address. Kokubu39: Two patients withdrew from the intervention group and one from the control group, however the reasons for this are not described. 387 Kokubu40: Five patients withdrew from the intervention group and four from the control group but the reasons for this are not described in the report according to the translation we have available. Ostojic32: Patients unfamiliar with text SMS or without consistent access to a cell telephone were excluded. After enrolment no patient withdrew from the study. Pinnock44: This was a phase IV implementation study and so there is no selection of participants. It instead takes place within the real-world situation of general practice where patients move, die or their asthma becomes more active or becomes inactive or is re-diagnosed as COPD or patients refuse to take part. Rasmussen48: 300 subjects were enrolled who fulfilled the criteria for asthma, 253 subjects completed the trial. Reasons for the withdrawals were not given. Van der Meer50: recruitment was undertaken from 37 general practices and continued until there were 200 patients entered in the study. Results were available for 183 patients. Twenty patients did not complete the three-month questionnaire, eight patients were lost to follow-up and nine patients withdrew consent. Vollmer45: It is difficult to assess withdrawal in this study as it was devised to compare usual care with automated telephone outreach and the outcome measures were taken from a representative sample. However, participation statistics were low for the intervention arm, only 38% participated in the first round of calls, 32% in the second round and 18% in the third round. Overall 47% of the intervention group had one call and 12.1% completed all three calls. Compared with those who did not participate in any call, participants were more likely to be female, take more inhaled corticosteroids and report worse asthma-specific quality of life at baseline. Interestingly 59.9% of the live-caller arm completed at least one call and 27.6% all three calls, suggesting that patients preferred a live call to the automated calls. 388 Willems33: 274 patients were assessed as potentially eligible from patient records. Eighteen did not have a house phone connection and so were excluded as not meeting inclusion criteria. One hundred and forty-seven refused to participate because of the following reasons: no time (n=63), not interested (n=28), no symptoms (n=18), too confronting (n=13) and other reasons (n=25). Therefore, 109 were enrolled, seven were lost to follow-up, six refused participation continuation and one person emigrated. Time off school or work Clark28: reported the yearly average number of days missed work per month as 3.0 (SD=7.1) for the control group and 2.3 (SD=6.2) for the intervention group (P=0.14). In the Donald29 trial, 24 intervention group participants worked or studied and lost 10 days over 12 months, and 25 of the control group participants lost a total of 11 days; the difference between the groups was not significant (P=0.58). In Guendelman,30 the odds of missing school in the past six weeks in the Health Buddy group were 0.74 (95%CI 0.37, 1.5). In summary, it appears that these telehealthcare interventions did not reduce time off work or school. PEF monitoring and diary monitoring 46 31 33 and32 report PEF flow monitoring. The initial improvements in inhaler technique seen in both arms of 46 can be attributed to the monitoring of participants by healthcare professionals. As monitoring by the health professionals dropped in the second 26 weeks of the study to a telephone call once a week in the control group, the participants’ inhaler technique scores dropped off. Mean peak flow (+/-SD) as a percentage of personal best was reported as 91.6% (+/- 27.2) in the intervention group and 100% (+/-17.6) in the control group at the end of the study. 389 Monitoring, or more accurately “prompting”, was also important in 41 a study in Brazil. A twice-weekly telephone call made by a trained student nurse to the patients in the experimental arm resulted in an inhaler adherence rate of 74.3 % whereas the rate in the control arm was only 51.9% (number needed to treat to benefit 4.5, i.e. telephone calls to 4.5 patients were required to prevent one non-adherence or missed inhaler dose). The content of the telephone call was individualised to each patient. In keeping with this advantage of monitoring, Kokubu39 also found that those patients who checked their peak flow most regularly tended to have the best peak flows. Patients who did not check or transmit peak flow data regularly remained poorly controlled. In Jan31 at week 12, children in both groups had a significantly increased PEF. Between-group differences are reported as non-significant. Willems33 reported rank correlations of lung function values (PEF and FEV1) with morning symptoms and evening symptoms as moderate to high. Absolute values were not reported in the study. Peak flow in L/min was measured by Ostojic32 as mean PEF by time of day (morning, afternoon, evening) none of which were significant. However, PEF variability (%) in the text message intervention group (16.12 +/- 6.93) and the control group (27.24 +/- 10.01) showed a significant difference (P=0.049). Therefore, overall it can be seen that telehealthcare improved PEF in some studies, but that this was not a consistent finding. However, monitoring in itself appeared to generate an advantage. Spirometry FEV1/FVC 46 49 32 48 and 50 reported data on FVC and FEV1 as follows. There were no differences between groups in FVC, forced expiratory volume in 1 second or forced expiratory flow in mid-expiratory phase at the end of the Chan46 study. FEV1 was similar at baseline in both groups in de Jongste49 and there was no significant difference in change from baseline over the course of the study between the groups: both groups improved. 390 Neither FVC nor FEV1 as a percentage of predicted was significantly different across the groups in Ostojic32. Rasmussen48 reported odds ratios for improved FEV1 > 300 ml over baseline of 3.26 (1.50 to 7.11) for Internet versus specialist, with a significance of P=0.002. The odds ratio of Internet versus GP group was also significant at P<0.001 (OR 4.86, 95%CI 1.97, 11.94). These odds ratios were calculated after six months of intervention. Van der Meer50 reported that mean pre-bronchodilator FEV1 changed by 0.24 L for the Internet and -0.01 L for the usual care group thus indicating an improvement in the telehealthcare arm’s FEV1. Patient satisfaction In the Cruz-Correia47 study, questionnaires were used to assess patient satisfaction and it was found that overall patients preferred the web-based system for monitoring their asthma to the paper based system. Gruffydd-Jones42 triaged patients in the intervention group by telephone call then allocated follow-up appointments accordingly. Of patients in this group 88% expressed a strong preference for care in this form compared to the previous standard form. Pinnock36 is an additional report of patients who had completed the RCT comparing telephone and face-to-face consultations for routine asthma reviews; they were sent a questionnaire to better understand their preference for future reviews. Thirty-three percent preferred telephone consultations for future reviews, 17% preferred surgery and 50% expressed no preference. Thematic analysis of the free-text responses identified five themes including convenience of telephone consultations, specific problems with telephone consultations (e.g. confidentiality when phoning from a public place), the human dimension of face-to-face encounters, that the mode of consultation 391 might change according to the clinical situation and wider implications such as using email for attachment of PEF information. In summary, patient satisfaction appeared to be high, but these newer approaches did not appear to suit all patients. In Bynum26 and Pinnock34 there was no significant difference in the satisfaction scores of each of the arms (Bynum26 P=0.132; and Pinnock34 P=0.51). Kokubu40 asked the following question: What did you think about the frequency of telephone consulting? 1. Adequate 87% 2. Too much 7% 3. Too few 3% 4. No response 3% Insufficient detail was reported by Chan46 to interpret the satisfaction scores published. Willems37 found no clinically important differences between a satisfaction questionnaire administered at four months and again at 12 months. Both arms of the trial used a PEF monitor and both arms answered the satisfaction questionnaire. Only 4% of patients experienced “a lot” of symptoms over the previous months. Forty percent of patients said that they would like to use the monitor again in the future; 36% said maybe. Ninety-four percent of patients were either much or very much appreciative that lung function deterioration is immediately noticed by the nurse. Eighty-four percent said that they felt “not at all less safe contacting the nurse instead of the doctor”. Eighty-seven percent found that they had a qualitative improvement by playing an active role in their asthma management. Sixty-five percent felt safer while using the monitor. Eighty-four percent said the monitor did not interfere with their daily activities. With regard to questions on the application of the monitor both children and adults were highly satisfied with using the monitor, 87% finding it “not difficult at all” to perform the spirometry test. 392 Costs from the healthcare perspective This refers to the costs of providing the intervention compared with differences in outcomes between the intervention and control groups (all currency conversions were undertaken in February 2010 to US dollars). In Pinnock36 the cost to the practices of face-to-face asthma reviews was $17.31 each per consultation achieved as there was a higher default rate than for telephone reviews. Telephone reviews reached more patients at $11.20 per consultation achieved. Access was therefore improved at lower cost per consultation, however overall costs were similar because more patients were interviewed in the telephone group. In Pinnock44 the cost per patient reviewed was significantly lower in the telephone option group; $15.63 versus $19.85. This generated a cost saving per patient reviewed of between $4.02 and $4.41 per patient reviewed (i.e. per six-monthly review achieved). Data on costs were published in the Donald35 paper. The overall cost of delivering the intervention from the healthcare perspective was $1693.91, i.e. the additional six phone calls each over the study time period for all intervention patients. These calls resulted in the intervention group having six readmissions overall, as opposed to the control group’s 20 readmissions to hospital. The total cost of the hospital readmissions in the control group was $35,878.52. Therefore there was a significant cost saving solely on the basis of reduced hospital admissions. The intervention group also showed a clinically significant increase in quality of life scores over the 12-month follow-up period in comparison to no change in the control group’s scores. Formal cost-effectiveness analysis looking at Quality Adjusted Life Years (QALYs) or Disability Adjusted Life Years (DALYs) was not undertaken. In the Kokubu40 study, it was estimated that a saving of $7074 per year per patient would be achieved if they were to use the telehealthcare system rather than conventional treatment, these savings largely being achieved by a reduction in hospitalisation costs. 393 In Willems38 costs from the societal perspective were calculated by costing from the healthcare perspective and adding the cost of over-the-counter medication, family informal care and productivity losses due to time off work. Cost-effectiveness data were calculated from the costs and the quality of life data from EQ-5D utility (adults and children) and the SF-6D utility (adults only). Cost-effectiveness depends on what society is prepared to pay per QALY gained. Among adults the healthcare costs were a mean of $695.54 higher in the intervention group. After adjustment for baseline differences by multiple regression an average 0.03 QALY were gained from the intervention. This equates to $42,520.39 (31,035 Euros) per QALY gained from the societal perspective. Among children the healthcare costs were $829.56 higher in the intervention group. On average 0.01 QALY were gained from the intervention, equating to $80,437.16 (59,071 Euros) per QALY from the societal perspective. Overall it therefore appears that the studies which analysed costs found that where hospitalisation was prevented, costs were favourable to continuing the intervention. However, this did not hold true for all studies. 14.4 Empirically demonstrated risks In Gruffydd-Jones42 study, two patients in the telephone triage arm died. Following contact with the author it was confirmed that these were not asthma-related deaths. Overall, studies did not report adverse events, with the exception of Rasmussen48 which found that an increase in corticosteroid dose was likely to result in oral candidiasis or dysphonia. In this study at follow-up an increase in inhaled corticosteroids was found in all groups, but significantly more patients in both the Internet and specialist groups were using inhaled corticosteroids than in the general practice group. 14.5 Implications for policy, practice and research This review included 21 trials measuring the effects of telehealthcare interventions. Overall, this review indicates that telehealthcare-based interventions do not have an appreciable impact on disease-specific quality of life or risk of emergency department attendance for asthma; they may, however, result in a reduction in the risk of hospitalisation for asthma, particularly in high-risk populations. 394 This latter finding is consistent with the high hopes that policymakers hold for telehealthcare in terms of its ability to prevent asthma patients requiring hospital admission. However, this meta-analysis is highly reliant on the results from the Kokubu40 and Donald studies. The Kokubu study40 selected a group of patients who had very poorly controlled asthma, requiring oral steroids at least three times in the previous 12 months or having had a previous hospital admission, and so they were much more likely to be admitted to hospital at baseline than patients in other studies which were based in primary care. In addition, our knowledge of this study is incomplete due to difficulties in translating this article. When the data were analysed without the Kokubu study40 the rate of hospitalisations over 12 months became non-significant (P=0.1). Similarly, the Donald35 study recruited from a group of adults who had been admitted to hospital with their asthma at some point previously. These two studies together suggest that telehealthcare might be most useful for high-morbidity asthma groups selected from secondary care over longer intervals (i.e. greater than 12 months). However, overall hospitalisations represent infrequent events and so generalisation from this limited number of patient outcomes should only be undertaken cautiously. Symptom scores data were inconsistent and often reported as within-group changes rather than across-group changes. In some instances telehealthcare related to improved symptoms, but mostly the reporting of data was not robust enough to draw any clear conclusions. There were few adverse events. There is always a risk in reducing the level of care from face-to-face to telehealthcare that if patients are inexpertly triaged they will receive insufficient support for their needs and their safety will be compromised. We, however, found no evidence of this having happened. This suggests that authors were aware of this risk and managed it appropriately. This review did identify a tendency for patients to abandon the technology and cease self-monitoring when they felt well; for example in Chan46 children’s adherence to submission of inhaler technique videos decreased over time. This observed pattern calls into question not only what works for asthma in telehealthcare and in what contexts, but for whom too. It seems that many of 395 the primary care population with asthma do not see themselves as having a chronic illness needing constant medication but as having an occasional acute inconvenience–this is a separate issue that is best explored using qualitative designs. Nine of the 21 studies included in this review studied the telephone as an intervention. However, Tulu54 found that only 6.6% of the telemedicine projects listed in the Telemedicine Information Exchange survey used the “Plain Old Telephone System” (sic). It would seem that more modern technologies such as video-conferencing, Skype® and web 2.0 technologies are attracting interest but have not yet made it into the asthma literature. There were several trials which had a very low participation rate for collection of their follow-up data, e.g. approximately 38% in Vollmer45 and 27% in Delaronde55. Studies varied in their recruitment either from hospital outpatients or emergency departments in which case patients with more severe forms of asthma were recruited than in primary care where the patients with milder asthma were recruited. The type of patient in the study had implications for the findings as we saw with the Kokubu40 and Donald35 studies. There is a plethora of ongoing research and research that has as yet only been published in abstract form. This emerging literature includes a number of studies looking at networked telehealthcare tools. It is worth noting that in the situations in which such solutions might prove most useful, e.g. remote and rural settings, there may be anticipated difficulties with broadband linkage required for networks. Similarly, frail, elderly people who could benefit from telehealthcare to help to maintain their independence may lack the cognitive skills to be able to adapt to network interfaces and so have difficulties using the technology at all. The research in this review has mostly taken place in a developed world setting with the equivalent of primary and secondary care, sometimes transferring the responsibility for care from secondary to primary, or setting up 396 a type of preventative secondary care. As such it is likely to be broadly transferable to other developed world settings. It is quite possible that some types of telehealthcare may work well in the developing world as well. This is because the developing world often has good mobile telephone network coverage. However, the devices for interfacing with patients require reliable electric power and skilled labour which might be more difficult to secure in such countries. In general, the biases seen included a lack of proper randomisation, problems with allocation concealment and a lack of overt statements of specific methodological processes, such as blinding. Therefore many of the risk of bias tables were often populated with the judgement ’unclear’ due to insufficient information. In summary, this review has found 21 randomised controlled trials researching the effects of telehealthcare intervention for asthma. In addition, we found 21 ongoing studies and unpublished trials. Despite some lack of consistency in the way telehealthcare has been researched thus far (see below), this represents a substantial body of reasonably high quality research in different settings internationally and assessing different technologies. Some positive conclusions can be drawn. Telehealthcare improves access and is no worse than normal care in carefully selected and triaged primary and secondary care patients. It does not, however, appear to have the desired impact on quality of life and these interventions have little or no significant impact on emergency department visits. However, given careful candidate selection conditions, telehealthcare may reduce hospital admission rates and associated costs. There is also some evidence for improved symptoms in telehealthcare trial arms where symptoms are dealt with quickly and exacerbations are prevented in a way not open to the control arm. One of the problems with telehealthcare research that we have seen in this review is that the control arm is not always usual care, but often receives an enhanced input from the clinicians–for example, a greater number of visits or face-to face contacts and so this results in greater adherence to medications 397 and surveillance of disease fluctuations in both arms of the trial and improved outcomes for both arms. Telehealthcare fits into the Medical Research Council’s model of a “complex intervention”20 and as such it seems that each telehealthcare intervention is very diverse. This can make it very hard to pinpoint the “active ingredients” of a telehealthcare intervention. Confusion also comes from the different modes of delivery within the bracket of telehealthcare. For example, both Internet- and telephone-based “helplines” improve access to clinicians for patients. In this example it is the function of improving access which leads to reduced hospitalisations–and so far the evidence shows that this is not to the detriment or advantage of quality of life. The form of the intervention, whether telephone or Internet, seems to have less of an impact than the function of the access which results in timely advice to prevent exacerbation in a way that outpatient clinics cannot deliver in comparison. It is important then in future studies that interventions are described as fully as possible, that relevant process or intermediary measures are studied, and consideration is also given to embed qualitative work within these complex intervention trials to help assess how these interventions exert their effects. Future telehealthcare interventions studied are likely to be even more complex and include the features of web 2.0. Such research is important to find how best to harness these innovative technologies without inadvertently causing harm. It is, however, important that these interventions are patientcentred, and that they are developed through close consultation both with patients, but also with healthcare professionals to maximise the chances of success.56 57 Acknowledgements: This chapter is based on our review which is published in The Cochrane Library: McLean S, Chandler D, Nurmatov U, Liu J, Pagliari C, Car J, et al. Telehealthcare for asthma. Cochrane Database Syst Rev. 2010 Oct 6;(10):CD007717. Margaret Fletcher translated the Ricci paper from Italian, for which we are grateful. We are grateful to our co-authors Joe Liu and David Chandler for their help with this Cochrane review. Finally, our thanks to colleagues in the Cochrane Airways Group for their help and support in undertaking this review. Permission to reproduce this material has been applied for. 398 References 1. BTS/SIGN. British Guideline on Asthma. British Thoracic Society, Scottish Intercollegiate Guidelines Network, National Guideline, 2008. 2. Editorial. A plea to abandon asthma as a disease concept. Lancet 2006;368(9537):705 3. Editorial. Asthma: still more questions than answers. Lancet 2008;372(9643):1009 4. Masoli M, Fabian D, Holt S, Beasley R. Global Burden of Asthma: The Global Initiative for Asthma, 2003. 5. Anderson H, Gupta R, Strachan D, Limb E. 50 years of asthma: UK trends from 1955 to 2004. Thorax 2007;62(1):85-90 6. Asher M, Montefore S, Bjorksten B, Lai C, Strachan D, Weilland S, et al. Worldwide time trends in the prevalence of symptoms of asthma allergic rhinoconjunctivitis and eczema in childhood: ISAAC Phases One and Three repeat muli-country cross-sectional surveys. Lancet 2006;368:733-743 7. Pearce N, Sunyer J, Cheng S, Chinn S, Bjorksten B. Comparison of asthma prevalence in the ISAAC and the ECRHS. European Respiratory Journal 2000;16:420-6 8. Anandan C, Nurmatov U, Sheikh A. Is the prevalence of asthma declining, systematic review of epidemiological studies. Allergy 2010;65(2):152-67 9. Punekar Y, Sheikh A. Establishing the incidence and prevalence of cliniciandiagnosed allergic conditions in children and adolescent using routinely collected data from general practices. Clinical Experimental Allergy 2009;39(8):1209-16 10. Simpson C, Sheikh A. Trends in the epidemiology of asthma in England: a national study of 333,294 patients. J R Soc Med 2010;103(3):98-106 11. Stewart A, Mitchell E, Pearce N, Strachan D, Weiland S, ISAAC. The relationship of per capita gross national product to the prevalence of symptoms of asthma and other atopic diseases in children (ISAAC). International Journal of Epidemiology 2001;30:173-9 12. Weiland S, Bjorksten B, Brunekreef B, Cookson W, von Mutius E, Strachan D. Phase II of the International Study of Asthma and Allergies in Childhood (ISAAC II): rationale and methods. Eur Respir J 2004;24(3):406-12 13. Marks G. Geographical heterogeneity of asthma International Journal of Epidemiology 2001;30:179-80 14. Anandan C, Gupta R, Simpson C, C F, Sheikh A. Epidemiology and disease burden from allergic disease in Scotland: analyses of national databases. Journal of the Royal society of Medicine 2009;102(10):431-42 15. Gupta R, Sheikh A, Strachan D, Anderson H. Burden of allergic disease in the UK: secondary analyses of national databases. Clinical Experimental Allergy 2003;34(4):520-526 16. Car J, Sheikh A. Telephone consultations. BMJ 2003;326:966-9 17. Car J, Sheikh A. Email consultations in healthcare: 1- scope and effectiveness. BMJ 2004;329:435-438 18. Car J, Sheikh A. Email consultation in healthcare: 2- acceptability and safe application. BMJ 2004;329:439-42 19. McLean S, Sheikh A. Does telehealthcare offer a patient-centred way forward for the community based management of long-term respiratory disease? Primary Care Respiratory Journal 2009;18(3):125-126 20. Craig P, Dieppe P, MacIntyre S, Michie S, Nazareth I, Petticrew M. Developing and evaluating complex interventions: new guidance.: Medical Research Council, 2008. 21. Finkelstein J. INHALED STEROID ADHERENCE IN MODERATE & SEVERE ASTHMA. CRISP (Computer Retrieval of Information on Scientific Projects) 2000;31/07/2005 End date 399 22. McKinstry B, Pinnock H, Sheikh A. Telemedicine for management of patients with COPD? Lancet 2009;374(9691):672-3 23. Currell R, Urquhart C, Wainwright P, Lewis R. Telemedicine versus face to face patient care:effects on professional practice and health care outcomes. Cochrane database of systematic reviews 2008;2000(2) 24. Mair F, Whitten P. Systematic review of studies of patient satisfaction with telemedicine. BMJ 2000;320:1517-1520 25. Whitten P, Mair F, Haycox A, May C, Willliams T, Hellmich S. Systematic review of cost effectiveness studies of telemedicine interventions. BMJ 2002;324:1434-7 26. Bynum A, Hopkins D, Thomas A, Copeland N, Irwin C. The Effect of Telepharmacy Counselling on Metered-Dose Inhaler Technique Among Adolescents With Asthma in Rural Arkansas. Telemedicine Journal and E-Health 2001;7(3):207-217 27. Barbanel D, Eldridge S, Griffiths C. Can a self-management programme delivered by a community pharmacist improve asthma control? A randomised trial. Thorax 2003;58(10):851-854 28. Clark NM, Gong ZM, Si JW, Lin X, Bria WF, Johnson TR. A randomized trial of a self-regulation intervention for women with asthma. Chest 2007;132(1):88-97 29. Donald KJ, McBurney H, Teichtahl H, Irving L. A pilot study of telephone based asthma management. Australian Family Physician 2008;37(3):170-3 30. Guendelman S, Meade K, Benson M, Chen YQ, Samuels S. Improving asthma outcomes and self-management behaviors of inner-city children: a randomized trial of the Health Buddy interactive device and an asthma diary. Archives of Pediatrics & Adolescent Medicine. 2002;156(2):114-20 31. Jan RL, Wang JY, Huang MC, Tseng SM, Su HJ, Liu LF. An internet-based interactive telemonitoring system for improving childhood asthma outcomes in Taiwan. Telemedicine Journal and e-Health 2007;13(3):257-268 32. Ostojic V, Cvoriscec B, Ostojic SB, Reznikoff D, Stipic-Markovic A, Tudjman Z. Improving asthma control through telemedicine: A study of short-message service. Telemedicine Journal & E-Health 2005;11(1):28-35 33. Willems D, m J, J H, van Duurling R, Wouters E, J S. The effectiveness of nurseled telemonitoring of asthma results of a randomized controlled trial. Journal of Evaluation in Clinical Practice 2008;14:600-608 34. Pinnock H, Bawden R, Proctor S, Wolfe S, Scullion J, Price D, et al. Accessibility, acceptability, and effectiveness in primary care of routine telephone review of asthma: pragmatic, randomised controlled trial. BMJ 2003;326(7387):477-9 35. Donald K, McBurney H, Teichtahl H, Irving L, Browning C, Rubinfeld A, et al. Telephone based asthma management; Financial and individual benefits. Australian Family Physician 2008;37(4):212-275 36. Pinnock H, McKenzie L, Price D, Sheikh A. Cost-effectiveness of telephone or surgery asthma reviews: economic analysis of a randomised controlled trial. Br J Gen Pract 2005;55(511):119-24 37. Willems DC, Joore MA, Hendriks JJ, van Duurling RA, Wouters EF, Severens JL. Process evaluation of a nurse-led telemonitoring programme for patients with asthma. Journal of Telemedicine & Telecare 2007;13(6):310-7 38. Willems DCM, Joore MA, Hendriks JJE, Wouters EFM, Severens JL. Costeffectiveness of a nurse-led telemonitoring intervention based on peak expiratory flow measurements in asthmatics: Results of a randomised controlled trial. Cost Effectiveness and Resource Allocation 2007;5 39. Kokubu F, Suzuki H, Sano Y, Kihara N, Adachi M. Tele-medicine system for highrisk asthmatic patients. Arerugi - Japanese Journal of Allergology 1999;48(7):700-12 40. Kokubu F, Nakajima S, Ito K, Makino S, Kitamura S, Fukuchi Y, et al. Hospitalization reduction by an asthma tele-medicine system. Arerugi - Japanese Journal of Allergology 2000;49(1):19 400 41. Chatkin JM, Blanco DC, Scaglia N, Wagner MB, Fritscher CC. Impact of a lowcost and simple intervention in enhancing treatment adherence in a Brazilian asthma sample. Journal of Asthma 2006;43(4):263-266 42. Gruffydd-Jones K, Hollinghurst S, Ward S, Taylor G. Targeted routine asthma care in general practice using telephone triage. British Journal of General Practice 2005;55(521):918-923 43. Khan MSR, O'Meara M, Stevermuer TL, Henry RL. Randomized controlled trial of asthma education after discharge from an emergency department. Journal of Paediatrics & Child Health 2004;40(12):674-677 44. Pinnock H, Adlem L, Gaskin S, Harris J, Sheikh A. Accessibility clinical effectiveness and practice costs of providing a telephone option for routine asthma reviews: phase IV controlled implementation study. British Journal of General Practice 2007;57(714-722) 45. Vollmer WM, Kirshner M, Peters D, Drane A, Stibolt T, Hickey T, et al. Use and impact of an automated telephone outreach system for asthma in a managed care setting. American Journal of Managed Care 2006;12(12):725-733 46. Chan DS, Callahan CW, Hatch-Pigott VB, Lawless A, Proffitt HL, Manning NE, et al. Internet-based home monitoring and education of children with asthma is comparable to ideal office-based care: results of a 1-year asthma in-home monitoring trial. Pediatrics 2007;119(3):569-78 47. Cruz-Correia R, Fonseca J, Lima L, Araujo L, Delgado L, Castel-Branco MG, et al. A comparison of web-based and paper-based self management tools for asthma: Patients' opinions and quality of data in a randomised crossover study. Journal on Information Technology in Healthcare 2007;5(6):357-371 48. Rasmussen LM, Phanareth K, Nolte H, Backer V. Internet-based monitoring of asthma: A long-term, randomized clinical study of 300 asthmatic subjects. Journal of Allergy & Clinical Immunology 2005;115(6):1137-1142 49. de Jongste J, Carraro S, Hop W, Group TCS, Baraldi E. Daily telemonitoring of exhaled nitric oxide and symptoms in the treatment of childhood asthma. American Journal of Respiratory and Critical Care Medicine 2009;179:93-97 50. van der Meer V, Bakker M, van den Hour W, Rabe K, Sterk P, Kievit J, et al. Internet-based self management plus education compared with usual care in asthma. Annals of Internal Medicine 2009;151:110-120 51. Juniper E, Guyatt GH, Cox FM, Ferrie PJ, King DR. Development and validation of the Mini Asthma Quality of Life Questionnaire. European Respiratory Journal 1999;14(32-8) 52. Juniper EF, Guyatt GH, Feeny DH, Ferrie PJ, Griffith LE, Townsend M. Measuring quality of life in the parents of children with asthma. Qualty of Life Research 1996;5:27-34 53. Pinnock H, Sheikh A, Bawden R, Proctor S, Wolfe S, Scullion J, et al. Cost effectiveness of telephone vs face to face consultations for annual asthma review: randomised controlled trial in UK primary care [Abstract]. European Respiratory Journal 2003;22(Suppl 45):Abstract No: [2250] 54. Tulu B, Chatterjee S, Maheshewari M. Telemedicine taxonomy: a classification tool. Telemed Journal E Health 2007;13(3):349-58 55. Delaronde S, Peruccio D, Bauer B. Improving asthma treatment in a managed care population. The American Journal of Managed Care 2005;11(6):361-368 56. Catwell L, Sheikh A. Evaluating eHealth interventions: the need for continuous systemic evaluation. . Plos med 2009;6(8):e1000126 57. Catwell L, Sheikh A. Information technology system users must be allowed to decide on the future direction of major national IT initiatives. But the task of redistributing power equally amongst stakeholders will not be an easy one. Inform Prim Care 2009;17(1):1-4 401 Chapter 15 Case study: Internet-based interventions for smoking cessation Summary • Smoking is the leading cause of preventable morbidity and mortality. • With the increasing penetration of the Internet, online resources represent a potentially important means of promoting smoking cessation, particularly amongst younger people. • A variety of educational and behavioural support tools have been developed to try and promote smoking cessation, but the effectiveness of these approaches remains unclear. • Our Cochrane review identified 20 eligible trials, which studied Internetbased interventions of a range of intensities and complexities, comparing these with either no intervention or a different Internet-based resource. • This review suggested that more engaging, tailored, intensive approaches may have modest short-term effects (at three months), particularly if used alongside other pharmacological measures; the impact on longer-term abstinence however remains unclear. • Future studies should focus on establishing the longer-term (i.e. over at least six and preferably 12 months) effectiveness and cost-effectiveness of Internet-based smoking cessation programmes. • It is important that future trials also seek to describe the likely mechanisms through which these interventions may (or may not) be exerting their effects–they should therefore also report data on patient satisfaction, changes in knowledge, motivation, dependency, quit attempts and safety considerations. 402 15.1 Introduction Tobacco smoking is a major preventable cause of death in both developed and developing countries. Every day over 13,000 people die from tobaccorelated diseases.1 If current trends continue, by 2025 tobacco will contribute to the death of 10 million people worldwide each year, with seven million of these deaths occurring in developing countries.2 People who smoke are more prone to developing various types of cancer, such as those of the oral cavity, larynx, bladder and particularly lung cancer. Tobacco smokers are also at substantially increased risk of developing heart disease, stroke, emphysema and other fatal diseases.1 Smoking also imposes a huge economic burden on society–currently up to 15% of the total healthcare costs in developed countries.3 Additionally, passive smoking is associated with serious morbidity.4 To reduce the growing global burden of tobacco-related mortality and morbidity, and the impact of tobacco use on economic indicators, tobacco control has become a world-wide public health imperative.1 Prevention and cessation are the two principal strategies in the battle against tobacco smoking. Nicotine is highly addictive.5 There is evidence that, for example, although 70% of US smokers say they want to quit, only five per cent are able to sustain cessation for one year.6 The balance between the individual's motivation to stop smoking and his or her dependence on cigarettes influences smoking cessation success. Dependence in smokers and their motivation to stop smoking can be assessed by simple questions.7 Maximising the delivery of smoking cessation interventions can achieve more in terms of years of life saved and economic benefits than most medical interventions for smoking-related illnesses.8 There is good evidence for the effectiveness of brief, therapist-delivered interventions, such as advice from a clinician.9 There appears to be additional benefit from more intensive behavioural interventions, such as group therapy,10 individual counselling11 and telephone counselling.12 However, these more intensive therapies are usually dependent on a trained professional delivering or facilitating the interventions. This is both expensive and time-consuming for the health providers, and often inconvenient to the 403 patient, because of lengthy waiting times and the need to take time off work. Another major limitation of these more intensive types of interventions is that they reach only a small proportion of those who smoke. It is estimated that in 2009, there were 1.73 billion Internet users worldwide,13 and the number of Internet users is likely to increase rapidly over a relatively short time-frame.14 The Internet has the potential to deliver behaviour change interventions.15-18 Internet-based material is an attractive intervention tool, because of relatively low costs per user, resulting in high cost-effectiveness.17 The Internet can be accessed in people's homes, in public libraries and through other public Internet access points, such as Internet cafes and information kiosks. The Internet is available 24 hours a day and 365 days a year, even in areas where there are not the resources for a smoking cessation clinic (such as some rural or deprived areas and low-income countries). Online treatment programmes are convenient from the users' perspective, because content can be accessed at any time, and they also offer a greater level of anonymity than in-person or phone-based counselling. They have the potential to reach audiences who might not otherwise seek support, because of limited health care provision or possible stigmatisation. Existing smoking cessation services such as advice from health professionals and nicotine replacement therapy (NRT) are under-utilised by young people.19 Internet use by young people has grown exponentially and has a powerful role in influencing youth culture. It may therefore reach a target population of young people who smoke more effectively than the more traditional providers are able to do. 15.2 Theoretical considerations 15.2.1 Potential benefits The Internet is a promising vehicle for delivering smoking cessation treatment either as a stand-alone programme or as an adjunct to pharmacotherapy.17 18 According to the Pew Internet & American Life Project,20 seven per cent of adult US Internet users, (approximately eight million people), reported having 404 searched online for information on 'how to quit smoking'. In the US, 18% of those with less than high school education searched the web for information on how to quit smoking, which represents a higher proportion than those with more education.20 Materials tailored for individual smokers are more effective than untailored ones, although the absolute size of the effect is still small.21 Internet programmes can be highly tailored to mimic the individualisation of one-to-one counselling. A web-based programme that collects relevant information from users and tailors the intervention to their specific needs had significant advantages over a web-based, non-tailored cessation programme.22 A Dutch study explored existing self-help materials which are currently available in the Netherlands, and found them to be ineffective for smoking cessation. However, this study suggested that computer-tailored interventions could potentially be successfully designed, and may be a promising means of communicating information on smoking and cessation.23 Another study on the efficacy of web-based tailored behavioural smoking cessation materials for nicotine patch users showed that participants in the tailored programme reported significantly higher 12-week continuous abstinence rates (22.8%) than those in a non-tailored programme (18.1%).22 15.2.2 Potential risks Using the Internet for smoking cessation programmes may also have limitations. There are a large number of smoking cessation web sites, but they do not all provide a direct intervention. Some studies of popular smoking cessation web sites and their quality24 25 suggest that smokers seeking tobacco dependence treatment online may have difficulty discriminating between the numerous sites available.25 In addition, web sites that provide direct treatment often do not fully implement treatment guidelines and do not take full advantage of the interactive and tailoring capabilities of the Internet.24 Furthermore, a study on rates and determinants of repeat participation in a web-based health behaviour change programme suggested that such programmes may reach those who need them the least. For example, older 405 individuals who had never smoked were more likely to participate repeatedly than those who currently smoke.26 The Internet is also less likely to be used by people on lower income, who are more likely to smoke.27,28 There have been two recent reviews of this area.29 30 Myung et al. evaluated both web-based and computer-based cessation programmes.29 The review included 22 trials in a single random effects meta-analysis that showed a significant effect on cessation; similar effects were also seen in a range of subgroups including one limited to nine trials that used a web-based intervention. They concluded there was evidence to support the use of web and computer-based cessation programmes for adults who smoked, but not adolescent smokers. Shahab et al focused on interactive online interventions, and also sought to identify treatment effect moderators and mediators.30 Their review included eleven randomised trials. The authors concluded that web based tailored and interactive interventions increased abstinence compared to booklet or email control, based on three trials. Both reviews identified large amounts of heterogeneity in both study designs and effect sizes. Our review focused on determining the effectiveness of Internet-based interventions for smoking cessation. Details of the methods employed can be found in our Cochrane review.31 15.3 Main findings 15.3.1 Description of studies Twenty studies met the inclusion criteria. Trialists were mostly based in the US and recruited participants based there. One trial was conducted in Switzerland, one in Norway, one in the Netherlands, one in England and one in the Republic of Ireland. The studies by Munoz and colleagues recruited from multiple countries.32 406 Recruitment was mainly web-based; participants found the sites through search engines and browsing. Several trials used press releases, billboards, television advertisements and flyers in addition to the web-based recruitment. Therefore, participants included in these trials were smokers motivated to quit smoking, who chose the Internet as a tool for smoking cessation support. Clark et al33 recruited people undergoing chest computerised tomography (CT) as a screening assessment for lung cancer at their first follow-up visits. Strecher et al.34 recruited from members of two health management organisations (HMOs) participating in the National Cancer Institute's Cancer Research Network, Swan et al.35 recruited participants from a large healthcare organisation. Sixteen studies recruited a full adult age range. An et al.36 recruited young adult college students and three studies recruited adolescents.37-39 Sample sizes ranged from less than 15037 38 to nearly 12,000.40 In some studies, participants were offered financial compensation for completing assessment surveys and for biochemical analysis.32 36 37 38 41 47 A range of types of Internet interventions was tested in the included studies, from a very low intensity intervention, providing a list of web sites for smoking cessation,33 to highly intensive interventions consisting of Internet, email and mobile phone delivered components.42,43 Tailored Internet interventions differed by the amount of tailoring, from a bulletin board facility,44 a multimedia component (McKay 2008),45 tailored and personalised access 22,46 ) to very high-depth tailored stories and highly personalised message sources.34 Sixteen studies assessed smoking status at least six months after the start of the intervention.15,32-39, 41-43,45-48 All of these except two34 41 also assessed smoking at one or more intermediate follow-up points. Four studies followed participants for less than six months.17,22,40,44 Studies reported a range of definitions of abstinence at the time of follow-up. Where studies reported abstinence rates for more than one definition we displayed the effect using the most conservative outcome (with the exception of An et al36). For most studies, seven-day smoking abstinence was the main outcome measure. 15,17,32-34,40,41,44,45,47,48 A few studies reported 30-day self-reported smoking 407 abstinence.35,38,39 One study assessed six-month prolonged abstinence from smoking;36 this study also reported seven-day and 30-day prevalence abstinence. We used 30-day rates as our primary outcome because the programme did not involve setting a quit date, and the prolonged abstinence was based on self-report of time since last cigarette rather than repeated assessments of abstinence. Three of the four short-term studies assessed self-reported point prevalence abstinence at three-month follow-up only17,40,44 whilst Strecher et al.22 assessed 28-day continuous abstinence rates at sixweek follow up, and 10-week continuous abstinence rates at 12-week follow up. In one study, seven-day smoking abstinence was a secondary outcome, while time spent on the website, utilisation of pages, cessation aids used in the past and during the study period were the main outcome measures.44 Due to the limited face-to-face contact and due to data collection through Internet or telephone interview, biochemical validation to confirm self-reported smoking abstinence was conducted in only five trials.15,33,36,38,39 All these measured carbon monoxide (CO) in expired air. Utilisation of the Internet site or programme use was a secondary outcome measure in 10 studies.15,17,33-35,38,43,45,46,48 Satisfaction with treatment condition was assessed in five trials.22,37,41,44,48 Use of NRT or other pharmacotherapies was a secondary outcome measure in five trials. 34,35,39,43,45 Two of the studies in adolescents assessed reductions in the number of cigarettes or in smoking frequency as secondary outcomes.37,39 Munoz et al48 reported the impact of method of obtaining follow-up data on quit rates; comparing phone with online follow-up procedure. We first grouped studies in subgroups according to whether: (1) they compared an Internet intervention with no intervention or a non-Internet intervention; or (2) compared a more complex or tailored Internet intervention with a less complex one. At the suggestion of peer reviewers we added a third comparison that included studies in which the intervention offered access to an interactive website and the control could be either a static website, or a control without Internet access. In five trials, all participants were using, or 408 were offered, pharmacotherapy15,22,34,35,43 and the Internet component was thus being evaluated as an adjunct to pharmacotherapy. These were grouped in comparisons based on the nature of the Internet component and the control. 15.3.2 Internet intervention compared to no Internet intervention or no intervention at all Ten trials compared an Internet intervention to a non-Internet based smoking cessation intervention or to a no intervention control.15,17,33,35-39,42,43 Six trials recruited adults, one targeted young adult university students 36 and three were conducted among adolescents.37-39 Of the six trials that targeted adults, two studies 42,43 evaluated 'Happy Endings', a 1-year programme delivered via the Internet and cell phone, consisting of more than 400 contacts by email, web pages, interactive voice response, and short message service (SMS) technology. Brendryen et al 42 recruited people attempting to quit without NRT, whilst in their other trial 43 they offered a free supply of NRT to all participants. Follow-up was after 12 months in both studies. Japuntich et al15 evaluated a web-based system incorporating information, support and problem-solving assistance; this was tested as an adjunct to bupropion and brief face-to-face counselling, with follow up after six months. Swartz et al.17 recruited participants via worksites to test a video-based Internet site that presented strategies for smoking cessation and motivational materials tailored to the user’s race/ethnicity, sex and age. This study only reported short follow up (i.e. at 90 days); after this time the control group had access to the programme. Clark et al.33 tested a very low intensity intervention for smokers having CT lung screening; a handout with a list of 10 Internet sites related to stopping smoking with a brief description of each site compared to printed self-help materials. Follow-up was at 12 months. Swan et al.35 was a three arm trial comparing an established proactive telephone counselling intervention, an interactive website based on the same programme, and a combination of phone and Internet components, all providing behavioural support in conjunction with 409 varenicline use. An et al36 recruited college students who reported smoking in the past 30 days; the sample smoked an average of four cigarettes per day. Intervention group participants received $10 a week to visit an online college magazine that provided personalized smoking cessation messages and peer email support. The control group received only a confirmation email containing links to online health and academic resources. Both groups were informed about a campus-wide ‘Quit&Win’ contest sponsored by University Health Service. The three small studies in adolescents recruited populations of relatively light smokers. Mermelstein et al38 evaluated the effectiveness of enhancing the American Lung Association’s Not on Tobacco programme (NOT) with a webbased adjunct (NOT Plus), which included access to a specially designed web-site for teenagers, along with proactive phone calls from the group facilitator to the participant. Twenty-nine high schools were randomly assigned to either the NOT programme alone or to the NOT Plus one. Selfreported smoking behaviour was assessed at the end of the 10-week programme and three months later. Patten et al39 compared a home-based, Internet-delivered treatment for adolescent smoking cessation with a clinicbased brief office intervention (BOI) consisting of four individual counselling sessions. Adolescents assigned to the Internet condition had access to the web-site for 24 weeks and abstinence was assessed at the end of this period. Woodruff et al.37 evaluated an Internet-based, virtual reality world combined with motivational interviewing, conducted in real-time by a smoking cessation counsellor. There was a measurement-only control condition involving four online surveys. Smoking status was assessed at baseline, postintervention and at three- and 12-month follow-up. 15.3.3 Comparisons between different Internet interventions Ten trials compared two or more different Internet interventions.22,32,34,40,41,44-48 Three of these studies had only short follow-up.22,40,44 In two studies all participants were using nicotine patch pharmacotherapy.22,34 410 A series of three studies by Munoz and colleagues evaluated adjuncts to an online resource, the 'Guia', a National Cancer Institute evidence-based intervention first developed for Spanish-speaking smokers. In separate English language47 and Spanish32 studies by Munoz et al the control condition was the provision of access to the Guia and 'Individually Timed Educational Messages' (ITEMs). The intervention tested was the addition of an online mood management course consisting of eight weekly lessons. Munoz et al48 also used the 'Guia' as the control condition, but in a four-arm design that evaluated the successive addition of ITEMs; the mood management condition used in their earlier studies;32,47 and a 'virtual group' asynchronous bulletin board. The study recruited English- and Spanish-speaking Internet users from 68 countries. In two trials,22,46 the control condition provided access to a relatively static Internet site whilst one or more intervention conditions provided more tailored and personalised access. Rabius et al46 compared five tailored and interactive Internet services with the targeted, minimally interactive American Cancer Society site providing stage-based quitting advice and peer modelling. Followup surveys were conducted four and 13 months after randomization. Strecher et al.22 assigned purchasers of a particular brand of nicotine patch to receive either web-based, tailored behavioural smoking cessation materials or webbased non-tailored materials. This study measured smoking behaviour at three-month follow-up. The remaining five studies were distinctive in their designs. Etter et al.40 compared the efficacy of two versions of an Internet-based, computer tailored cessation programme; the control group received a shorter version modified for use by those smoking and buying NRT over the counter, although use of NRT was not a condition for enrolment. Follow-up was at 2.5 months. Stoddard et al.44 evaluated the impact of adding a bulletin board facility to the 'smokefree.gov' cessation site. This study measured smoking behaviour three months after enrolment. Strecher et al.34 identified active psychosocial and communication components of a web-based smoking cessation intervention and examined the impact of increasing the tailoring 411 depth on smoking cessation among nicotine patch users. Five components of the intervention were randomised using a fractional factorial design: high versus low depth tailored success story, outcome expectation versus efficacy expectation messages; high versus low personalised source; and multiple versus single exposure to the intervention component. Abstinence was assessed after six months. McKay et al.45 compared the Quit Smoking Network (QSN), a web-based tailored cessation programme with a multimedia component, with Active Lives, a web-based programme providing tailored physical activity recommendations and goal setting in order to encourage smoking cessation. Abstinence was assessed after six months. Te Poel et al.41 compared tailored to untailored cessation advice letters sent by email, after participants had completed an online survey. We included this study in the review because information was gathered via a website. 15.3.4 Risk of bias in included studies Allocation The majority of studies did not explicitly describe the way in which the randomisation sequence was generated or concealed until patient enrolment. In eleven studies, computer randomisation was used to assign participants to intervention or control condition.17,22,32,35,40-44,47,48 Although there was little information about allocation concealment, when investigators used computerised randomisation and had minimal interaction with participants we judged there to have been a low risk for selection bias. Munoz et al32,4748 used the baseline questionnaire to automatically implement stratified randomisation by gender and major depressive episode (MDE) status (no MDE history, past MDE, current MDE) to the two conditions. In the two studies37,38 that randomised schools to conditions there was the potential for bias due to the way in which individual students were recruited once their school was randomised. In both there were differences in the baseline smoking behaviour of intervention and control participants. The two studies also needed to take account of the non-independence of outcomes for students clustered within schools. Mermelstein et al. linear modelling to allow for clustering. Woodruff et al. 37 38 used hierarchical assessed baseline 412 variable intraclass correlations and average cluster sizes. Intraclass correlations were generally small (0.1 or less) and the magnitude of the effect sizes was below two, so analyses were conducted at the individual level without a school-level cluster term. Incomplete outcome data All studies included in this review used ITT analysis, reporting analyses based on the total number randomised, with drop-outs and participants lost to follow up classified as smoking. In one study33 there were no drop-outs at one-year follow up, as all study participants attended their annual review at that point. Five studies 15,34,36,42,43 ascertained smoking status for over 80% of participants at follow-up. Five studies ascertained smoking status for 50-80% of participants at follow-up.17,22,35,37-39 Eight studies had over 50% loss to follow up.32,40,41,44-48 All studies reported similar proportions loss to follow up in each group except in one study where survey non-response was higher among intervention participants then among controls.37 We tested the sensitivity of our results to the assumption that lost participants were still smoking, by excluding those not reached at follow-up from the denominator in the analyses, and report this below. 15.3.5 Effects of interventions Internet intervention compared to no Internet interventions or no interventions at all There were six trials in adult populations.15,17,33,35,42,43 Both the Happy Endings trials detected a significant effect of the Internet intervention on sustained abstinence at 12 months whether it was compared to a self-help control,42 RR 2.94, (95%CI 1.49, 5.81) or tested as an adjunct to NRT43 RR 1.71, (95%CI 1.10, 2.66). Whilst the relative benefit of the intervention was similar, prolonged abstinence at 12 months was higher, 13%, amongst control group participants in Brendryen et al.,43 who were offered NRT, than amongst controls who were not given pharmacotherapy (seven per cent in Brendryen et al.).42 The authors noted that the relative effects were smaller for the outcome of point prevalence abstinence at 12 months, rather than repeated abstinence, because abstinence increased over time in the control group in 413 both trials. Japuntich et al.15 did not detect a significant effect of the Internet component as an adjunct to counselling and bupropion (RR 1.27, 95%CI 0.70, 2.31) whilst Swan et al.35 did not detect an effect of an Internet component either compared to, or as an adjunct to, telephone counselling and varenicline (RR 0.94 95%CI 0.79, 1.13, combining web only and web plus phone arms). All three conditions achieved similar quit rates, ranging from 27.4% to 30.6% for 30 day abstinence at six months. Clark et al.,33 which was only a minimal intervention, also did not detect an effect (RR 0.45, 95%CI 0.14, 1.40). Relative effects were similar in these trials at shorter follow-up points. The study by Swartz et al.17 with only short-term follow-up detected a significant effect of Internet intervention compared to no intervention at all (RR 2.46, 95%CI 1.16, 5.21). The trial by An et al.36 in a population of college students detected a significant effect on 30-day abstinence at 30-week follow-up (RR 1.95, 95%CI 1.42, 2.69) although rates of prolonged abstinence were only six per cent and did not differ between groups. Patten 200639 compared a home-based Internet delivered intervention (SOS) to a brief office intervention (BOI) for adolescent smoking cessation, and did not detect a difference in abstinence. Rates at 24 and 36 weeks follow-up were higher for BOI (RR 0.44, 95%CI 0.14, 1.36 at 36 weeks). Mermelstein et al.38 detected a significant effect of the web-based adjuncts to the groupbased approaches for adolescent smoking cessation (crude RR 1.96, 95%CI 1.02, 3.77; also reported as significant, (P<0.05), using mixed model logistic regression to account for clustering within schools). Woodruff et al.37 recruited eligible adolescents based on a report of smoking in the past month; at baseline some described themselves as 'former' smokers or had not smoked in the past week. Intervention participants had lower past week abstinence rates at baseline than controls (14% vs. 29%). At the post-assessment, they had significantly higher abstinence rates than controls (35% vs. 22%), but by the final 12-month follow-up, the two groups had almost identical past-week abstinence rates (RR 0.93, 95%CI 0.60, 1.44). The interaction term considering all four assessments was not significant. Intervention participants 414 (68%, n=52) completed a five item questionnaire assessing their satisfaction with the programme immediately after the post-test assessment; 89% of participants reported they would recommend the programme to another person who smoked. Comparison of different Internet intervention None of the three Munoz studies detected significant benefits of adding a mood management intervention, and pooling the comparable arms did not make a difference at 12 months (RR 0.90, 95%CI 0.70, 1.15 ; I²=51%) or in the shorter term. In fact Munoz et al.47 found that the more complex intervention (GUIA + ITEMS + MM) yielded significantly lower quit rates at 12 months. All four arms of Munoz et al.48 had similar long- and short-term quit rates, ranging from 19.1% to 22.7% at 12 months, with no evidence that the more tailored conditions had any incremental benefit over the static website control. Authors of this trial also assessed user satisfaction with the treatment condition, and reported that there were significant differences in satisfaction across conditions at all follow-up assessments, with groups 3 and 4 generally reporting greater satisfaction at each time point. Strecher et al.22 and Rabius et al.46 also compared tailored to static sites but Strecher 200522 used pharmacotherapy and only reported short-term outcomes, and Rabius et al.46 in their later trial compared multiple sites. Rabius et al.46 detected no significant differences between a static cessation site and any of the other Internet sites included in the evaluation. Approximately 10% of enrolled participants assigned to the static site reported abstinence after 13 months compared with 8% to 12% among those assigned to any of the five different interactive sites (RR 1.12, 95%CI 0.92, 1.36 pooling any interactive website versus the control site). At baseline, 30% of participants reported an indicator of depression. Post-hoc analyses found that this subgroup had significantly lower 13-month quit rates than those who did not have signs of depression (8% vs. 12%, P <0.001). Amongst the 70% of participants who did not report an indicator of depression at baseline, the more interactive, tailored sites, as a whole, were associated with higher quitting rates than the less interactive American Cancer Society site (13% vs. 415 10%, P =0.04). This exploratory analysis suggested that tailored, interactive websites might help smokers who do not report the indicator of depression at baseline to quit and maintain cessation. Strecher et al.22 reported higher continuous abstinence rates in the tailored condition than the control, as an adjunct to NRT. At 12 weeks, continuous abstinence rates were 22.8% vs. 18.1% respectively (RR 1.26, 95%CI 1.10, 1.44). Satisfaction with the programme was also significantly higher in the tailored than in the non-tailored programme. McKay et al.45 detected no difference at three or six-month follow-up between two cessation interventions, one with a focus on physical activity presenting very few strategies for quitting smoking. Te Poel et al.41 detected evidence of a benefit from a tailored email letter compared to a non-tailored one, (RR 2.48, 95%CI 1.11, 5.55). All participants in this study accessed a website once to complete a questionnaire about their smoking. Stoddard et al.44 compared the publicly available version of smokfree.gov, designated as usual care condition (UC), to an identical-looking website that included an asynchronous bulletin board (BB) and found that quit rates for participants in both conditions were similar after three months (RR 0.95, 95%CI 0.64, 1.40). Satisfaction with the website was high and did not differ significantly between conditions (UC: 90.2%, BB: 84.9%, P=0.08). Strecher et al.34 compared multiple conditions in a fractional factorial design so results are not displayed in the analyses. Participants could receive up to three high-depth components, addressing efficacy expectations, outcome expectations and success stories, as part of their tailored web-based intervention. Tailoring depth was marginally related (P<0.08) to 6-month smoking cessation in ITT analyses and was significantly related to cessation using per-protocol analysis (excluding participants who used other cessation aids during the follow-up period). The adjusted 6-month cessation rates among participants receiving all the three high-depth tailored components was 416 27.7% in the ITT analysis. There was some evidence that high-depth tailored success stories had a particular influence on participants with lower levels of education although this interaction was not significant in the ITT analysis. Participants reported using an average of 5.1 weeks of their supply of nicotine patches, with 26.7% using the patch for the full 10 weeks. Comparison of interactive Internet sites with static sites or non-Internet controls From the studies discussed above, eight studies with long-term outcomes32,36,42,43,45-48 and another three with only short-term17,22,40 yielded data that allowed pooling to assess whether sites that give smokers content tailored to their needs and interests are more useful than a control. These summary statistics do however need to be interpreted with caution because of a high degree of statistical heterogeneity (I²=75% for long-term outcomes, Figure 15.1, I²=76% for short-term outcomes, Figure 15.2). Figure 15.1 Meta-analysis of effectiveness of tailored interactive vs. nontailored/non-Internet interventions on smoking abstinence at the longest follow-up assessment point 417 Figure 15.2 Meta-analysis of effectiveness of tailored interactive vs. nontailored/non-Internet interventions on smoking abstinence at the shortterm follow-up assessment 15.4 Implications for policy, practice and research The Internet, with all its richness of options and opportunities for communication and sharing information, has now become a regular part of daily life for the majority of people in many countries. Therefore it is appropriate to consider using it as a tool to increase choice and access to smoking cessation support. Online treatment is convenient in that it can be accessed anywhere, at any time; it also offers the option of anonymity. For healthcare providers it has the potential for being very cost-effective if provided as an automated service. Internet interventions for smoking cessation can be provided in conjunction with other cessation support such as individual or group counselling and NRT or other pharmacotherapy. We identified 20 trials yielding data from nearly 40,000 participants. Despite this large volume of data, this remains a relatively new field of research, with all trials published since 2004. The studies included in this review assess ways to help people quit smoking using a variety of Internet cessation programmes. Trials varied in the interventions studied and in the duration of the interventions; they also varied in the timing of follow-up assessments and the way in which abstinence was defined. 418 There are only a small number of studies comparing the long-term effects of Internet to a no-Internet or no intervention control. The results of these studies have been mixed so there is as yet only limited evidence to support a longterm effect of programmes that use the Internet. Two studies assessed whether an Internet intervention was more effective than a self-help booklet for smoking cessation.42,43 These have shown evidence of both short- and longer-term effect with up to one-year follow-up, but the programme delivered to the treatment group was fully automated, relatively intensive and proactive, and was delivered via both the Internet and mobile phone. The three other studies reporting long-term results did not detect a benefit, but these were relatively less intensive programmes with fewer proactive elements to encourage programme use. One additional shortterm study providing tailored online cessation videos did detect an increase in abstinence at three months.17 When considering trials comparing different intensities of Internet support, there is some evidence of a short-term beneficial effect of individually tailored Internet programmes compared to static web-sites or non-tailored programmes. For example, Etter et al found that an original more tailored programme versus a modified and less tailored programme increased quit rates and helped maintain short-term abstinence.40 Strecher et al found that continuous abstinence rates at six and 12 weeks were significantly higher in the tailored intervention than in the non-tailored intervention.22 In contrast, however, one study found there was no significant difference in abstinence rates between participants assigned to interactive sites and participants assigned to static sites, even though interactive sites led to higher utilization of the websites.46 There are also reported benefits of interactive sites with regard to satisfaction of the users.22,37,48 One form of tailoring is to encourage website use by sending tailored email messages. Three studies by Munoz and colleagues used these as a component of an Internet programme, but none of them evaluated the effect compared to a non-Internet control, and only Munoz et al.48 used a static site as a control; this study failed to detect any additional benefit of this particular tailoring. All three studies evaluated the 419 effect of adding a mood management component and none detected any evidence that this was helpful, and the trend in two studies was for a reduction in success. In a post-hoc analysis we explored pooling studies that evaluated tailored, interactive Internet-based interventions, whatever the nature of the control, but found there to be too much between study heterogeneity in effect sizes. Pharmacotherapies such as NRT and bupropion can help people who make a quit attempt increase their chances of success.49,50 Face-to-face behavioural support has an independent benefit, but many people are not willing to attend or cannot access group-based programmes or multi-session counselling. Internet programmes can be used to deliver additional behavioural support to people using pharmacotherapy, but there is so far little evidence of an effect of Internet intervention in addition to pharmacotherapies. Both Japuntich et al15 and Swan et al.35 failed to detect any benefit, although Strecher et al.22 suggested a short-term advantage of a tailored site over a static one. Strecher et al in their later trial34 identified components of tailored sites that could assist cessation. Studies conducted among adolescents and young adults vary not only in the Internet intervention used, but also in methodology and setting. All three trials in adolescents were relatively small. Two suggested that the Internet condition had less benefit than the comparison condition, while one reported a marginally significant benefit sustained after the end of the programme.38 In this study, participants in the web arm also had phone calls from a counsellor so the independent effect of the web component is uncertain. These studies show that Internet assistance is attractive and suggest that tailored web sites are more popular among young people.36-38 To achieve success in smoking cessation programmes, interventions must be accessible, efficacious and cost-effective and transportable. Only two of the included studies in this review reported any information about costeffectiveness of their intervention.40,46 Etter et al estimated that the total cost 420 of implementing the website, for a reach of 8000 participants in computer tailored programmes and for 600,000 visitors per year to the website, is comparable to the cost of running a small smoking cessation clinic which would treat about 50 smokers a month. Therefore, Internet services provide greater potential for cost-efficiency because they can provide assistance to many smokers at a very low cost. Internet interventions can also be delivered alongside other more traditional smoking cessation programmes, providing smokers who are motivated to quit smoking with different tools which increases their overall choice.40 Rabius et al found Internet assistance for smoking cessation to be cost-effective, since four days of programming at a cost of less than US $2000 allowed approximately 5000 additional users for services from the five tailored interactive service providers, comparing with the much larger cost of serving 1000 new clients with telephone counselling (approximately US $100,000).46 The trials enrolling adults generally relied on self-reported data on smoking status. Biochemical validation of self-reported cessation was only attempted in two trials where participants had face-to-face contact during the follow-up visit.15,33 The Society for Research on Nicotine and Tobacco subcommittee on biochemical verification in clinical trials51 considers that verification is not necessary when a trial includes a large population with limited face-to-face contact, and where the optimal data collection methods are through the mail, telephone, or Internet. There is a recommendation that biochemical verification be used in studies of smoking cessation in special populations, including adolescents.51 Only one of the four studies in adolescents and young people did not use biochemical verification of self-reported abstinence.37 Four included studies followed participants for less than six months.17,22,40,44 It is hoped that reporting six-month outcomes will become routine.52 Conducting research via the Internet provides opportunities to generate very large sample sizes, but it is also methodologically challenging, because of threats to internal and external validity such as selection bias or differential drop-out.53,54 Although there was limited detail about procedures for sequence 421 generation and allocation concealment, we judged that the likelihood of selection bias was small in studies that recruited participants online. Rates of loss to follow-up were varied and were high in some large online studies. In our primary analyses we followed the convention of assuming that all those lost to follow-up continued to smoke. This can be argued to be a reasonable approach with a volunteer sample and low attrition but if attrition rates are high, or differential across conditions, the assumption may be wrong in some cases and introduce bias.55 We undertook a sensitivity analysis ignoring losses to follow-up by omitting them from the denominators, and did not find any important differences in relative effects. A range of alternative assumptions about those lost to follow-up could be tested, but it is important to recognise that bias in the relative effect will only occur if there the proportion of quitters amongst those lost to follow-up differs between intervention and control group. Determining the contribution of a specific website presents a difficult challenge, since Internet users appear to access different sites when searching for information or support to a specific topic. For example, contamination in control groups may be difficult to prevent because of unrestricted access to the Internet, whilst on the other side we cannot be sure that the intervention group is using only the intended intervention.53,56 The two other recent reviews in this area drew somewhat less cautious conclusions about the strength of the evidence for the effectiveness of Internet interventions. One review29 pooled data from a number of studies of both web- and computer-based interventions, and concluded that there is now sufficient evidence to support the use of both categories of intervention for adult smokers. Their estimate for Internet interventions based on nine studies, using a random effects model, was RR 1.40, 95%CI 1.13, 1.72. Shahab et al also suggest that interactive, web-based cessation can be effective in aiding cessation,30 based on 11 studies, all but one of which is included in our review (we were able to include longer-term data for the studies by Pike et al.57 and Rabius et al.46). We excluded the study by Prochaska et al,58 because we were unable to confirm all data with the authors. Shahab et al.30 pooled the 422 studies in a number of subgroups; the intervention (tailored/untailored); length of treatment; motivation to quit; and whether the intervention was fully automated or not. They estimated interactive web-based smoking cessation interventions to be effective compared to untailored booklet or e-mail interventions (random effects RR 1.8, 95%CI 1.4, 2.3), but this was based on just three trials; they also estimated that tailored interventions increase sixmonth abstinence by 17%. They also suggest that only interventions aimed at smokers motivated to quit were effective (RR 1.3 95%CI 1.0, 1.7). We consider that our more cautious conclusions are due to a conservative approach to subgroup analyses, and our preference not to pool, even with a random effects approach, when there is evidence of substantial heterogeneity. Further studies of the long-term effects of Internet-based cessation interventions are clearly needed, and there are several ongoing studies in this area. Future trials and reviews should include analyses of participants according to socio-demographic data, in order to be able to identify the types of smokers who seek Internet assistance in quitting smoking. Feil et al detected that a large proportion of women was recruited in their study.53 According to our findings there is some evidence that females are more interested in smoking cessation programmes delivered via Internet; only three of the included trials reported that more males enrolled,32,37,47 and three had equal number of male and female participants.33,39,43 In the future there may be an interest from patients with depression seeking Internet assistance for quitting smoking. Although there is evidence that depression is an important factor in smoking cessation,59 and that depression inhibits quitting success by decreasing self-efficacy,60 only a few studies evaluated the impact of Internet interventions among the subgroup of smokers reporting depression. Rabius et al found no overall difference in quit rate among smokers assigned to six experimental groups (five interactive and one static site), but they also found that those who reported an indicator of depression and were assigned to interactive site had lower cessation rates than those assigned to the static site, although this difference was not 423 significant. The authors attribute these findings to the increased time investment required from participants of interactive sites.46 There are a small number of studies which provide very limited evidence of long-term benefits for programmes delivered only by the Internet compared to no-Internet controls. There is some evidence that tailored Internet interventions are more effective than non-tailored interventions. This should therefore not for the present be a priority for the NHS. Looking ahead, more rigorous studies comparing the long-term effects of Internet interventions with non-Internet interventions or no intervention at all are needed in order to determine the true long-term effectiveness of the Internet as a tool for smoking cessation. These should, where possible, assess outcomes using objective measures, and also assess costeffectiveness considerations. It is important that future trials also seek to describe the likely mechanisms through which these interventions may (or may not) be exerting their effects–they should therefore also report data on patient satisfaction, changes in knowledge, motivation, dependency, quit attempts and safety considerations. Researchers should aim to assess smoking status after six months as a minimum, so that the longer-term benefit of programmes can be determined and meta-analyses of outcomes across studies be facilitated. Acknowledgements: This chapter is based on a Cochrane review (Civljak M, Sheikh A, Stead LF, Car J. Internet-based interventions for smoking cessation. Cochrane Database Syst Rev. 2010 Sep 8;(9):CD007078) and we take pleasure in acknowledging our gratitude to Marta Civljak and Lindsay Stead for their expert contributions to this work; our thanks are also due to colleagues in The Cochrane Tobacco Addiction Group for their support. Permission to reproduce this material has been applied for. 424 References 1. World Health Organization. Building blocks for Tobacco Control - A Handbook. Serial, 2004. 2. Mackay J, Eriksen M, Shafey O. The Tobacco Atlas. 2nd Edition. American Cancer Society, 2006. 3. Parrot S, Godfrey C. Economics of smoking cessation. BMJ 2004; 328: 947-49. 4. Scientific Committee on Tobacco and Health and HMSO. Report of the Scientific Committee on Tobacco and Health. London: Her Majesty's Stationery Office, 1998. 5. United States Public Health Service. The Health Consequences of Smoking Nicotine Addiction: a report of the Surgeon General. Office of the Surgeon General 1988. 6. Schroeder SA. Conflicting dispatches from the tobacco wars. NEJM 2002; 347: 1106-09 7. West R. Assessment of dependence and motivation to stop smoking. BMJ 2004; 328: 338-39. 8. Coleman T. Cessation interventions in routine health care. BMJ 2004; 328: 63133. 9. Stead LF, Bergson G, Lancaster T. Physician advice for smoking cessation. Cochrane Database Syst Rev 2004; 2: 10.1002/14651858.CD000165.pub3 10. Stead LF, Lancaster T. Group behaviour therapy programmes for smoking cessation. Cochrane Database Syst Rev 2005; 2: 10.1002/14651858.CD001007.pub2 11. Lancaster T, Stead LF. Individual behavioural counselling for smoking cessation. Cochrane Database Syst Rev 2005; 2: 10.1002/14651858.CD001292.pub2 12. Stead LF, Lancaster T, Perera R. Telephone counselling for smoking cessation. Cochrane Database Syst Rev 2006; 3: 10.1002/14651858.CD002850.pub2 13. Internet 2009 in numbers. Available from: http://royal.pingdom.com/2010/01/22/internet-2009-in-numbers/ (last accessed 2/1/11) 14. Modis T. The end of the internet rush. Techn Forecast Soc Change 2005; 72: 938-43. 15. Japuntich SJ, Zehner ME, Smith SS, Jorenbv DE, Valdez JA, Fiore MC, et al.Smoking cessation via the internet: a randomized clinical trial of an internet intervention as adjuvant treatment in a smoking cessation intervention. Nicot Tobac Res 2006; 8: S59-67. 16. Strecher VJ, Shiffman S, West R. Moderators and mediators of a web-based computer-tailored smoking cessation program among nicotine patch users. Nicot Tobac Res 2006; 8: S95-101. 17. Swartz LH, Noell JW, Schroeder SW, Ary DV. A randomised control study of a fully automated internet based smoking cessation programme. Tobac Contr 2006; 15: 7-12. 18. Graham AL, Cobb NK, Raymond L, Sill S, Young J. Effectiveness of an internetbased worksite smoking cessation intervention at 12 months. J Occ Environ Med 2007; 49: 821-28. 19. Rodgers A, Corbett T, Bramley D, Riddell T, Wills M, Lin RB, et al. Do u smoke after txt? Results of a randomised trial of smoking cessation using mobile phone text messaging. Tobac Contr 2005; 14: 255-61. 20. Fox S. Health information online: eight in ten internet users have looked for health information online, with increased interest in diet, fitness, drugs, health insurance, experimental treatments, and particular doctors and hospitals. Available from: http://www.pewinternet.org/Reports/2005/Health-InformationOnline/1-Summary-of-Findings.aspx (last accessed 2/1/11) 425 21. Lancaster T, Stead LF. Self-help interventions for smoking cessation. Cochrane Database Syst Rev 2005; 3: 10.1002/14651858.CD001118.pub2 22. Strecher VJ, Shiffman S, West R. Randomized controlled trial of a web-based computer-tailored smoking cessation program as a supplement to nicotine patch therapy. Addiction 2005; 100: 682-88. 23. Dijkstra A, De Vries H, Roijackers J. Targeting smokers with low readiness to change with tailored and nontailored self-help materials. Prev Med 1999; 28: 20311. 24. Bock B, Graham A, Sciamanna C, Krishanamoorthy J, Whiteley J, CarmonaBarros R, et al.Smoking cessation treatment on the internet: content, quality, and usability. Nicot Tobac Res 2004; 6: 207-19. 25. Etter JF. A list of the most popular cessation web sites and a comparison of their quality. Nicot Tobac Res 2006; 8: S27-34. 26. Verheijden MW, Jans MP, Hildebrandt VH, Hopman-Rock M. Rates and determinants of repeated participation in a web-based behaviour change program for healthy body weight and healthy life style. J Med Intern Res 2007; 9: e1. 27. Eysenbach G. Poverty, human development and the role of e Health. J Med Intern Res 2007; 9: e34. 28. Kontos EZ, Bennett GG, Viswanath K. Barriers and facilitators to home computer and internet use among urban novice computer users of low socioeconomic position. J Med Intern Res 2007; 9: e 31. 29. Myung SK, McDonell DD, Kazinets G, Seo HG, Moskowitz JM. Effects of weband computer-based smoking cessation programs: meta-analysis of randomized controlled trials. Arch Intern Med 2009; 169: 929-37. 30. Shahab L, McEwen A. Online support for smoking cessation: a systematic review of the literature. Addiction 2009; 104: 1792-1804. 31. Civljak M, Sheikh A, Stead LF, Car J. Internet-based interventions for smoking cessation. Cochrane Database Syst Rev 2010; CD007078. 32. Munoz RF, Lenert LL, Delucchi K, Stoddard J, Perez JE, Penilla C, et al. Toward evidence-based Internet interventions: A Spanish/English Web site for international smoking cessation trials. Nicot Tobac Res 2006; 8: 77-87. 33. Clark MM, Cox LS, Jett JR, Patten CA, Schroeder DR, Nirelli LM, et al.Effectiveness of smoking cessation self-help materials in a lung cancer screening population. Lung Cancer 2004; 44: 13-21. 34. Strecher VJ, McClure JB, Alexander GL, Chakraborty B, Nair VN, Konkel JM, et al.Web-based smoking-cessation programs: results of a randomized trial. Am J Prev Med 2008; 34: 373-81. 35. Swan GE, McClure JB, Jack LM, Zbikowski SM, Javitz HS, Catz SL, et al.Behavioral counseling and varenicline treatment for smoking cessation. Am Prev Med 2010; 38: 482-90. 36. An LC, Klatt C, Perry CL, Lein EB, Hennrikus DJ, Pallonen UE, et al.The RealU online cessation intervention for college smokers: a randomized controlled trial. Prev Med 2008; 47: 194-99. 37. Woodruff SI, Conway TL, Edwards CC, Elliott SP, Crittenden J. Evaluation of an Internet virtual world chat room for adolescent smoking cessation. Addict Behav 2007; 32: 1769-86. 38. Mermelstein R, Turner L. Web-based support as an adjunct to group-based smoking cessation for adolescents. Nicot Tobac Res 2006; 8 Suppl 1: S69-76. 39. Patten CA, Croghan IT, Meis TM, Decker PA, Pingree S, Colligan RC, et al.Randomized clinical trial of an Internet-based versus brief office intervention for adolescent smoking cessation. Patient Educ Couns 2006; 64: 249-58. 40. Etter JF. Comparing the efficacy of two Internet-based, computer-tailored smoking cessation programs: a randomized trial. J Med Intern Res 2005; 7: e2. 426 41. Te Poel F, Bolman C, Reubsaet A, de Vries H. Efficacy of single computertailored e-mail for smoking cessation: results after 6 months. Health Educ Res 2009; 24: 930-40. 42. Brendryen H, Drozd F, Kraft P. A digital smoking cessation program delivered through internet and cell phone without nicotine replacement (Happy Ending): Randomized controlled trial. J Med Intern Res 2008; 10: e51. 43. Brendryen H, Kraft P. Happy Ending: a randomized controlled trial of a digital multi-media smoking cessation intervention. Addiction 2008; 103: 478-84. 44. Stoddard JL, Augustson EM, Moser RP. Effect of adding a virtual community (bulletin board) to smokefree.gov: randomized controlled trial. J Med Intern Res 2008; 10: e53. 45. McKay HG, Danaher BG, Seeley JR, Lichtenstein E, Gau JM. Comparing two web-based smoking cessation programs: randomized controlled trial. J Med Intern Res 2008; 10: e40. 46. Rabius V, Pike KJ, Wiatrek D, McAlister AL. Comparing internet assistance for smoking cessation: 13-month follow-up of a six-arm randomized controlled trial. J Med Intern Res 2008; 10: e45. 47. Lenner L, Munoz RE, Stoddard J, Delucchi K, Bansod A, Skoczen S, et.al Design and pilot evaluation of an internet smoking cessation program. J Am Med Inform Assoc 2003; 10: 16-20. 48. Munoz RF, Barrere AZ, Delucchi K, Penilla C, Torres LD, Perez-Stable EJ. International Spanish/Inglish Internet smoking cessation trial yields 20% abstinence rates at 1 year. Nicot Tobac Res 2009; 11: 1025-34. 49. Hughes JR, Stead LF, Lancaster T. Antidepressants for smoking cessation. Cochrane Database Syst Rev 2007; Issue 1: 10.1002/14651858.CD000031.pub3. 50. Stead LF, Perera R, Bullen C, MMant D, Lancaster T. Nicotine replacement therapy for smoking sessation. Cochrane Database Syst Rev 2008; Issue 1: 10.1002/14651858.CD000146.pub3]. 51. SRNT Subcommittee on Biochemical Verification. Outcome criteria in smoking cessation trials proposal for a common standard. Nicot Tobac Res 2002; 4: 149159. 52. West R, Hajek P, Stead L, Stapleton J, Outcome criteria in smoking cessation trials: proposal for a common standard. Addiction 2005; 100: 299-303. 53. Feil EG, Noell J, Lichtenstein E, Boles SM, McKay HG. Evaluation of an Internetbased smoking cessation program: Lessons learned from a pilot study. Nicot Tobac Res2003; 5: 189-94. 54. Cobb NK, Graham AL, Bock BC, Papandonatos G, Abrams DB. Initial evaluation of a real-world Internet smoking cessation system. Nicot Tobac Res 2005; 7: 20716. 55. Hall SM, Delucchi KL, Velicer WF, Kahler CW, Ranger-Moore J, Hedeker D, et al. Statistical analysis of randomized trials in tobacco treatment: Longitudinal designs with dichotomous outcome. Nicot Tobac Res 2001; 3: 193-201. 56. Eusenbach G. Issues in evaluating health websites in an Internet-based controlled trial. J Med Internet Res 2002; 4: e17. 57. Pike KJ, Rabius V, McAlister A, Geiger A. American Cancer Society’s Quit Link: randomized trial of Internet assistance. Nicot Tobac Res 2007; 9: 415-20. 58. Prochaska JO, Butterworth S, Redding CA, Burden V, Perrin N, Leo M, et al. Initial efficacy of MI, TTM tailoring and HRI’s with multiply behaviours for employee health promotion. Prev Med 2008; 46: 226-31. 59. Niaura R, Abrams DB. Smoking cessation: progress, priorities, and prospects. J Consult Clin Psychol 2002; 70: 494-509. 60. Haukkala A, Uutela A, Vartiainen E, McCalister A, Knekt P. Depression and smoking cessation: the role, motivation and self-efficacy. Addict Behav 2000; 25: 494-509. 427 Chapter 16 Human factors and human factors engineering Summary • The study of human factors is the science of human behaviour and performance. Human factors engineering is the application of human factors principles and insights for the purpose of matching people’s technological, organisational and social environments to their goals, abilities and needs. • Human factors issues relevant for eHealth include: users’ work practices and workflow; the nature of the tasks to be supported by the eHealth system; users’ capabilities and skills; training programmes; and the wider work and organisational context in which the system will be deployed. • Healthcare has been slow to incorporate understanding of human factors into eHealth projects, despite the increasing dependency on eHealth applications for delivery of care and growing evidence of patients being put at risk. • Users should be involved in all stages of design, development and deployment of eHealth applications. Feedback from users should not only be facilitated, but must also be actively encouraged to ensure that new applications are fitfor-purpose and to minimise risks to patient safety. • Ease of use (“usability”) and fit with working practices are as important as the functionality and reliability of eHealth applications such as the electronic health record, computerised decision support and ePrescribing systems. Confusion and frustration arising from poor usability will interfere with user acceptance, with an adverse subsequent knock-on effect on implementation; it may also endanger the safety of patient care. • Meaningful integration of human factors’ considerations is only likely to be achievable if eHealth projects are approached as a co-production of healthcare and IT professionals, such that users are given the opportunity to be more fully involved throughout the design and development phase. • Embedding human factors engineering principles and thinking is not free; policymakers and commissioners need to ensure that adequate time, resources and prioritisation are given to this so as to maximise the chances of success of eHealth initiatives. 428 16.1 Introduction Despite widespread acknowledgement that involvement of healthcare professionals in the design and development of new medical informatics systems is crucial to their successful implementation and adoption, this is often still not given sufficient attention by system commissioners and Information Technology (IT) professionals. Drawing on insights from human factors’ engineering (HFE), usability research and participatory design, this chapter seeks to provide an overview of current thinking on user involvement in IT projects and the challenges of achieving it effectively in the context of eHealth. We illustrate these challenges with examples of usability issues arising in the design and implementation of a range of eHealth applications, primarily the electronic health record (EHR) and its various associated applications such as computerised decision support systems (CDSS) and ePrescribing. We conclude with suggestions for how the involvement of healthcare professionals in eHealth projects may be strengthened and made more effective. 16.2 Definition, description and scope In the most general sense, human factors (HF) (also sometimes known as ergonomics) is the multi-disciplinary study of how people’s performance is influenced by their technological, organisational and social environments.1 Human factors’ engineering is the methodical application of HF principles to the design of these environments. The goal is to ensure the safe, comfortable, and effective use of technologies.2 To achieve this, HF set out to, in the words of Wilson, ‘… understand people and their interactions, as well as the relationships between these interactions, and to improve those interactions in real-life settings.3 Though the start of HF pre-dates information technology (IT) by many decades, it is now most closely associated with the design of IT-based systems. As IT has become more widely used, so too has the scope of HF widened, from an initial, narrow focus on understanding the physical, 429 perceptual and cognitive aspects of human performance, to a broader agenda that includes the social and organisational contexts in which the work that an IT system is to support is embedded.4 The aim of HFE is to apply HF principles to achieve a good fit between an IT system and its users’ capabilities and needs. The term “usability” is often used as a short-hand for the measure of the quality of this fit. The broad areas of interest and scope of HFE practice today are outlined in Box 16.1. Box 16.1 The scope of human factors Physical and perceptual factors: These include the bounds of human performance such as the accuracy of movements such as pointing, response times to simple stimuli and capacity to discriminate between levels of brightness, different colours and their implications for the design of the physical interface. Cognitive factors: These include human performance relating to the speed of information processing and decision-making, recognition of and memory for information, the causes of errors, learning times and styles. The aim is to inform guidelines for design of user interface layouts, the representation of information, the sequencing of activities and measures to reduce the likelihood and impact of errors. Motivational factors: These focus on people’s attitudes, and beliefs and expectations of technologies and how these may be influenced by a person’s status, role, profession, age, etc. One important aim is to try to arrive at an appropriate allocation of function between user and system such that the former is able to derive satisfaction from using the system while ensuring its safe and effective use. Situational factors: These describe the social and organisational context within which the individual is expected to perform. They include how roles and divisions of labour are managed within a group or team, the collaborative dimensions of the work, how awareness and coordination is achieved and the implications of new technologies and work re-design for safety and reliability of the overall socio-technical system. 430 While technical reliability is an essential requirement for any eHealth application, successful deployment, adoption and dependability in day-to-day use draws on a much wider range of insights and disciplines.1 2 It is the recognition of the socio-technical character of this challenge that has motivated the incorporation of HFE within IT systems design and development. HFE is of critical importance to eHealth, not only to reduce the risks that poor design may pose to patient safety, but also to reduce the risk that new IT systems may be rejected by healthcare professionals.1 5 Studies confirm that when HFE is applied early and consistently throughout an IT project, it can greatly increase the chances of higher productivity and process improvements. It can also provide the foundations for planning an effective deployment strategy, which may over time decrease staffing and training costs and reduce the risks of user resistance to the new system.1 In the case of equipment such as medical devices and surgical tools, where attention to safety is reinforced through standards and regulation,6 evidence of the use of HFE is generally encouraging.7 Influenced by high profile cases such as the Therac-25 accidents,8 it is now widely understood that errors in the use of medical devices often have their origins in poorly designed user interfaces; and that design-induced errors can lead to patient injuries and deaths.9 User interfaces that are misleading or illogical can induce errors even by the most skilled users.9 Good user interface design requires more than the following of HF guidelines on use of colour and screen layout, however. It is equally important to have a detailed understanding of how a device will be used and the work environment to gauge the types of errors that may arise in use and thereby to be able to design to eliminate or, at least, mitigate their impact.6 Beyond the medical device arena, which is to say beyond a focus on physical, perceptual and cognitive aspects of usability, the impact of HFE on the design of IT systems has been less evident, despite the increasing reliance by 431 healthcare providers on IT for the management and delivery of healthcare.10 As IT systems become ever more deeply embedded in healthcare, the challenges that HFE will need to address can only grow in complexity.11 Inadequately conceived objectives, poor design and lack of change management planning continue to lead to wasted investment in new IT systems that are not fit for purpose or to rejection by their prospective endusers;12 for example, the London Ambulance Computer Aided Dispatch System failure,13 where patients’ lives were put at risk (see Box 16.2). Box 16.2 The Ambulance Computer Aided Dispatch System failure The London Ambulance Service Computer Assisted Dispatch (LASCAD) system was intended to replace a manual system and improve communication, location and dispatch of vehicles to improve the timeliness of medical treatment. The system went live on 26th October 1992, covering all of London and stopping use of the manual backup. Although on the first two days all functionality was provided, some response times were less than adequate, and manual intervention to correct problems was difficult. One of the direct results of the failure of the system included an ambulance arriving after the patient had died and had already been taken away by the undertaker. Estimates of the total number of fatalities caused differ from 10 upwards, even through no coronial findings included the late arrival of an ambulance as the direct cause. The investigation blamed a wide range of factors, including technical, managerial, human and environmental issues. Some blame was placed upon incomplete software, and inadequate testing (particularly the lack of adequate load testing) and optimisation of the system the use of tools meant for prototyping and not for safety-critical systems. The human factors aspects included that staff had little or no confidence in the system and had not been trained in its operation. Managerial issues included the lack of change as a result of earlier problems. In this chapter, we provide an overview of the basic principles of usability. We then consider different approaches to the problem of setting usability goals 432 and determining if they are being met. The failure of many healthcare IT projects to deliver their promised benefits leads us to examine what we argue is the central question that this failure raises: how to involve healthcare professionals effectively in IT systems design and development processes. 16.3 Usability 16.3.1 Usability principles At its most basic level, usability is concerned with ensuring that IT system design is compatible with users’ physical, perceptual and cognitive capacities.9 Beyond these basic usability goals lies an increasingly complex and interrelated set of concerns and HF orientations such that the pursuit of acceptable levels of usability tends to resist a simple guideline-based approach. In reality, the usability of an IT system is a complex, multidimensional problem that is not well-bounded. Usability cannot be defined simply by human performance parameters, but must be set in a context in which issues such as user preferences, working practices and environment are taken into consideration. A design response to this broader agenda is to give users some scope to customise systems so that that they can plan and select actions according to their individual preferences or requirements. An example would be to allow users to choose the level of drug safety alerts in ePrescribing. This will enable users to familiarise themselves with and cope with so-called “spurious alerts” and associated “alert fatigue”, problems that had not been anticipated when these systems were first introduced (see Chapter 10).14 There may, however, be a tension between user choice and safety. In the case of the ePrescribing example, high customisability may lead to potentially serious prescribing mistakes and omissions (discussed below in evaluating performance and safety) and it is for this reason that some software developers, based on the legal advice they have received, have restricted the degree of customisation they allow. It will be clear from the above that design decisions that may influence usability should therefore be subjected to a thorough evaluation process. Usability evaluation attempts not only to identify problems, but also their causes so that they can be eliminated. A number of different usability 433 evaluation techniques are outlined in Table 16.1 below. These techniques can be used both ‘formatively’ (i.e. to identify usability requirements) and ‘summatively’ (i.e. to verify that usability goals have been met): Table 16.1 Comparison of usability evaluation techniques Method Advantages Usable early in the Analytical design process Few resources Heuristic Walkthrough required Strongly diagnostic Overview of whole High potential return Disadvantages Narrow focus Broad assumptions of users’ cognitive behaviour Problems getting experts Cannot capture real behaviour Rich data yields in- Observational Ethnography Data collection and depth understanding analysis can be Capable of detecting time-consuming subtle issues Natural, high validity of results Survey Can be used on Low response rates large groups Possible interviewer Addresses opinions Questionnaire Interviews and understanding bias Interviews time consuming Laboratory Powerful studies Quantitative data High reproducibility High resource demands Artificial, questionable validity of results Based on Dix et al. (2003) 43 Permission to reproduce has been applied for 434 16.3.2 Evaluating usability and safety Improving patient safety, as in the case of England’s NHS Care Records Service (NHS CRS), is often given as a key objective of eHealth applications (see Chapters 3 and 18). As Reason has noted, however, ‘IT does not eliminate error, it relocates it and can also change its form.’15 Clearly, such unintended and hence unanticipated consequences of the introduction of IT must be taken seriously, especially if eHealth applications become more closely integrated with work practices, workflows and organisational goals. Where the quality and safety of care may be adversely affected, then usability evaluation methods must be capable of assessing the impact of a new IT system in a clinically meaningful way. This, as, amongst others, Heathfield and Wyatt,16 Heathfield and Buchan,17 Heathfield et al.,18 Kaplan19 and Sharit20 have argued, remains a challenging area for eHealth. Healthcare has developed the clinical trial to measure the efficacy and safety of new medicines and it is a methodology that also has its place in the evaluation of eHealth applications (see Chapter 20). However, CDSSs provide a good example of how relying on clinical trials alone can fail to identify how eHealth applications are actually used in practice and the consequences of that use (see Chapters 8 and 10). For example, in a review of clinician responses to drug safety alerts in ePrescribing, van der Sijs et al. found that with the exception of serious alerts for overdose, safety alerts were over-ridden in 49-96% of cases. The reasons for over-riding alerts included:21 • Alert fatigue: this was the most important reason for overriding, caused by too many false positive alerts affecting clinicians’ judgement. • Disagreements: professional disagreement with the recommendations being made: clinicians’ faith in their own knowledge or other information sources obtained, incorrect information, patients’ resistance to drug change. • Poor presentation: alerts were too long and difficult to interpret and the clinical consequences of over-riding them were not clear. 435 • Lack of time: insufficient time to pay adequate attention to the messages being generated and unnecessary workflow interruptions. • Knowledge gaps: lack of understanding about importance of the alert. Similarly, a review by Eslami et al. on the appropriateness of alerts, assessed the impact of ePrescribing on the produced, accepted and ignored, alerts from two points of view: system weakness and user response.22 The same review reported that most of the alerts (55-91%) were ignored by the clinicians, with perceived “clinical irrelevance” being the main reason for over-riding alerts. The requirement for measurement and for repeatability favours an approach to usability evaluation that is able to control and isolate assumed key factors. But the issue is whether the results have ecological validity–that is, do they reflect the outcomes that will be found in real situations of use? This question becomes especially important where the introduction of a new eHealth application is likely (either by design or as an unanticipated adverse effect) to lead to changes in working practices. Alberdi et al. have sought to address this issue by using a multi-disciplinary approach combining quantitative data from controlled clinical trials with qualitative data from ethnographic (i.e. observational) studies of clinical decision-making.23 Applying this approach to an evaluation of a decision support tool for mammographic screening, Alberdi et al.’s findings suggest that the ethnographic studies of how radiologists actually used the CDSS in practice enabled a better understanding of the sometimes subtle interactions between human decision-making and computer-generated evidence. 16.4: User involvement in eHealth projects The challenges in setting usability goals and evaluating whether they have been met raises fundamental questions about the nature and scope for the meaningful and effective involvement of healthcare professionals in the design of eHealth applications, and throughout the development, implementation and adoption process. This is an issue on which HFE has increasingly focused, but experience to date confirms that this is not easily 436 achieved. The practical issue is whether it is sufficient just to get user input at the beginning (i.e. formatively) and at the end (i.e. summatively) of a project. Based on evidence such as that above, we would argue that user involvement needs to be taken considerably further, both in terms of scale and scope. Thanks in part to the efforts of the participatory design community,24 the case for user involvement in IT projects is now accepted as standard software engineering practice.25 In this time, a number of recurring themes have emerged as evidence of the challenges of achieving this effectively in practice while, at the same time, questioning whether current practice goes far enough. Firstly, as has frequently been observed by IT professionals, users often “do not know what they want”. Not surprisingly, this problem seems to be especially common where the aim of introducing a new system is to facilitate major changes in work practices. This, of course, is often a goal of eHealth projects, with IT often being used explicitly as a change agent as is the case with the NHS Connecting for Health’s (NHS CFH) programme.26 In such cases, while high level requirements may be easy to identify, the details are likely to prove much more difficult to define and may be based on an oversimplified and abstract model of the work practices involved. Actual work processes are typically much more complex and even quite different from the procedures documented in organisational manuals.27 28 Clarke et al., for example, studied how hospital managers monitor the availability of beds and how this relies on various important, but often taken-for-granted aspects of the workplace whose significance is often overlooked when IT systems are designed.28 29 Similarly, the lack of attention paid by IT professionals to the real world of work–for example, to how clinical information is actually recorded and used–has afflicted EHRs and other healthcare information management and integration projects for many years.30 31 Healthcare professionals, in their turn, often have limited knowledge of the technical possibilities and may find it difficult to re-conceptualise what they do in ways that are compatible with what is technically feasible. 437 We argue that this lack of a common language and understanding between IT professionals and healthcare professionals is a consequence of the “silos” in which they operate. Dealing with this calls for radical changes in the ways in which users and IT professionals work together.32 IT professionals need to make greater efforts to familiarise themselves with users’ work context and practices. One way of doing this that has found increasing favour is to undertake detailed, observational studies of users’ working practices right from the outset of an IT project.33 34 Equally, it is essential that users have sufficient time and opportunities to experiment with the system and learn as the project unfolds which highlights the need for IT projects to support users with prototyping and iteration, and the benefits of staging implementation, by, for example, piloting a small part of the overall system. Even where changes in working practices are not intended, the introduction of a new system can still have unanticipated consequences, sometimes because of poor design decisions, on other occasions because, over time, its users discover more effective ways of using the system.35 Proactively eliciting details of these workarounds from end-users is, we argue, important to inform further iterations of the application. When considering how to improve user involvement in a project, there is no single best technique, nor are techniques mutually exclusive; rather, different techniques may deliver the greatest benefit when used in combination. What there is rather less agreement on, however, is which of these techniques for user involvement are necessary or sufficient for a given project and how this user involvement should be scheduled and managed. For project management, the concern is typically to achieve a balance: ensuring, on the one hand, that user input is adequate for the purposes of establishing and (tracking, possibly changing) user requirements while, on the other, preventing the project being thrown off schedule by a seemingly never ending series of demands for changes. The study by Martin et al. provides a detailed account of project management issues in the context of an EHR project and why ensuring systematic and effective user involvement is challenging.36 Martin et al. note, on the one 438 hand, the concerns of the project team that user participation may become unmanageable. Diverse and possibly conflicting opinions about requirements are almost inevitable in large-scale projects such as the implementation of the NHS CRS, where the range of users is necessarily very broad. On the other hand, Martin et al. point to the difficulties of securing the commitment of users to give their time to the project. User involvement in project work is likely to be discretionary and such work may be taken on voluntarily. Users may experience difficulty in maintaining commitment to a project because the benefits may seem remote and intangible. Even if commitment is strong at the beginning, it may subsequently wane if there is a perceived lack of progress and/or difficulties in using the new system. Unless such issues are addressed, the inevitable result is that what is achievable in terms of user participation “in the wild” is almost always less than what the HFE guidelines call for. Jenkings has addressed a number of these issues in his study of another EHR project.37 One reason why the notion of what constitutes “best-practice” for user involvement in IT projects continues to be elusive is because the problem it is trying to address keeps changing and becoming more complex.38 To put it simply, as IT systems become progressively more deeply embedded within workplaces and organisations, and are increasingly seen (rightly or wrongly) as vehicles for innovating work practices, then uncertainty about what users’ requirements really are grows. This is not just because there may be more and different kinds of users (“stakeholders”) with whom application developers must deal and who may have different–and possibly conflicting–interests (although this is, of course, an important factor). Perhaps more important, however, is that the likelihood that requirements will be difficult to establish a priori, but will change when the system is deployed and users get to use it “for real”. Furthermore, innovation often has unpredictable consequences. For eHealth applications, it is imperative that unforeseen consequences for patient safety are quickly detected and their causes resolved. The challenge is how to evolve applications so that an adequate fit with work practices is maintained. As Suchman argues, ‘…system function and human work processes must be addressed together.’33 439 As responses to these problems, new ideas are beginning to emerge that treat user involvement more seriously. So-called “agile” methods for IT systems design and development follow an incremental planning approach that allows changes to be made according to evolving user requirements and have a commitment to ongoing design throughout the software lifecycle.39 In these ways, agile methods provide an opportunity for IT projects to be undertaken more as a partnership or “co-production” between users and IT professionals in which the boundaries between system design, development and use are broken down.32 38 40 In a similar vein, and drawing on lessons from the NHS CRS, Eason has argued that the pace of project implementation must be geared more strongly to users’ capacity to accommodate and contribute meaningfully, supported by a technical approach that emphasises building upon what is already in place with discrete and phased implementations.41 16.5 Conclusions We have argued that HF considerations are of central importance to the achievement of effective and safe implementation goals in healthcare systems. There can be little doubt, for example, that the lack of attention to HF issues hitherto in eHealth applications such as ePrescribing and the EHR has limited their effectiveness and adoption and, quite possibly, posed a risk to patient safety. In this chapter, while noting the breadth of the HF agenda for IT systems design, we have focused on questions that ensuring the usability of eHealth applications raises for healthcare professionals’ involvement. Patient safety demands that usability requirements and evaluation techniques be driven by clinical agendas and are capable of providing results which are meaningful in practice. It seems inconceivable to us that this can be achieved without close and effective participation of healthcare professionals throughout the IT project lifecycle. The argument that effective involvement of healthcare professionals in design and development is imperative to ensure that eHealth applications are fit for purpose is, we believe, irrefutable. It is also clear that 440 major problems remain to be tackled if the HFE agenda is to be integrated effectively within IT system design and development practice. Much progress in increasing the effectiveness of user involvement in IT projects has already been achieved. However, in the face of increasingly complex and multi-faceted eHealth applications–many of which aim to bring about radical changes in complex work processes and which imply greater integration and interdependencies between services and organisations– healthcare and IT professionals must seek new and more productive ways of working together if the benefits of eHealth are to be realised and patient safety is not to be jeopardised: to put it another way, in eHealth the time has come to stop treating users as “human factors” and, instead, to acknowledge their contribution as “human actors”:42 ‘Part of the problem resides in an implicit view of ordinary people which, if surfaced, would seem to treat people as […] at best, simply sets of elementary processes or “factors” that can be studied in isolation in the laboratory […] Understanding people as “actors” in situations, with a set of skills and shared practices based upon work experience with others, requires us to seek new ways of understanding the relationship between people, technology, work requirements, and organisational constraints in work settings.’42 We argue that this is only achievable if eHealth projects are approached as a co-production of healthcare and IT professionals, such that users are given the opportunity to be more fully involved throughout the design and development phase. Moreover, such enhanced forms of user involvement need to be carried through into the deployment phase so that the fit between work practices and system functionality is maintained as these co-evolve in use. As, following the pattern now evident in other sectors, eHealth applications become steadily more “organisationally embedded”, understanding the organisational context assumes even greater importance for their successful adoption. In the following chapter, we explore the organisation issues 441 surrounding the implementation and adoption of eHealth applications and we then, in Chapter 18, seek to draw together these socio-techno-cultural considerations through a detailed case study exploring the timely and important challenge of introducing the NHS CRS into secondary care settings. References 1. Federal Aviation Administration. Human Factors Engineering and Safety Principles & Practices. FAA System Safety Handbook. 2000.17-1-17-16. 2. Saathoff A. Human factors considerations relevant to CPOE implementations. J Healthc Inf Manag 2005; 19: 71-78. 3. Wilson JR. Fundamentals of ergonomics in theory and practice. Appl Ergon 2000; 31: 557-67. 4. Grudin J. The Computer Reaches Out: The historical continuity of the interface. In: Proceedings of the ACM SIGCHI Conference on Human Factors in Computer Systems (CHI), Seattle, Washington, 1990. 5. Higson G. Medical Device Safety: The Regulation of Medical Devices for Public Health and Safety. Institute of Physics Publishing, 2002. 6. Kaye R, Crowley J. Guidance for Industry and FDA Premarket and Design Control Reviewers. Medical Device Use-Safety: Incorporating Human Factors Engineering into Risk Management. 2000. 7. Sawyer D. Do It By Design: An Introduction to Human Factors in Medical Devices. 1996. US Department of Health and Human Services, Public Health Service, Food and Drug Administration, Center for Devices and Radiological Health. 8. Leveson N. Medical devices: The Therac25. Available from: http://sunnyday.mit.edu/papers/therac.pdf 1995. (last accessed 31/12/10) 9. Sawyer MA, Lim RB, Wong SY, Cirangle PT, Birkmire-Peters D, Sawyer MA et al. Telemonitored laparoscopic cholecystectomy: a pilot study. Stud Health Technol Inform 2000; 70: 302–08. 10. Barach P, Small S. How the NHS can improve safety and learning. BMJ 2000; 320: 1683–84. 11. Button G. A New Perspective on the Dependability of Software Systems. In: Clarke K, Hardstone G, Rouncefield M, Sommerville I, editors. Trust in Technology: a Socio- Technical Perspective. Dordrecht: Springer, 2006: ix–xxv. 12. Heeks R, Mundy D, Salazar A. Why Healthcare Information Systems Succeed or Fail. Institute for development Policy and Management Working Paper Series No 9 University of Manchester, Manchester 1999. 13. Finkelstein A, Dowell J. A comedy of errors: the London Ambulance Service case study. Available from: http://people.cs.uct.ac.za/~gaz/teach/hons/papers/lascase.pdf (last accessed 31/12/10) 14. Corley ST. Electronic prescribing: a review of costs and benefits. Top Health Inf Manage 2003; 24: 29–38. 15. Reason J. Human Factors & Patient Safety: System Issues. Annual Conference of the Care Record Development Board, November 2006. Available from: www.connectingforhealth.nhs.uk/crdb/boardpapers/2006conference/workshop_i mproving_patient_safety.ppt (last accessed 31/12/10) 16. Heathfield H, Wyatt J. Philosophies for the design and development of clinical decision307 support systems. Methods Inf Med 1993; 32: 1–8. 17. Heathfield H, Buchan I. Current evaluations of information technology in health care are often inadequate. BMJ 1996; 313: 1008. 442 18. Heathfield H, Pitty D, Hanka R. Evaluating information technology in health care: barriers and challenges. BMJ 1998; 316: 1959–61. 19. Kaplan B. Evaluating informatics applications—some alternative approaches: theory, social interactionism, and call for methodological pluralism. Int J Med Inform 2001; 64: 39–56. 20. Sharit J. Perspectives on computer aiding in cognitive work domains: Toward predictions of effectiveness and use. Ergonomics 2003; 46: 15. 21. van der Sijs H, Aarts J, Vulto A, Berg M. Overriding of drug safety alerts in computerized physician order entry. J Am Med Inform Assoc 2006; 13: 138–47. 22. Eslami S, Abu-Hanna A, de Keizer NF. Evaluation of outpatient computerized physician medication order entry systems: a systematic review. J Am Med Inform Assoc 2007; 14: 400–06. 23. Alberdi E, Povyakalo A, Strigini L, Ayton P, Hartswood M, Procter R. Use of computer-aided detection (CAD) tools in screening mammography: a multidisciplinary investigation. Brit J Radiol 2005; 78: S31–S40. 24. Greenbaum J, Kyng M (eds.) Design at work: Cooperative Design of Computer Systems. Hillsdale, NJ: Lawrence Erlbaum, 1991. 25. Sommerville I. Software Engineering. 8th ed. Harlow: Addison-Wesley, 2007. 26. NHS Connecting for Health. Available from: http://www.connectingforhealth.nhs.uk/ (last accessed 31/12/10) 27. Clarke K, Hughes J, Rouncefield M, Hemmings T. When a bed is not a bed: calculation and calculability in complex organizational settings. In: Clarke K, Hardstone G, Rouncefield M, Sommerville I, editors. Trust in Technology: a Socio-Technical Perspective. Dordrecht: Springer, 2006: ix. 28. Pappas Y, Seale C. The opening phase of telemedicine consultations: an analysis of interaction. Soc Sci Med 2009; 68: 1229-37. 29. Heath C, Luff P. Documents and professional practice: ‘bad’ organizational reasons for ‘good’ clinical records. In Proceedings of CSCW’96 (Boston MA, November), ACM Press, 1996: 354-62. 30. Hartswood M, Procter R, Rouncefield M, Slack R. Making a case in medical work: implications for the electronic medical record. CSCW 2003; 12: 241–66. 31. Anderson S, Hardstone G, Procter R, Williams R. Supporting the Evolution of Organizational Information Systems. In: Ackerman M, Erickson T, Halverson C, Kellog W, editors. Evolving Information Artefacts. Springer, 2006. 32. Hartswood M, Procter R, Rouncefield M, Slack R, Voss A. Co-realization: Evolving IT Artefacts by Design. In: Ackerman M, Erickson T, Halverson C, Kellog W, editors. Evolving Information Artefacts. Springer, 2007. 33. Suchman L. Representations of work. Commun ACM 1995; 38: 33–34. 34. Hughes J, O’Brien J, Rodden T, Rouncefield M. Ethnography, communication and support for design. In: Luff P, Hindmarsh J, Heath C, editors. Workplace Studies: Recovering Work Practice and Informing System Design. Cambridge: Cambridge University Press, 2000: 187–214. 35. Voss A, Hartswood M, Procter R, Slack R, Rouncefield M, Büscher M. Configuring user-designer relations: Interdisciplinary perspectives. Springer, 2009. 36. Martin D, Mariani J, Rouncefield M. Practicalities of Participation: Stakeholder involvement in an electronic patient records (EPR) project. In Voss A, Hartswood M, Procter R, Slack R, Rouncefield M, Büscher M (Eds.) Configuring UserDesigner Relations: Interdisciplinary Perspectives, Springer, 2009. 37. Jenkings N. User-Designer Relations in Technology Production: The Development and Evaluation of an ‘Animator’ tool to Facilitate User Involvement in the Development of Electronic Health Records. In Voss A, Hartswood M, Procter R, Slack R, Rouncefield M, Büscher M (Eds.) Configuring User-Designer Relations: Interdisciplinary Perspectives, Springer, 2009. 443 38. Hartswood M, Procter R, Rouchy P, Rouncefield M, Slack R, Voss A. Working IT Out in Medical Practice: IT Systems Design and Development as Co-Realisation. Meth Inform Med 2003; 42: 392-97. 39. Beck K. Extreme Programming Explained: Embracing Change. Addison Wesley, 2000. 40. Catwell L, Sheikh A. Information technology (IT) system users must be allowed to decide on the future direction of major national IT initiatives. But the task of redistributing power equally amongst stakeholders will not be an easy one. Inform Prim Care 2009; 17: 1–4. 41. Eason K. Local sociotechnical system development in the NHS National Programme for Information Technology. J Inform Technol 2007; 22: 257–64. 42. Bannon L. From human factors to human actors: The role of psychology and human computer interaction studies in system design. In J. Greenbaum and M. Kyng (eds.), Design at Work: Cooperative Design of Computer Systems. Hillsdale, NJ: Lawrence Erlbaum, 1991. 43. Dix A, Finlay J, Aboud G, Beale R. Human-Computer Interaction. Prentice Hall. UK, 2003. 444 Chapter 17 Importance of organisational issues in the implementation and adoption of eHealth innovations Summary • The study of organisational issues as they pertain to eHealth innovations is a multi-disciplinary field utilising bodies of knowledge from organisational psychology, management and in particular change management and human factors with clinical and information technology expertise. • There is a general consensus that organisational issues are at the root of many of the problems with the implementation and adoption of technological innovation in healthcare, particularly those innovations that are likely to have a major discernible impact on care processes. • While there is at present no overarching conceptual or theoretical framework in relation to the implementation and adoption of eHealth innovations, research consistently emphasises the importance of technical, human and organisational factors, and the inter-relationships between these. • Early and ongoing user nvolvement, relative advantage of the technology and early demonstrable benefits, a close fit with organisational priorities and processes, training and support, and effective leadership and change management seem to be particularly important factors for success. • There is a need for research that moves beyond the existing largely descriptive work to the development and testing of interventions seeking to promote implementation and adoption. • This research is particularly important to undertake in contexts in which organisational factors are likely to have a major impact on the success or failure of the initiative–these include, for example, implementations of technologies that are: likely to have a major clinical impact on care delivery; “pushed” into the organisation rather than “pulled” by endusers and their respective bodies; “commercially procured” with little 445 opportunity for customisation when compared to “home-gown” products with a greater sense of local ownership and buy-in; unreceptive environments; and little opportunity for local “working out” of how these systems can best be used in practice. 17.1 Introduction There is now widespread acknowledgement that healthcare has, when compared to other industries, been slow to adopt information technology (IT). This in part results from the ethos of many healthcare professionals, who tend to emphasise the importance of the more human dimensions of care,1 but also the long history of failures associated with attempts at implementing IT into health settings.2-4 This latter failing often relates to a failure to adequately consider the organisational dimensions of technology, which can, depending on the nature of the innovation and the way in which it is introduced into a healthcare system, be substantial.5 The previous chapter focused on the need to take account of human factors’ onsiderations to ensure that the IT tools that are developed actually meet the needs of clinicians and patients, and furthermore that those that are developed are usable. Building on these deliberations, we shift our focus in this chapter to reviewing the wider social and organisational considerations associated with IT deployment. The case study in the following chapter (Chapter 19) brings together these important–but all too often neglected– human and organisational dimensions in the context of a detailed reflection on the implementation and adoption of the National Health Service Care Records Service (NHS CRS). We begin by discussing what we mean by organisational issues, what bodies of knowledge contribute to their study, and their scope for use. We then discuss the theoretical basis for addressing organisational issues when implementing eHealth innovations, summarise the evidence-base and reflect on its limitations. This is followed by discussion of the themes we have identified as important in implementation to facilitate effective adoption and 446 integration, and the potential for assessing organisational readiness for technological innovation in healthcare. Finally, we discuss the implications for practice, policy and research. 17.2 Definitions, description and scope for use 17.2.1 Definitions There are a number of related terms that are often used as synonyms or in a less than clear way, which can make navigating and interpreting this body of evidence somewhat confusing. adoption, deployment, These include inter-related terms such as diffusion, implementation, infusion, integration, normalisation and routinisation. The glossary provides detailed definitions of these terms, but in essence they all relate to the processes by which innovation is introduced and is then incorporated into care process such that its use by professionals and/or patients becomes part of everyday practice. Rather than these being straightforward linear processes, there is increasing appreciation that the process of embedding technology into health systems is complex and dynamic in nature involving a good deal of iteration and “working out” as the technological, social and organisational dimensions gradually accommodate each other (or not) over time. 17.2.2 Description The study of organisational issues in relation to eHealth is not a clearly defined field of study, but rather a problem-based approach centring on the interaction between organisations or, more accurately, the people working within these organisations and technology. The following bodies of knowledge are potentially useful in contributing to the understanding of organisational issues in the context of IT implementation and usage: • Human factors: An all-embracing term that covers: the science of understanding the properties of human capability (human factors science), the application of this understanding to the design and development of innovations (human factors engineering), and the art of ensuring successful application of human factors engineering to IT (see Chapter 16). 447 • Organisational/occupational/social psychology: A subset of psychology that is concerned with the application of psychological theories, research methods, and intervention strategies to workplace issues, relevant topics include: personnel psychology (e.g. behaviour and attitudes, changes in what jobs entail/working patterns and effects on the individual); motivation and leadership; employee selection; training and development; organisation development and guided change; organisational behaviour; and work and family issues.6-16 • Management and, in particular, organisational change management: A structured approach to change in individuals, teams, organisations and societies that enables the transition from a current state to a desired future state. Often focuses on increasing organisational effectiveness and on identifying barriers and facilitators to reaching the desired future state.17 • Information systems: An academic discipline that is concerned with the uses of information and information technology (IT) in organisations and, more generally, society. This area has emerged from Systems Theory, which assumes that the world consists of complex systems, which have complex inter-relationships with each other and the world at large. The defining feature here is that a system is viewed as being more than the sum of its parts.18-22 Additionally, technological innovation in healthcare requires expertise in technical IT considerations and clinical practice. These bodies of knowledge contribute to the design, development and deployment of eHealth innovations. 17.2.3 The case for attending to organisational issues An appropriate appreciation of the importance of human factors and organisational considerations should help in ensuring that technological innovations that are designed are not only useful and usable, but that they also support the organisations or systems within which patients and/or professionals operate. These innovations should therefore support care provision as well as organisational functioning. Recognition of the fact that change can however prove disruptive–particularly in the short- to medium448 term–means that steps should be taken to minimise potential adverse effects, whilst at the same time, maximising the chances of successful integration with workflow processes and organisational requirements. This is particularly relevant in the context of eHealth applications that are likely to have an impact on day-to-day clinical considerations such as electronic health record (EHR) systems (see Chapter 6), rather than those which have more of an infrastructure function such as the National Network for the NHS (N3) (see Chapter 3). Technological innovations are often employed to enable organisational change in healthcare–as was the explicit aim of the National Programme for Information Technology (NPfIT) for example–in which case attending to organisational issues is also important so as to maximise the chances of organisations successfully achieving the desired outcomes. This is particularly important in the context of NPfIT, where organisations have to fulfil national demands (e.g. achievement of targets) whilst simultaneously having to balance these with organisational demands (e.g. resources), social demands (e.g. user requirements) and technical demands (e.g. interoperability and performance). Managing change effectively reduces the potential for failure, but conversely a failure to give due regards to social and cultural dimensions often proves to be a recipe for disaster. 17.3 Theoretical considerations There is potentially a wide range of theoretical considerations that can be drawn on when conceptualising the interaction between technology, humans and the organisations in which they function. Reviewing these in detail is beyond the scope of this chapter, but here we introduce a few of the more prominent and potentially most useful theories from the literature, which will, where appropriate, be referred to when interpreting the empirical evidence: • Social Shaping of Technology (SST): Approaches studying the SST highlight the importance of wider macro-environmental factors in influencing technology and its implementation and adoption into organisations.23 This has emerged as a response to studies focusing on the social consequences of technology implementation, and in doing so increasingly shifts the focus to viewing technology itself being as shaped by social processes.23 These social processes are assumed 449 to consist of choices made by key stakeholders throughout the stages of design, implementation and adoption.23 Different choices arising from particular interests (which may be transitory or more stable) can have different consequences for different stakeholders– examining these is a central task in SST.23 • Diffusion of innovations/Diffusion of Innovations in Health Service Organisations: These are approaches that focus on how innovations spread in organisations over time and how this spread can be facilitated.24 25 Key components of these models include: o A focus on end-user perceptions and innovation attributes o Strategies on targeting/communicating with and engaging endusers o Attention to both micro and macro contexts. • Socio-technical and sociotechnical changing: These approaches conceptualise change as a non-linear, unpredictable and context dependent process. They assume that both social and technical dimensions shape each other over time in a complex and itself evolving environment. Socio-technical systems have been defined as “any interlinked, systems based mixture of people, technology and their environment engaged in goal directed behaviour which, ideally, leads to the emergence of productivity and well-being”.26 Socio-technical approaches have emerged out of many examples of failed technology implementations and the recognition that social factors are often neglected when introducing technology into organisations.26-28 The aim is to create a receptive organisation that pays attention to both technical requirements as well as social and psychological needs of workers in the changing organisational environment that results from technology introduction.26 29 • Technology Acceptance amongst Physicians (TAM): This assumes that adoption/usage of a system is determined by the attitude towards use and perceived usefulness ("the degree to which a person believes that using a particular system would enhance his or her job performance") and perceived ease of use ("the degree to which a 450 person believes that using a particular system would be free from effort") of the application.13 30 • Normalisation Process Model: This is a theoretical framework describing how complex interventions in healthcare are routinely incorporated into the day-to-day work of healthcare workers (or “normalised”).31 The model highlights the importance of social processes, and the organisational context in shaping outcomes. Several factors are assumed to influence the normalisation of an innovation: o The effect of the intervention on interactions between people and practices o How interventions relate to existing knowledge and relationships o The effect of the intervention on division of labour o The way the intervention relates to the organisational context. • Sensemaking: It is assumed that here individuals in organisations discover meanings of the status quo (frequently as a result of some kind of change), often by transforming situations into words (expressed in language or texts) and then displaying a resulting action as a consequence of their interpretations.32 The underlying assumption is that organisations are not existing entities as such, but are “talked into action” or produced by sensemaking activities (and also the other way around); the very way in which they are talked about defines their existence. • The notion of “fit”: These models emphasise that one not only needs to consider human, technological and work process factors in isolation, but also the extent to which these align with each other.33 The better the fit, the more likely the implementation is assumed to be successful and the higher levels of adoption amongst users are likely to be. Fit can be facilitated through management’s actions. For example, the fit between users and technology can be improved by providing additional training in how to use the software, whilst the fit between task and technology can be improved by increasing the availability of computers in the work environment. Several authors have used the notion of fit to 451 explain their findings, arguing that models relying on factors of success fail to pay attention to dynamics between individual dimensions and the way they affect each other.34-36 For example, Aarts and colleagues explain an unsuccessful implementation of a computerised physician order entry (CPOE) system in a Dutch hospital as being due to a lack of fit between technological and environmental factors.34 They argue that a perfectly good system is worth nothing if it is not used. Our brief list is by no means exhaustive, but our intention is to illustrate the range of different theoretical considerations in relation to the introduction of IT in the healthcare context. Overall, the main tenants of these various theoretical considerations seem to be: (1) a focus on relatively linear stages and integration of technology over time, with some models focusing on exploring one particular aspect of the lifecycle in detail; (2) a focus on individual adopters; (3) a focus on complexity and unpredictability characterising the change process; and (4) a mixture between the above with models trying to be as inclusive as possible. 17.4 Evidence-base and its limitations To date, much of the available primary literature concerning organisational issues in relation to eHealth innovations is anecdotal and retrospective in nature stemming from single organisation experiences of implementing a specific eHealth application. These tend to be descriptive accounts, without much in the way of attention to relevant theoretical considerations, which makes drawing generalisable lessons from such reports difficult.37 Another important feature of this literature is that there are as yet, for obvious reasons, very few accounts of the experiences of implementing major programmebased IT innovations into healthcare systems of the scale and complexity of those that are now being pursued in England and elsewhere. We begin by highlighting key findings from systematic reviews (SRs) and then turn to other key reports in the literature. 452 We found 13 SRs focusing on the issues of implementation and adoption of IT in healthcare (see Appendices 4 and 5).38 38-49 We in addition identified three SRs, which focused more generally on related questions of innovation in healthcare settings.24 50 51 Also of relevance were a number of important additional reports and reviews based on a combination of experience and expert opinion.34 36 44 52-78 Overall, the evidence from the SRs draws attention to the importance of a number of inter-related technical, human and organisational factors that can help describe and explain potential contributing factors to success and failure (or the perceptions of these)54 in attempts at implementing technology into healthcare systems. We summarise below some of the recurring findings in relation to these themes. 17.4.1 Technical features It is now widely accepted that the majority of end-users are not technically averse per se, but they are averse to technology that is seen as inadequate or worse still interferes with their values, roles and aspirations.55 66 A key feature of the technology therefore, as demonstrated by the review of 101 studies of adoption by Gagnon et al,39 is that it needs to be useful and offer relative advantages over existing ways of doing things. Particularly important in this respect is that IT should preferably speed-up tasks, but should at-least be as quick as the existing way of doing things. Such is the importance of this that Bates et al, in their widely cited ‘Ten commandments’ review, refer to this as the first commandment.52 In their extension to the TAM model, Yarbrough and Smith highlight the importance of the related perceived ease of use, which can then shape attitudes and influence behavioural intentions.46 Other features of technology that have found to be important determinants of success includes the extent to which it fits and is interoperable with existing technology in the organisation, costs, complexity and the extent to which it has already been trialled and early problems have had an opportunity to be 453 ironed out.39 53 Given the constantly changing nature, leadership and priorities of complex systems such as the NHS, it is also important that the technology is able to evolve or to be adapted to support change in this way.34 56 62 72 17.4.2 Human factors A number of human factors in relation to technology have been reviewed in the previous chapter. Building on these, the themes that have been highlighted as increasing the chances of successful implementation include the IT literacy and general competencies of staff,57 personal attitudes towards IT and also those of colleagues and patients,58 financial considerations,38 and the extent to which the technology supports inter-professional roles and working.42 Technologies which inadvertently undermine perceived social standing or professional autonomy, for example, are less likely to be accepted than those which support such functioning.59 Involvement of individuals at the conception and early design stages, together with the opportunity for field testing of early prototypes and open communication channels, should help to identify key human dimensions that can help in implementing systems that are likely to be valued and used by professionals and patients.60 61 17.4.3 Organisational factors Larger, more complex health systems have, as predicted by Diffusion of Innovations theory, proven particularly receptive to the introduction of innovation.25 53 This is in part because of the human, organisational and financial capital and “slack” that such organisations have when compared to smaller organisations. These institutions tend however also to have more complex management structures with greater degrees of hierarchy and as such available evidence suggests that it is very important that senior leadership and lead professionals or “champions” are signed-up to the vision and have ownership of the implementation plans.34 36 44 56 60 62-70 73 74 These champions often need to play roles as “boundary spanners”, bridging the gulfs that often exist between and within IT staff, management and clinicians.24 They can also be very helpful in leading the redesign of workflows, providing training and support to new users, highlighting issues, and trouble-shooting, all of which are core considerations in the context of major IT deployments.39 454 The initial implementation is often disruptive as parallel systems are still in operation and staff attempt to make sense of the new workflows.72 Making time available to clinicians and administrative staff by, for example, proactively reducing workload during this time period and/or introducing the technology when there is not other major upheavals in the system has also been highlighted as important.71 Strong leadership and management are also necessary to ensure that a realistic assessment of the likely benefits and trade-offs is conveyed, the anticipated timeframe to realise these, the avoidance of “scope creep”, that interoperability considerations are addressed and that an appropriate implementation approach is adopted–for example, a slow and incremental “soft landing” or a one-off “big bang”.44 68 69 73-76 Leadership also needs to plan for contingencies, including those that are extreme, such as the technology failing, and ensuring that such planning pervades the system.79 Whilst much of the literature focuses on the importance of the “inner context” of the organisation, additional wider contextual factors can also be important, particularly, as discussed in the next chapter, in the context of governmentinitiated programmes such as the NPfIT.24 17.4.4 Ensuring “fit” between these technical, human and organisational dimensions As will be obvious from the discussions above, these three dimensions are closely related. The key it seems is to ensure an adequate fit between these considerations. This point is exemplified in the SR by Yusof et al. who, after reviewing 55 studies using the Human, Organisation and Technology-fit (HOT-fit) framework, concluded that “all three technology, human, and organisational factors are equally important, in addition to the fit between them.” 36 47 This alignment appears much easier to achieve in the case of small-scale, organic, incrementally developed, “home grown” systems, in contrast to the larger, more ambitious, eHealth projects that are now being 455 parachuted into complex environments according to highly pressurised political timelines (see Chapters 3 and 18).80 There is furthermore a growing realisation that for a new technology to embed itself in an organisation, there needs to be opportunities for the technology to shape the organisation and also for those working in the organisation to shape the technology–this therefore needs to be a reciprocal relationship in which new ways of working are allowed to emerge.77 This is challenging for those working using linear perspectives, but unless such experimentation and re-invention is allowed, and indeed encouraged, then it may be that technology never fulfils its potential.24 31 81 82 Although a time-consuming and expensive process, attention to unanticipated consequences is important and need to be studied, both in relation to those which may ultimately prove to be advantageous and of course also those which may inadvertently increase the risk of harm.69 78 17.4.5 A chronological perspective The SR by Keshavjee et al. is interesting in that in addition to the technology, people/organisational and process considerations highlighted above, it takes a chronological perspective, focusing on the importance of pre-, during and post-implementation considerations; this allows the interplay between these different dimensions at different stages of the implementation journey to be studied.43 Based on a critique of 55 reports of implementation of electronic health record (EHR) systems, they identified how the focus of activity of change management changes as implementation proceeds, beginning with, for example, extensive stakeholder discussions when making key decisions on software procurement to the creation and nurturing of user support groups during the early post-implementation phase.43 83 Extending this system development lifecycle-based view, Gruber et al. found that the availability or lack of end-user support during and in the aftermath of “go live” was a particularly important determinant of success or failure.41 Catwell and Sheikh have also highlighted the need for continuous evaluation cycles during the various stages from system conception to integration.84 456 17.5 Conclusions: Implications for practice, policy and research Despite previous work, it is clear, as Lorenzi et al. note, that organisational issues have not received due attention and they identify reasons why these might be initially discounted.85 Organisational issues are experienced subjectively and in different ways by different actors. As a result, they are difficult to measure objectively, difficult to predict and are time consuming to plan for. As a consequence, technical staff often responds by downplaying the importance of organisational issues or by refusing to take responsibility for them. A further factor is that dealing with organisational issues is often viewed as delaying the more immediate implementation-related work. Nonetheless, organisational issues are coming to the forefront of the eHealth agenda due to a general consensus within the field that technological innovation is not designed, developed nor deployed in a vacuum. The importance of organisational issues has important implications for practice, policy and research. Theories need ideally to describe, explain and predict. Findings from the study of organisational issues have however yet to be prospectively applied with any rigour. An ideal place to start with would be a model specific to eHealth to test empirically (see Chapter 18). With enough empirical testing and refining, development of best practice guidelines for implementation is possible; for example, guidelines for involving users during the design and implementation of large-scale systems. In contrast to the points made by Lorenzi et al. on the unpredictability of innovations making strategies for success too difficult to pin-down, Braude notes that ‘…predictability comes from research. When a sufficient body of research in a field is available, it becomes possible to predict outcomes based on prior experiences. Research into people and organisational issues surrounding implementation of innovations is equally scattered throughout the literature of different disciplines with the amount of research is still relatively small.’86 457 The numerous disciplines or bodies of knowledge which contribute to the study of organisational issues are rich in potential to facilitate implementation and adoption of innovations in an ever increasingly complex health services system.87 With that in mind, research employing expertise in these fields is central to furthering knowledge on organisational adoption and best practices for implementation. This prospective work needs to focus in particular on the contexts in which, based on current models, there is the highest risk of problems. Building on the themes introduced in this and the preceding chapter, we in the following chapter focus on human and organisational considerations in the context of a detailed case study on the introduction of the NHS’ CRS. This chapter will also consider the importance of wider contextual factors in impacting on the issues discussed in above. References 1. Stevenson JE, Nilsson GC, Petersson GI, Johansson PE. NursesGÇÖ experience of using electronic patient records in everyday practice in acute/inpatient ward settings: A literature review. Health Informatics Journal 2010; 16(1):63-72. 2. Massaro TA. Introducing physician order entry at a major academic medical center: 1. Impact on organizational culture and behaviour. Acad Med 1993; 68:20-25. 3. Sicotte C, Denis JL, Lehoux P. The Computer Based Patient Record: A Strategic Issue in Process Innovation. Journal of Medical Systems 1998; 22(6):431-443. 4. Southon G, Sauer C, Dampney K. Lessons from a failed information systems initiative: issues for complex organisations. Int J Med Inform 1999; 55(1):3346. 5. Bali RK, Wickramasinghe N. Achieving successful EPR implementation with the penta-stage model. International Journal of Healthcare Technology and Management 2008; 9(1):97-105. 6. Symon G, Cassell C, Dickson R. Qualitative Methods in Organizational Research and Practice. Psychology Press, 2001. 7. Symon G, Cassell C. Neglected perspectives in work and organizational psychology. Journal of Occupational and Organizational Psychology 2006; 79:307-314. 8. Tharenou P. The relevance of industrial and organisational psychology to contemporary organisations: How far have we come and what needs to be done post-2000? Australian Psychologist 2001; 36(3):200-210. 9. Cassell C, Symon G. Essential Guide to Qualitative Methods in Organizational Research. Sage, 2004. 10. Shavinina L. The International Handbook on Innovation. Elsevier, 2003. 458 11. Silvester J, Anderson N, Patterson F. Organizational culture change: An intergroup attributional analysis. Journal of Occupational and Organizational Psychology 1999; 72(1):1-23. 12. Korpela M, Mursu A, Abimbola H. Two Times Four Integrative Levels of Analysis: A Framework. 2001. Proceedings of the IFIP TC8/WG8.2 Working Conference on Realigning Research and Practice in Information Systems Development: The Social and Organizational Perspective, Kluwer. 13. Taylor S, Todd P. Understanding information technology usage: A test of competing models. Information Systems Research 1995; 6(2):144-176. 14. Davis FD, Bagozzi RP, Warshaw PA. User acceptance of computer technology: a comparison of two theoretical models. Management Science 1989; 35(8):982-1003. 15. Karsten R, Schmidt D. Social Psychology for IT professionals: a proposed model. Proceedings of the Academy of Information and Management Sciences 2004; 8(1):37-41. 16. Bandura A. Self-efficacy: The exercise of control. New York: W.H. Freeman & Company, 1997. 17. Cassell C, Symon G, Buehring A, Johnson P. The role and status of qualitative methods in management research: an empirical account. Management decision 2006; 44:290-303. 18. Moore Trauth E. Qualitative Research in IS: Issues and Trends. 2002. Idea Group Inc (IGI). 19. Kaplan B, Maxwell A. Qualitative Research Methods for Evaluating Computer Information Systems. Evaluating the Organizational Impact of Healthcare Information Systems. Springer New York, 2006: 30-55. 20. Myers MD. Qualitative research in information systems. MIS Quarterly 1997; 21(2):241-242. 21. Dalcher D, Drevin L. Learning from information systems failures by using narrative and ante-narrative methods. ACM International Conference Proceeding Series 2003; 47. 22. Orlikowski W, Gash D. Technological frames: making sense of information technology in organizations. ACM Transactions on Information Systems (TOIS) 1994; 12(2):174. 23. Williams R, Edge D. The social shaping of technology. Research Policy 1996; 25(6):865-899. 24. Greenhalgh T, Robert G, Macfarlane F, Bate P, Kyriakidou O. Diffusion of Innovations in Service Organizations: Systematic Review and Recommendations. Milbank Q 2004; 82(4):581. 25. Rogers E. Diffusion of innovations. New York: Free Press, 1983. 26. Butson R. SocioTechnical Approach - STS. Available from: https://russell.wiki.otago.ac.nz/SocioTechnical_Approach_-_STS (last accessed: 21/11/2008). 2008. 27. Mitchell VL, Nault BR. The emergence of functional knowledge in sociotechnical systems. Haskayne School of Business, University of Calgary, Calgary, Alberta, Canada. Available online at: http://www.haskayne.ucalgary.ca/haskaynefaculty/files/haskaynefaculty/emer gence.pdf, accessed 31 October 2008. 2003. 28. Association for Information Systems. Theories used in IS research: sociotechnical theory. Available from: http://home.aisnet.org/displaycommon.cfm?an=1&subarticlenbr=242 (last accessed: 21/11/2008). 2008. 29. Procter R, Williams R. HCI: Whose Problem Is It Anyway? 385-396. 1992. Proceedings of the IFIP TC2/WG2.7 Working Conference on Engineering for Human-Computer Interaction. 459 30. Davis FD. A Technology Acceptance Model for empirically testing new enduser infromation systems: theory and results. Doctoral Dissertation, Sloane School of Management, Massachusetts Insitute of Technology. 1989. 31. May C. A rational model for assessing and evaluating complex interventions in health care. BMC Health Serv Res 2006; 6:86. 32. Weick K, Sutcliffe K, Obstfeld D. Organizing and the Process of Sensemaking. OrganizationalScience 2005; 16(4):409. 33. Ammenwerth E, Iller C, Mahler C. IT-adoption and the interaction of task, technology and individuals: A fit framework and a case study. BMC Medical Informatics and Decision Making 2006; 6:13. 34. Aarts J, Doorewaard H, Berg M. Understanding Implementation: The Case of a Computerized Physician Order Entry System in a Large Dutch University Medical Center. J Am Med Inform Assoc 2004; 11(3):207-216. 35. Pare G. Implementing clinical information systems: A multiple-case study within a US hospital. Health Services Management Research 2002; 15(2):7192. 36. Yusof MM, Kuljis J, Papazafeiropoulou A, Stergioulas LK. An evaluation framework for Health Information Systems: human, organization and technology-fit factors (HOT-fit). International Journal of Medical Informatics 2008; 77(6):386-398. 37. Ovretveit J, Scott T, Rundall TG, Shortell SM, Brommels M. Improving quality through effective implementation of information technology in healthcare. International Journal for Quality in Health Care 2007;(5):259-266. 38. Boonstra A, Broekhuis M. Barriers to the acceptance of electronic medical records by physicians From systematic review to taxonomy and interventions. BMC Health Services Research 2010; 10:231. 39. Gagnon MP, Desmartis M, Labrecque M, Car J, Pagliari C, Pluye P et al. Systematic Review of Factors Influencing the Adoption of Information and Communication Technologies by Healthcare Professionals. Journal of Medical Systems 2010;1-37. 40. Gagnon P, Pluye P, Desmartis M, Car J, Pagliari C, Labrecque M et al. A systematic review of interventions promoting clinical information retrieval technology (CIRT) adoption by healthcare professionals. International Journal of Medical Informatics 79[10], 669-680. 1-10-2010. 41. Gruber D, Cummings G, LeBlanc L, Smith D. Factors Influencing Outcomes of Clinical Information Systems Implementation: A Systematic Review. Computers Informatics Nursing 2009; 27(3). 42. Gurses AP, Xiao Y. A Systematic Review of the Literature on Multidisciplinary Rounds to Design Information Technology. J Am Med Inform Assoc 2005; 13(3):267-276. 43. Keshavjee K, Bosomworth J, Copen J, Lai J, Kucukyazici B, Lilani R et al. Best practices in EMR implementation: a systematic review. AMIA Annual Symposium Proceedings/AMIA Symposium 2006;982. 44. Ludwick DA, Doucette J. Adopting electronic medical records in primary care: Lessons learned from health information systems implementation experience in seven countries. International Journal of Medical Informatics 2009; 78(1):22-31. 45. Mair FS, May C, Finch T, Murray E, Anderson G, Sullivan F et al. Understanding the implementation and integration of e-health services. J Telemed Telecare 2007; 13(suppl_1):36-37. 46. Yarbrough AK, Smith TB. Technology acceptance among physicians: A new take on TAM. Med Care Res Rev 2007; 64(6):650-672. 47. Yusof MM, Stergioulas L, Zugic J. Health information systems adoption: findings from a systematic review. Stud Health Technol Inform 2007; 129(Pt 1):262-266. 460 48. Alexander G, Staggers N. A Systematic Review of the Designs of Clinical Technology: Findings and Recommendations for Future Research. Advances in Nursing Science 2009; 32(3). 49. Gruber D, Cummings G, LeBlanc L, Smith D. Factors Influencing Outcomes of Clinical Information Systems Implementation: A Systematic Review. Computers Informatics Nursing 2009; 27(3). 50. Robert G, Greenhalgh T, Macfarlane F, Peacock R. Adopting and assimilating new non-pharmaceutical technologies into health care: a systematic review. J Health Serv Res Policy 2010; 15(4):243-250. 51. Länsisalmi H, Kivimäki M, Aalto P, Ruoranen R. Innovation in Healthcare: A Systematic Review of Recent Research. Nursing Science Quarterly 2006; 19(1):66-72. 52. Bates DW, Kuperman GJ, Wang S, Gandhi T, Kittler A, Volk L et al. Ten commandments for effective clinical decision support: making the practice of evidence-based medicine a reality. J Am Med Inform Assoc 2011; 10(6):523530. 53. England I, Stewart D, Walker S. Information technology adoption in health care: when organisations and technology collide. Aust Health Rev 2000; 23(3):176-185. 54. Lorenzi NM, Riley RT. Organizational issues=change. International Journal of Medical Informatics 69[2], 197-203. 1-3-2003. 55. Tonnesen AS, LeMaistre A, Tucker D. Electronic medical record implementation barriers encountered during implementation. Proceedings AMIA Annual Symposium 1999;624-626. 56. Doebbeling BN, Chou AF, Tierney WM. Priorities and strategies for the implementation of integrated informatics and communications technology to improve evidence-based practice. Journal of General Internal Medicine 2006; 21(SUPPL. 2):S50-S57. 57. Halamka J, Aranow M, Ascenzo C, Bates DW, Berry K, Debor G et al. EPrescribing Collaboration in Massachusetts: Early Experiences from Regional Prescribing Projects. J Am Med Inform Assoc 2006; 13(3):239-244. 58. Garside P. Organisational context for quality: lessons from the fields of organisational development and change management. Qual Health Care 1998; 7 Suppl:S8-15. 59. McDonald R, Waring J, Harrison S. At the Cutting Edge? Modernization and Nostalgia in a Hospital Operating Theatre Department. Sociology 2006; 40(6):1097-1115. 60. Dagroso D, Williams PD, Chesney JD, Lee MM, Theoharis E, Enberg RN. Implementation of an obstetrics EMR module: overcoming user dissatisfaction. Journal of Healthcare Information Management 2007; 21(1):87-94. 61. Puffer MJ, Ferguson JA, Wright BC, Osborn J, Anshus AL, Cahill BP et al. Partnering with clinical providers to enhance the efficiency of an EMR. Journal of Healthcare Information Management 2007; 21(1):24-32. 62. Ash J, Berg M. Report of conference Track 4: socio-technical issues of HIS. International Journal of Medical Informatics 2003; 69(2-3):305-306. 63. Austin CJ, Hornberger KD, Shmerling JE. Managing information resources: a study of ten healthcare organizations. Journal of Healthcare Management 2000; 45(4):229-238. 64. Bossen C. Test the artefact-Develop the organization. The implementation of an electronic medication plan. International Journal of Medical Informatics 2007; 76(1):13-21. 65. Fenton SH, Giannangelo K, Stanfill M. Essential people skills for EHR implementation success. Journal of Ahima 2006; 77(6):60. 461 66. Greenhalgh T, Stramer K, Bratan T, Byrne E, Mohammad Y, Russell J. Introduction of shared electronic records: multi-site case study using diffusion of innovation theory. BMJ 2008; 337(oct23_1):a1786. 67. Lorenzi NM, Smith JB, Conner SR, Campion TR. The Success Factor Profile for clinical computer innovation. Studies in Health Technology & Informatics 2004; 107:1077-1080. 68. Pare G, Sicotte C, Jaana M, Girouard D. Prioritizing the risk factors influencing the success of clinical information system projects. A Delphi study in Canada. Methods of Information in Medicine 2008; 47(3):251-259. 69. Sicotte C, Pare G, Moreault M-P, Paccioni A. A Risk Assessment of Two Interorganizational Clinical Information Systems. J Am Med Inform Assoc 2006; 13(5):557-566. 70. Walker AR. Case study: leading change across two sites: introduction of a new documentation system. Nursing leadership 2006;(4):34-40. 71. Sirkin HL, Keenan P, Jackson A. The hard side of change management. Harv Bus Rev 2005; 83(10):108-18, 158. 72. Rose J, Jones M, Truex D. Scandinavian Journal of Information Systems. Scandinavian Journal of Information Systems 2005; 17(1):133-152. 73. Duggan C. Implementation evaluation. HIM professionals share their experiences bringing health IT online. Journal of Ahima 2006; 77(6):52-55. 74. Adler KG. How to successfully navigate your EHR implementation. Family Practice Management 2007; 14(2):33-39. 75. Delany R. Smiling all the way. Healthcare Review Online 2004; 8(2). 76. Van Ginneken AM. The computerized patient record: balancing effort and benefit. [Review] [104 refs]. International Journal of Medical Informatics 2002; 65(2):97-119. 77. Czarniawska B. A Golden Braid: Allport, Goffman, Weick. Organization Studies 2006; 27(11):1661-1674. 78. Harrison MI, Koppel R, Bar-Lev S. Unintended Consequences of Information Technologies in Health Care--An Interactive Sociotechnical Analysis. J Am Med Inform Assoc 2009; 14(5):542-549. 79. Johnson C. Case Studies in the Failure of Healthcare Information Systems. Available from: http://www.dcs.gla.ac.uk/~johnson/papers/AHRQ/case_study.pdf (last accessed 14/12/09). 2009. 80. Kreps D, Richardson H. IS Success and Failure—The Problem of Scale. Political Quarterly 2007; 78(3):439-446. 81. Cornford T, Doukidis G, Forster D. Experience Outcome with a Structure, Process and Framework for Evaluating an Information System. Omega, Int J Mgmt Sci 1994; 22(5):491-504. 82. Eason K. Local sociotechnical system development in the NHS National Programme for Information Technology. Journal of Information Technology 2007; 22(3):257-264. 83. Ash JS, Gorman PN, Lavelle M, Payne TH, Massaro TA, Frantz GL et al. A cross-site qualitative study of physician order entry. J Am Med Inform Assoc 2003; 10(2):188-200. 84. Catwell L, Sheikh A. Evaluating eHealth interventions: the need for continuous systemic evaluation. PLoS Med 2009; 6(8):e1000126. 85. Lorenzi NM, Riley RT. Organizational issues = change. Int J Med Inform 2003; 69(2-3):197-203. 86. Braude RM. People and organizational issues in health informatics. J Am Med Inform Assoc 1997; 4(2):150-151. 87. Lorenzi NM, Riley RT, Blyth AJ, Southon G, Dixon BJ. Antecedents of the people and organizational aspects of medical informatics: review of the literature. J Am Med Inform Assoc 1997; 4(2):79-93. 462 Chapter 18 Case study: Lessons learnt from the literature in relation to the design, implementation and adoption of the NHS Care Record Service in secondary care Summary • Many information technology innovations fail to realise their potential and this unfortunately has also been true with respect to the history of eHealth applications. • Major factors contributing to these failures—some of which are spectacular—include the lack of appreciation and attention being paid to socio-techno-cultural factors during product development and deployment. • There is a burgeoning change management literature, relating to the implementation and adoption and the diffusion of innovations and technology. • No one existing model in the literature seems to be comprehensive enough to cover all aspects that need to be taken into account, when considering the implementation and adoption of electronic health records in the context of the National Programme for Information Technology. • Drawing on lessons from the existing literature, we have therefore developed our own model and discuss this in relation to the introduction of the NHS Care Record Service in secondary care. • Based on this, we make 28 recommendations that the Department of Health may wish to consider when thinking about the future strategy of implementing the NHS Care Record Service into English hospitals. • The main components of our model relate to technical, human/social, organisational and macro-environmental dimensions – all inter-related and situated within specific contexts and more or less prominent at particular points in the implementation and adoption circle. 463 Summary continued… • The main components of our model relate to technical, human/social, organisational and macro-environmental dimensions–all inter-related and situated within specific contexts and more or less prominent at particular points in the implementation and adoption circle. • The Department of Health has already taken on board several key lessons, mainly relating to implementation and adoption, but there is still a need to address these more thoroughly in line with recent developments in implementation strategy, the constantly evolving literature and lessons learned from early implementations of electronic health record systems. • We argue that in this respect the most important components of our model are a technology that is perceived as fit for purpose and used by endusers. Recognition of the importance of these issues may, to some extent, have been lost due to the politically driven nature of the programme and its large scale. • Given that the national implementation of NHS computer systems has somewhat changed in strategic direction, it is important to reconsider the evolving literature in this area and incorporate lessons into ongoing policy deliberations. 18.1 Introduction As discussed in Chapter 3, the National Health Service Care Record Service (NHS CRS) represents the backbone of the National Programme for Information Technology (NPfIT) and as such represents a potentially fundamentally transformative and, conversely, also potentially very disruptive eHealth innovation. Considering its centrality within the Programme, it is of considerable importance that in the development and deployment of this technology—which has the potential to yield great benefits to patients— implementers are cognisant of the range of factors that are likely to have a major impact on the acceptability and likely effectiveness of this innovation. It is particularly important to reconsider these factors in the light of the evolving literature base and the constantly evolving implementation strategy. In this case study, drawing on the existing theoretical and empirical evidence, we 464 consider in detail potentially important socio-techno-cultural issues (see Chapters 16 and 17) that may impact on the successful implementation and adoption of NHS CRS in secondary care. In so doing, we hope to identify possible strategies and approaches that implementers might wish to take into account when considering the future strategic direction of the NHS CRS, whilst at the same time contributing to the theoretical and relatively limited empirical base for understanding IT adoption in healthcare. We begin by briefly reviewing the nature, structure and the evolving implementation and policy context of NHS CRS, before turning our attention to theoretical considerations. We then introduce an inclusive theoretical model of implementation and adoption, based on our ongoing review of the literature. We then use this model to reflect on the current and potential future approach to helping to ensure successful deployment of the NHS CRS, outlining areas which, we believe, based on this model, need further attention. 18.2 Context: The NHS Care Records Service and the National Programme for Information Technology 18.2.1 The National Programme for Information Technology The English NPfIT is a ten year strategy (which began in 2002) to modernise the NHS through the introduction of computer systems. In 2005, the Department of Health established NHS Connecting for Health (NHS CFH) as a designated body charged with overseeing the implementation of the programme. NHS CFH has responsibility for setting standards surrounding the implementation and systems, for providing a range of implementation resources to individual organisations and Strategic Health Authorities (SHAs), and also has overall responsibility for liaising with national software suppliers and accrediting software to be implemented as part of the national Programme. In this capacity, it holds national contracts with two Local Service Providers (LSPs), which implement the national solutions, subcontract software developers and deliver the software into Trusts. 465 The Programme was conceived during a time of economic prosperity and aimed to deliver national software systems through NHS CFH as a centralised implementation agency; the underpinning political vision was to modernise the NHS. This implementation strategy is therefore often described as “top-down”, with limited capability for individual Trusts to choose or develop their own software solutions, having to adhere to a national agenda. The underlying approach was a “one size fits all” model, the hope being that the introduction of these standard solutions would minimise the risk of inter-operability problems that had hitherto plagued the NHS (see Chapter 5). 18.2.2 The NHS Care Record Service The NHS CRS is an electronic health record (EHR) that is currently being introduced as the quintessential headline deliverable of NPfIT. This is a complex innovation that will in due course consist of several inter-related components, these include the following:1 2 • National Spine, containing the basic capabilities of the system • National Network for the NHS (N3), a very large virtual private network allowing electronic data exchanges between the applications across organisations • Personal Demographics Service (PDS), containing demographic patient details • Images in Picture Archiving and Communication Systems (PACS) • Summary Care Record (SCR), which is held on the National Spine and contains a record of essential clinical information • Detailed Care Record (DCR), containing comprehensive clinical information on individual patients seen and managed in secondary care • Secondary Uses Service (SUS) for integration of data from different sources and then making this available for audit and research purposes • Choose and book, an electronic booking service giving patients the possibility to choose both location and time of appointments • An Electronic Prescription Service (EPS), an electronic service that allows sending of prescriptions from primary care to pharmacies. 466 The LSPs delivering the NHS CRS currently include Computer Sciences Corporation (CSC) and British Telecom (BT), who implement software developed by Cerner (Cerner Millennium), iSoft (Lorenzo) and CSE (RiO). The implementation of the NHS CRS in secondary care is planned as a staged approach with software releases being implemented incrementally with increasing capability. However, there are some differences between clusters due to variations in software solutions. Lorenzo, the solution being implemented into the North, Midlands and Eastern (NME) region of England, has been developed “from scratch” as it is implemented and does not exist as yet in its full form; Cerner Millennium is in contrast a “finished” product that has been in routine use in the United States (US) for many years. Although both solutions are being implemented incrementally, there are some differences in the scale of implementations. Lorenzo has so far only been deployed as a so-called “soft-landing” on a small scale with a limited number of users and capabilities. Cerner Millennium, on the other hand, is characterised by “big bang” implementations, with whole organisations moving overnight from paper-based to electronic solutions. The introduction of the EHRs into secondary care represents a dramatic change from the current model of working as English clinicians working in hospital settings are currently typically using paper-based records (with all the associated idiosyncrasies of storing, retrieving and interpreting written records). The introduction of the new paperless system is therefore likely to have a significant effect on organisations, healthcare teams and individual practitioners, as well as patients.3 Although it is possible that the introduction of the new application will, if successfully implemented and adopted, in the longer-term result in costsavings, the initial investment is significant, estimated at anywhere between £6-12 billion.2 Given the scale of the financial spend and the fact that so many other aspects of the Programme depend on the success of this initiative, it is vitally important that the implementation of NHS CRS is successful. 467 However, historically the introduction of new IT applications into healthcare organisations has proved problematic, which is a major concern. For example, Sicotte et al. describe the introduction of EHRs into four US hospitals.4 The cost of this introduction was considerable at $45 million, but the venture failed due to the application being rejected by healthcare staff who refused to use it as they felt the application did not fit in with existing care processes. This, and other examples,5 6, illustrate the importance of considering the human dimensions of implementation technologies into existing organisational structures. 7 8 of innovative These factors do not only relate to practicalities and technical issues, but crucially to the design considerations and socio-cultural dimensions of organisational change.9 18.2.3 The present situation Overall, as discussed in Chapter 3, the national Programme has made significant progress in some respects (e.g. implementations of the Spine, N3, and PACS have been completed), other aspects–and in particular the implementation of NHS CRS–is considerably behind schedule. Although there have been hospital-wide implementations of national EHR software with limited capability, particularly in the London area, these implementations have tended to have major problems. These have included issues with lack of user engagement, questions being asked about the software’s fitness-for-purpose and a general lack of support on the ground for the national strategy. At present, particularly the sharing of records between care settings has not been realised as yet and the more advanced clinical functionalities (such as ePrescribing) have not been implemented as part of the national solutions. These developments have been taking place against a changing financial climate, with a resulting in the tightening of NHS budgets in general and, more specifically, concerns about the escalating costs of the Programme. This led to considerable political debate about the need for radical changes in 468 direction. For example, in December 2009, the then Health Secretary Andy Burnham announced cuts of £600m. The recent change in government means that the future direction of the national strategy is therefore at present still somewhat uncertain.10 11 This includes the role of NHS CFH, whose responsibility has been integrated with the Department of Health’s Informatics Directorate. The new coalition government has, despite announcing budget cuts and an estimated £20 billion in NHS efficiency savings, not published a new IT strategy for the NHS as yet, but announced a general NHS strategy in July 2010.10 This included a plan to establish new regulatory structures to monitor progress and quality, increased control of local GPs coupled with a decreased input from commissioners and the government, and plans to abolish SHAs. Nevertheless, despite uncertainties surrounding the detailed future implementation strategy, the implementation of EHRs in secondary care is likely to continue as significant investment has already been made. It is likely that this new strategy will mean a move away from centralised implementation of national systems towards increased local decisions with respect to both the procurement and implementation strategy. 18.3 Socio-techno-environmental dimensions of organisational change—theoretical considerations As discussed in the previous chapter, several authors have attempted to devise models of technology implementation in healthcare settings including those based on critical success/risk factors (with an ingredients-to-successtype of approach) and temporal stages.7 12-14 Most appropriately, the introduction of technology into healthcare is, however, viewed as a non-linear, unpredictable and context dependent process of change where the technology and organisation shape each other (the so-called “socio-technical” view) during the different stages of design, implementation and adoption.12 1517 Central to this is the notion of “organising”.18 This includes the assumption that a complex system is characterised by a constant process of change, in which related structures re-order themselves continuously.19 In this context, an organisation is conceptualised as an accumulation of events and actions 469 by different actors, taking the focus away from an organisation as a structure towards an organisation as a process. Nevertheless, and despite the messy nature of the process surrounding EHR implementations, a range of factors are repeatedly found to be important. These factors have been conceptualised to include the following:20-22 • Innovation attributes (usability, user-friendliness, flexibility, integration with existing work practices) • A focus on end-user perceptions (attitudes, perceived benefits and concerns) • Strategies on targeting/communicating with and engaging of end-users (training and support, effective communication between different stakeholders, effective change management and leadership, evaluation) • A receptive context and planning (e.g. clear setting of goals/vision and preparing the organisation for change, incentives) • Attention to both macro and micro contexts. In order to reflect the process and complexity of implementation and adoption, the present case study avoids using the available models in a concrete way and incorporates components of existing models to be as inclusive as possible. The following sections will summarise the main lessons from the evolving literature, outlining which factors have been found to influence the successful implementation of IT systems in healthcare in general and apply these to the NHS CRS. In so doing, we will attempt to give an indication of the main recurring technical, social/human, organisational and wider macro/environmental dimensions that are likely to play a role and make recommendations relating to the future implementation strategy of the NHS CRS.14 17 23-28 470 The figure below illustrates a model of dimensions and factors identified throughout our emerging understanding of the literature and by observing the evolving implementation landscape. It has, however, to be kept in mind that the division into headings is subjective and that some factors may fit under two or more categories. Also, the complexity of the process and variations in context mean that the factors identified are not prescriptive for implementation success. TIME - STAGES OF DESIGN, IMPLEMENTATION, ADOPTION, EMBEDDED USE Wider macro-environmental dimension4 Organisational dimension³ Social/human dimension² Future state Local context of implementation Technical dimension¹ ¹Includes usability, system performance, integration and interoperability, stability and reliability, adaptability and flexibility, cost, accessibility and adaptability of hardware ² Includes attitudes and concerns, resistance and workarounds, expectations, benefits/values and motivations, engagement and user input in design, training and support, champions, integration with existing work processes ³ Includes getting the organisation ready for change, planning, leadership and management, realistic expectations, management and user ownership, the implementation team, complexity and management in the NPfIT, teamwork and communication, learning and evaluation 4 Includes other healthcare organizations, industry, policy, professional groups, independent bodies and the wider economic environment Figure 18.1 A refined model emerging from our emerging understanding of the literature Using the above model, we aim to build on existing work through undertaking a detailed and up-to-date case study of the implementation of the NHS CRS into hospital-based care. In so doing, we aim to reflect, using existing complementary frameworks, on the utility of the above model. Our aim throughout is to identify, where possible, additional strategic steps that might 471 usefully be taken to maximise the likelihood that the NHS CRS will successfully be implemented into existing organisational structures and, most importantly, be used by hospital staff. 18.4 Technical dimension (the technology…) Implementation of an EHR involves learning to use a new technology and in some instances replacing an existing one, which may result in negative user attitudes.29 Choosing the right software is important as it needs to fit in with organisational and individual user requirements so that it does not need to be replaced again in the near future.12 30 Factors under the technical dimension relate to the design of the technology itself, which is commonly considered to play an important part in impacting on adoption behaviour of users as technical issues can lead to significant levels of frustration.31 32 Equally, user satisfaction and adoption can also be positively influenced by the design of the technology.33-35 The way humans and technology interact is commonly referred to as ergonomics or human factors. 18.4.1 Usability Usability problems can pose significant barriers to adoption of IT in healthcare, as systems that are difficult to use are often rejected by users (see Chapter 16).12 27 31 36-39 Conversely, if a system is perceived to be userfriendly and easy to learn it is likely to be adopted more readily.17 40 Usability issues are commonly conceptualised to be due to a poor alignment between work processes and software specifications.33 41 This is often exacerbated by the complexity of EHRs, which makes them difficult to navigate and, in particular, to obtain an overview of a patient or particular care activities.29 42 Usability is also often impacted on by the number of interfaces in EHR applications, which can significantly slow down work.15 27 36 43-45 The difficulty for designers therefore lies in balancing the complexity of information needed for day-to-day healthcare activities without increasing the overall number of screens.35 46 47 In doing so, they also need to bring together the 472 large amount of information required by healthcare staff within a limited amount of screen space and to ensure that the interface in any one screen is not too complex (i.e. that the items on the screen are not too close together).14 35 Usability can be improved by making the software design as intuitive possible.48-52 This can, for example, be achieved through an increased use of graphics or “visual hierarchies”, but also through standardised symbols, alerts and reminders.35 53 Hardware usability can be improved by portable technology (e.g. tablet computers).12 Van Ginneken describes what clinicians are looking for in an EHR in relation to usability.45 It is, for example, important that the design of the record provides an overview (which means finding quickly for what one is looking for), has a predictable structure (e.g. through use of headings and consistent buttons, use of tables and graphs) and is clinically relevant (i.e. meeting the need to be able to find clinically relevant data quickly). Issues with usability also often occur if the EHR system does not fit in with existing work practices of users (discussed in more detail below). For example, Jaspers and colleagues conducted an evaluation on the usability of a new and an old EHR system in the Netherlands in secondary care through standardised questionnaires given to clinicians.46 They found that adoption of EHRs is influenced to a large degree by perceived usability. The more in line the system was with existing work processes and the more intuitive it was the more satisfied were the clinicians. In this respect it is also important to consider that different users have different work practices, which has implications for individual usability. Rose and colleagues used a combination of task analysis and focus groups with clinicians and some nurses to inform the usability of an EHR in US primary care.35 They discovered that users found the EHR difficult to use and navigation was viewed as too complicated and not in line with existing workflows. Users also felt that EHR access needed to be quicker. Usability 473 problems resulted in feelings of frustration and in some cases users had developed workarounds that were perceived to have adverse effects on patient safety. The authors recommend therefore that designers need to design the system paying attention to different types and different needs of different user groups (e.g. experienced/inexperienced users and different professional groups). Testing systems with users to identify issues has also been recommended to improve usability and avoid frustration.35 54 The literature suggests that the best way of achieving application usability is through a close collaboration between the designers of application and endusers.3 55 This may take the form of continuous testing of prototypes in different groups of end-users and re-design if necessary.56-58 Hartswood et al. have applied an ethnographic method called “co-realisation” to help achieve user-informed IT design.59 They argue that a facilitator is important in this context in developing a partnership between users and designers of the IT application. NHS CFH has already made efforts of increasing user involvement during deployment, which will be discussed in more detail below. There have thus far however been only limited efforts focused on incorporating user input into application design. Such engagement needs to occur with a range of user groups. Issue 1: NHS CRS software is often not perceived as usable by end-users.60 Recommendation 1: Start with a usable system, paying attention to human factors’ considerations. Alternatively, start with a system that can be “made usable” with input from users and prompt incorporation of suggested changes. A close working relationship between designers and users is essential to achieve this. 18.4.2 System performance Frequently, EHR users need to rely on service providers to tell them how the system should perform and what it can and cannot do in line with technical specifications.38 This is complicated by the immaturity of many existing EHR systems, including the NHS CRS.27 32 39 474 When considering the necessary integration of the system into the complex and fast-moving nature of clinical work, users can be frustrated and unhappy for a variety of performance-related reasons. The most common one is slow speed of the application.14 15 23 27 28 39 53 61 62 In this context, it is important that using the application is not significantly slower than the old system, whether paper-based or electronic.63 This may include system downtime and long login times, which are frequently reported to delay users.29 39 Slowness of the system has been reported by many NHS CRS software users, which is of concern. Whilst logging on with smartcards and pin number to verify identity is a good way of addressing concerns surrounding confidentiality, the resulting length of the logging-in process may compromise valuable time in the case of emergencies. There may here be a potential for using biometric technologies, but this has obvious additional cost implications and privacy considerations. Conversely, systems need to be in place to ensure that after use, healthcare staff will be logged-out after a certain amount of time to minimise risk of inadvertently breaching confidentiality.55 If the system is not perceived as improving performance, it is likely to be rejected. Generally, the new system should therefore help to make work more efficient rather than slowing it down.36 For example, Granlien and colleagues report on the implementation of an electronic medication record in Danish hospitals.64 Three years after implementation it was still only used by the minority of healthcare staff despite increased efforts to facilitate adoption. The authors conclude that the system was reported to be slow, not user-friendly and that there was a perceived lack of training on how to use the system. However, it also has to be kept in mind that an initial (temporary) slow down in work processes is to be expected. This should ideally ease off as users get used to the new system as performance also often depends on the skill level of users.65 For example, Quinzio and colleagues report on the results of a survey about user acceptance of an anaesthesia information management system at a German hospital after the system had been used for a period of five years.66 Although no time savings were reported, the authors suggest that 475 this may have been due to other factors such as users’ lack of typing skills. They suggest typing classes to increase speed of data entry. Issue 2: NHS CRS software is perceived to slow down the work of endusers.60 Recommendation 2: There is a need to communicate that using an electronic system will never be as fast as using paper. In the best case scenario, it will take the same amount of time. Whilst it should be expected that systems will slow down work processes initially, there also needs to be some consideration as to what level of slow performance is acceptable and ways to address unnecessary levels of bureaucracy. 18.4.3 Integration and interoperability A new EHR should integrate relatively easily with existing IT systems (see Chapter 5).23 28 42 49 67 The development of standards for interfacing and integration with existing systems has been found to facilitate implementation across studies.12 26 36 51 52 67-70 Conversely, if there are problems with integration, this can slow down clinical work and result in increased user frustration.14 15 27 37 65 66 71 For example, Miranda and colleagues report that a lack of standards inhibited sharing of information across specialties.54 Also, Sicotte and colleagues, reporting on two longitudinal case studies of EHR system implementations in Canada, found that one case used a data warehouse, which meant that new standards and new interfaces had to be developed to connect systems.14 This led to delays in the implementation timeline and additional costs. It also impacted on user perceptions of the system. However, it has also to be kept in mind that whilst interoperability means that systems have to a certain degree to be standardised, these have to be flexible enough to fit in with the complex individual work practices in healthcare.38 44 45 68 The discussion as to exactly how much standardisation is needed in different contexts is ongoing. 476 Stead and colleagues outline three different types of integration projects with increasing levels of complexity. These range from a single database, to integration of similar sources and systems, and integration of sources and systems that are by nature very different.72 The problem is that individual commercial systems are often not designed to integrate with other systems. Integration and interoperability between systems is particularly important and complex in large-scale IT implementations across different care settings, as in such scenarios the number of interfaces required is higher than in smaller and more contained settings (such as for example GP settings).73 The NHS CRS has been set up as an integrated solution by centrally procuring selected software applications. However, recent changes in the direction of the Programme mean that the market is likely to open up so that hospitals can chose other software suppliers. Therefore, issues surrounding integration and interoperability will become increasingly important. Significant progress has already been made by NHS CFH in developing standards for interfacing of systems (the dm+d standard), which vendors supplying systems have to adhere to. Standards have also already been adopted to authenticate the identity of users and to ensure the secure transfer of information across applications (electronic Government Interoperability Framework) and to outline the technical requirements regarding the specification of the application (OBS—Output Based Specifications).74 75 Issue 3: The market of NHS CRS software is likely to open up with an increasing number of vendors. Recommendation 3: There is a need to build on NHS CFH’s efforts to set standards for interoperability. A central authority is extremely important for achieving this. 477 18.4.4 Stability and reliability Some studies also refer to the importance of system stability. The more stable a system is, the higher the likelihood of successful implementation and adoption.28 32 Conversely, a lack of system reliability (e.g. crashes) can result in negative user attitudes.29 In this context, Bossen describes the implementation of an electronic medication plan as part of a national EHR implementation in three Danish hospitals.28 Data on adoption were collected through observations and interviews with key stakeholders. During the initial test period, several issues emerged. Log-in times were viewed as too long, computers froze after a certain period of inactivity and had to be “rebooted”, and the network was not stable enough. NHS CRS software itself seems to be relatively stable, but existing networks and infrastructures to support the software are often not fit for purpose and different from the environment in which the software is tested. This, in combination with constant system upgrades, tends to give users the impression that the system “crashes”. Issue 4: Existing infrastructures do not support the software adequately. Recommendation 4: Focus on building adequate infrastructures before introducing a system. Testing should be done in an environment that mirrors existing infrastructures. It however needs to be acknowledged that no system can ever be fail-safe. It will therefore be necessary to have systems in place and disseminate a plan of action of what to do in such situations. This will mean devising alternative forms of accessing and storing data in collaboration with application designers and may take the form of reverting to paper processes. However, this requires that staff are trained on what to do in such situations.76 478 Issue 5: No system is failsafe Recommendation 5: There is a need to have back-up systems in place and plan for such situations in advance. 18.4.5 Adaptability and flexibility A common barrier to adoption is that IT systems in healthcare are often not flexible enough to fit in with the nature of clinical responsibilities and local needs.28 41 71 77 Conversely, space for adaptation of newly introduced systems can facilitate adoption and increase user acceptance as the system needs to in with existing work practices of users (discussed in more detail below).15 23 34 35 Some authors have in this context referred to the “design reality gap”, which points to the differences in intended use of a technology by designers, as opposed to actual use (in context by healthcare staff).78 This gap needs to be closed for the new EHR to be successfully integrated. For example, Aarts and colleagues describe an in-depth qualitative study of the implementation of a computerised physician order entry (CPOE) system into a Dutch hospital.15 Here, the purchased US system needed to be (and was) adapted to the Dutch context through translation of terms and specification of different clinical pathways. This is also relevant in relation to the NHS CRS as Cerner Millennium is a US system. Adaptability to the UK context is therefore crucial for successful implementation. However, this need for adaptability and flexibility also needs to be balanced with standardisation and interoperability considerations, especially in relation to national systems such as the NHS CRS.40 79 80 Designers therefore need to find a balance between locally customised and shared parts of the system.81 The potential for re-invention is another important described attribute. Healthcare staff needs to be able to adapt the use of NHS CRS to their particular profession’s requirements. This may be done via design features and requires active efforts of involving staff. Supporting the usefulness of this construct, Ash et al. interviewed health professionals at sites where CPOE 479 was successfully implemented and identified the ability to adapt the application to local needs as a facilitator for adoption.82 Issue 6: There are trade-offs between customisation and standardisation Recommendation 6: There needs to be a common agreement on the shared element of the NHS CRS, whilst there is also a need to allow flexibility for customisation and adaptation. 18.4.6 Cost Several studies indicate that costs of the system and the implementation process can be a significant barrier to adoption.14 27 34 37 38 40 45 47 48 51 54 62 65 77 83 84 For example, a recent systematic literature review to identify barriers to EHR adoption amongst primary care physicians has suggested that financial barriers are frequently cited and include high perceived initial investment cost, high ongoing costs, uncertainty surrounding return of investment, and lack of financial resources.29 Cost as a barrier to implementation has also been identified specifically in relation to the NPfIT, although Trusts have to date in theory received the NHS CRS system “for free” as it is centrally procured.85 Nevertheless, ongoing costs (e.g. for training and maintenance) and a lack of quantifiable, and particularly long-term net, returns from these investments are still a concern in this context.23 34 51 71 84 86 There is also currently no business case for EHR implementation and its cost-effectiveness is based on speculation in the longterm (as outlined above).38 Issue 7: There is a cost involved in implementing the NHS CRS for both local organisations and the country as a whole. Recommendation 7: There is a need to realise that implementing the NHS CRS should not be based primarily on cost savings as these may take a long time to or possibly may never materialise. 480 18.4.7 Accessibility and adaptability of hardware Another crucial factor is providing an adequate workspace for users of the new application,57 as the lack of access to computers has previously been identified as a barrier for use.39 56 87 This may, for example involve healthcare professionals waiting for computer terminals to become available, which again slows down work. This means thinking through the positioning of computers throughout the setting and may involve access testing for simulated emergency situations. Once users reach a terminal, it is important to consider that a variety of different individuals may need to use the same computer terminal. This means that settings at the workstation therefore need to be easily changeable.70 Similarly, a commonly cited barrier to EHR adoption is a lack of hardware to support the new system.29 In the context of the NHS CRS, issues surrounding accessibility and adaptability of hardware are likely to become more important over time as local systems become more developed. It is, however, important to consider these issues when planning for EHR implementations, particularly when considering that the majority of NHS buildings have simply not been built with computers in mind. 18.5 Social/human dimension (those that interact with the technology…) In order for successful implementation and adoption to occur, a strategic targeting of end-users (adopters) of the innovation is crucial. The NHS CRS is likely to be used by a variety of healthcare professionals as well as patients in a range of settings. Groups of adopters may, however, differ in terms of particular needs and or existing skills which may in turn influence individual adoption. A potential problem is that end-users of the NHS CRS have thus far not been mapped out and can therefore not be systematically targeted and motivated to use the application. There is also a lack of research investigating how adoption of eHealth technologies can successfully promoted.88 481 Once prospective end-users have been mapped out it is important to consider in which capacity they will utilise the NHS CRS. This is best informed by users themselves in order to develop a sufficiently rounded understanding.89 A potential problem is, however, that there is still some confusion of how exactly the SCR will be used—this uncertainty within NHS CFH has a knockon effect on professionals and patients—and there is a pressing need to address this.2 There is also still confusion over who exactly will use DCR applications and which organisations will be enabled to share information. The NHS Care Records Service Registration Authority is a designated authority for registering staff to use the NHS CRS and may have a role to play in this context. Most existing studies in the existing literature have focused on doctors and some on nurses. There is clearly a need to investigate factors that are relevant to other users such as administrative staff and allied healthcare professionals as their work processes and resulting needs are likely to vary from those of clinicians. For example, a study by Lium and colleagues found that secretarial staff found adapting to a new EHR system easier than clinical users.42 This may be due to this group of healthcare staff being used to working with computers. There is also another stakeholder group, which although of prime importance, has often been neglected: the patients. It is acknowledged that they too are important, but patient views are not the focus of the current review. Individual differences between healthcare staff also exist in ways of working, patterns of communication, backgrounds and place in the hierarchy.12 24 These factors are likely to result in each user having his/her own perspective of the system in the context of his/her own work and at a particular time.41 Issue 8: There are many users and many different contexts of use of the NHS CRS. 482 Recommendation 8: It is important to map these out and communicate consistent messages as to what exactly the functionality will consist of so that appropriate plans can be made. 18.5.1 Attitudes and concerns It is generally acknowledged that for change initiatives to be successful change needs to have a receptive context that is characterised by positive attitudes of stakeholders.90 In line with this, attitudes towards the EHR amongst users are important in determining adoption decisions. Accordingly, many studies have found that many EHR implementations fail due to a lack of user acceptance and satisfaction.7 16 17 27 35 40 45 47 50 51 53 65 66 70 71 77 91 Although significant efforts have been made to assess patient and user attitudes towards the Programme, some have argued that these were not conducted with representative samples.80 Nevertheless, there now appears to be a widespread unfavourable attitude towards the Programme, especially amongst doctors.92 Clinicians tend to support the underlying aims of shared EHRs, but often not the implementation strategy; of concern is that support for the Programme in this group is decreasing.93 Concerns mostly voiced amongst healthcare staff in relation to EHRs–and the NHS CRS in particular– include those surrounding confidentiality and security. 83-85 94 12 14 29 32 45 51 62 65 67 69 77 Some of these concerns have to an extent been addressed by initiatives such as the NHS Care Record Guarantee, which is setting standards to protect confidentiality.75 It has however been pointed out that if work is disrupted through security measures, this can decrease user motivation to carry out a certain task.95 Role-based access features (that define the level of access to patient information according to profession) can help address these issues, but this may delay implementation and can result in reduced user satisfaction due to the limited amount of information provided by the system to each individual user, and lengthy log-in times.14 28 483 Other concerns amongst healthcare professionals relate to a fear: of increased workload; that breakdown in computers may lead to threats to patient safety, about how the system will influence individual roles and responsibilities (e.g. professional control over patient information); that the system will become outdated relatively quickly; and that the system will negatively impact on doctor-patient relationships (e.g. less conversation, less eye contact).14 17 29 29 31 38 41 45 49 51 65 66 71 83 85 One case study described how concerns surrounding the impact on doctorpatient relationship were diffused by offering users the possibility of portable computers, whilst confidentiality concerns were addressed by explaining the system and how it protects confidentiality.49 Another study found that doctors felt apprehensive towards modernisation including technology in the NHS and tended to feel that their professional identity and deep rooted values were to some degree threatened by modernisation.96 In relation to the NHS CRS, it has therefore been argued that there is a need to address existing concerns and map out how CRS can contribute to safety and quality of care with robust evidence in order to address existing concerns.22 Demographic and individual factors such as level of experience with technology, salary status, personal values and norms also play an important role in determining attitudes to technology.17 77 There also seem to be differences in attitudes between younger and older users with seniors showing lower levels of acceptance.42 One survey indicated that clinicians felt that the real use of computers would come as the old generation of users would retire and the younger generation who are more used to using them would become more prominent.85 Similarly, in a Dutch study, Van Der Meijden and colleagues investigated the attitudes of future users of an EHR for stroke patients before the system was implemented.50 They found more negative attitudes amongst inexperienced computer users and a general expectation that the EHR would improve the perceived negative aspects of paper records. Usability, speed and reliability were high priorities in this context as discussed above. 484 However, despite this, it has to be kept in mind that change is unpredictable and attitudes amongst players/parties can change over time.97 Issue 9: There are widespread negative attitudes towards the Programme, with particular concerns surrounding the issues of confidentiality, increased workload and a negative impact on care provision. Recommendation 9: There is a need to face these concerns “head-on”. Mapping of stakeholders is likely to facilitate this as concerns are often profession-specific. 18.5 2 Resistance and workarounds Negative attitudes and the feeling that change is externally imposed on users can lead to resistance to use a system or result in workarounds. Some have argued that resistance to technology itself is rare; rather, the problem is more often resistance to the management introducing the technology. 95 Especially, high levels of resistance among clinical users can threaten the implementation as this staff groups tends to have a high degree of professional autonomy.14 28 31 98 For example, Doolin describes the implementation of an information system (that monitored the performance of doctors) into a hospital in New Zealand. Here, doctors resisted using the system which ultimately resulted into changing the system itself to encourage clinical use.99 However, Ferneley and Sobreperez argue that resistance to use has traditionally been viewed negatively, but is not necessarily so.100 It may, for example, be a response to an inadequate system and can in certain situations be adaptive (e.g. in compensating for inadequate technology).80 101 If resistance occurs then “workarounds” can emerge as a response on the individual level. Workarounds are a result of the demands/rules that a certain organisation imposes through a system, which cannot be accommodated with everyday practice. 485 Ferneley and Sobreperez outline three types of workarounds that can result from resistance.100 The first type is labelled “hindrance workarounds”. These tend to occur when using a system is perceived as so cumbersome that users either enter partial data, wrong data or enter data retrospectively. This type of workaround may also result in avoidance to use a system. The second type is labelled “harmless workarounds”. Here, users tend to use the system in a different way to that which was intended, but the outputs of the system are not affected (e.g. conducting work in different sequence, sub-dividing tasks). The third type comprises of “essential workarounds”. These are workarounds that are viewed as important by users (e.g. retrospective data entry in an accident and emergency (A&E) care setting). An example relating to workarounds in healthcare IT comes from a study conducted by Vogelsmeier and colleagues evaluating the introduction of an electronic medication administration record. Here, the authors argue that workarounds used by nurses can be divided into two main categories.102 The first category included workarounds as a result of a “work flow block” introduced by the system (e.g. a warning asking the user to reconsider a certain action) and the second category included the failure to integrate organisational processes with the technology (e.g. if the system was too slow). If the block in the system was perceived as unjustified or the system was perceived as too slow, then staff tended to revert to using paper systems. However, the authors conclude that workarounds such as these may compromise the effectiveness and integrity of the electronic system as here staff deal with an emerging problem without trying to address the underlying causes. Similarly, in a recent literature review, Stevenson and colleagues found that nurses tended to view EHR documenting practices as taking time away from patient care and close interaction with patients.39 Therefore, notes were often taken on paper at the bedside and then later transferred onto the EHR system. 486 Some have argued that in relation to the national Programme, technology may be resisted as it is not viewed as central to, and/or disruptive to healthcare professional work.80 98 Nevertheless, it has to be kept in mind that resistance is typically expressed from a certain viewpoint, typically by those who want others to comply with the change.103 Issue 10: Introduction of the NHS CRS may be resisted by users. Recommendation 10: There is a need to seek to understand why this is the case, as it may be due to managerial views of users finding it difficult to cope with change, but it may also be a result of inadequate technology. 18.5.3 Expectations Potential assumed benefits of EHRs include improved patient safety and work processes, time savings and improvement of communication between different settings.40 45 Users do tend to have pre-conceived ideas and expectations about the potential benefits of the technology and how the software and hardware will look.104 These can influence adoption behaviour negatively, especially if expectations of users and the product delivered do not align.27 41 54 105 106 Greenhalgh and colleagues state that this tends to be most often the case with “off the shelf” systems, which often do not meet expectations of users.80 The result is an increased risk of implementation failure.80 101 107 Karsten and Laine have explored expectations through interviews with healthcare staff in a Finnish hospital.41 They have identified user expectations relating to the technology itself, expectations of how it influences work processes, expectations of how it influences the organisation and expectations in relation to usability and training. Negative attitudes and expectations are often also exacerbated by previous negative experiences with technology and the many examples of failed IT implementations in the healthcare sector.14 27 38 40 47 65 487 It is, however, important to keep in mind that different stakeholders tend to have different expectations. In the context of the national Programme, there are numerous groups of potential users and stakeholders of EHRs, making it difficult all of these expectations to be met.80 100 108 This is further complicated by the large scale of the NPfIT and a “strategic optimism” on the part of policy makers.97 This results in aims that cannot be achieved and perhaps too high levels of public expectations, neglecting the complex nature of such programmes. Others have supported this notion by outlining the often very high expectations of managers and implementers, who assume that safety and quality of care can be improved through increased availability of information,109 that clinical information will be entered at the point of care,110 and that clinicians to support implementation of NPfIT.93 Many have viewed the expectations from management as unrealistic and too ambitious, which may contribute to implementation failure.101 107 108 For example, there still seems to be a lack of belief amongst users that EHRs will result in improvements to patient care in line with a lack of quantitative data to support this.29 Issue 11: Expectations of what the NHS CRS will achieve are often not realistic. Recommendation 11: There is a need to find out what expectations stakeholders have and assess to what extent these can realistically be met. “Strategic optimism” is suitable for getting an initiative “off the ground”, but will, if sustained and not matched with reality, lead to disillusioned stakeholders. 18.5.4 Benefits, value and motivations In order to address expectations and make these realistic, it is important to consider and explicitly state the potential value to users and implementing organisations, the potential trade-offs and anticipated effects on work processes. In this context, Klein and Sorra argue that the success of an innovation in an organisation is determined by organisational climate and the 488 degree of fit between the innovation and user values. To strengthen “innovation-value fit” the organisation should involve users in the decision to introduce an innovation and help employees identify the need for an innovation by educating them about it. This should involve making the benefits arising from the innovation explicit.111 It is assumed that if new systems address an identified user need and fit in with existing work practices, this creates a “user pull” and facilitates implementation.110 In doing so, management should not make assumptions that people understand the vision.91 Otherwise, there is a danger that the system will be rejected by users.14 17 23 27 31 32 34 37 42 44 49 52 71 Ownership needs to be developed allowing users to identify individual benefits to them and their patients (not only to the organisation or healthcare in general).104 110 Therefore, one may wish to state separate organisational and individual benefits as well as gains from different aspects of the system.40 42 49 This is because different gains may be motivators for different groups of stakeholders. Mapping potential benefits and anticipated changes to work processes for each group of stakeholders and before implementation can help to anticipate some problems.30 70 For example, Nikula reports on a qualitative study at two Swedish hospitals with EHRs and found that managers viewed the EHR from an organisational perspective (as facilitating organisational processes) whilst clinicians viewed it as facilitating clinical processes.112 Similarly, the focus of clinicians is mainly on usability issues, whilst the focus of management tends to be on business process issues.47 Here, the vision for the future was mainly present in managers but not clinicians and the author argues that it is crucial that additional efforts are spent on communicating the vision more effectively to clinicians. Van Ginneken considers the alignment of relative efforts and benefits in more detail.45 The author argues that when considering the implementation of electronic record systems, efforts of users will only be invested if the resulting benefits are motivation enough to invest. Therefore, clinicians are more likely to adopt if they can see the benefit of using an EHR for themselves. In this context, it is important that the relationship of benefits and efforts is relatively equal for all stakeholders. For example, benefits of the system are most 489 obvious if clinicians are inputting structured data directly, but often clinicians are reluctant to do so as it increases their workload and does not yield visible benefits in the patient encounter itself.29 The benefits of structured data entry are primarily organisational and governmental (e.g. cost saving). Several other studies have supported the notion that if desired benefits cannot be observed and the system is perceived to be of limited value to users, then this can increase user resistance and inhibit adoption.15 53 113 Conversely, other studies also support the notion that an identified individual need for the system can facilitate adoption.30 49 65 In relation to the NHS CRS, there seems to be a lack of tangible benefits for staff on the ground, which is of concern.62 114 Aligning the NHS CRS with user and organisational values is clearly difficult, due to the national and top-down implementation strategy. Although NHS CFH has invested a considerable amount of effort into benefits realisation, these efforts were mainly based on business process issues, lacking a focus on individual users. There has also so far been a lack of attention paid to measuring risks and trade-offs. An identified need for a system can help to increase staff motivation to use it. 115 Motivations for using the new system therefore need to be identified and systematically targeted whilst concerns about the new system and barriers to using it need to be voiced, openly discussed and addressed as early as possible in order to facilitate user ownership. 116 12 17 25 34 42 49 51 54 61 70 80 94 113 114 If requests or feedback can not be incorporated, users need to be informed and reasons why need to be communicated in order to prevent frustration.42 54 This also needs to involve outlining the benefits of using the system for different user groups.12 42 61 For example, Miranda and colleagues describe lessons learned during implementation and expansion of a nursing information system in secondary care.54 Actively involving all different users in the pre-implementation phase in order to help to define what advantages the system would have for different user groups was viewed as crucial in order to create a vision and to help define system requirements so that it would fulfil users’ needs. Similarly, Smaltz and colleagues describe a case study of the implementation planning of a US secondary care core clinical system 490 (including CPOE, results management etc.).117 They argue that the essence of a good vision is that it integrates a plan with different goals defined for different clinical areas and stakeholders. This was achieved through an assessment of the hospital’s current practices before implementation in order to help define the vision (e.g. through work process mapping) and to identify potential areas for improvement. The importance of vision will be discussed in more detail below. Issue 12: There appears to be a lack of tangible benefits for individual users of the NHS CRS. Recommendation 12: There is a need to assess how both small scale and large scale benefits can be aligned. There is also a need for benefits realisation teams to measure both benefits and risks to individual users, organisations and nationally. 18.5.5 Engagement and user input in design As attitudes and a lack of perceived benefits can lead to resistance, user involvement and engagement at every stage of the design and implementation process is crucial to increase ownership and prevent alienation.12 14 21 25 28 33 34 36 37 40 45 47 67 68 112 114 116 118-120 This is particularly important when users feel that on them and the decision to implement has been made without their consultation as could be the case with NHS CRS software (resulting in a lack of psychological ownership).8 23 25 44 48 51 113 For example, Mair and colleagues have conducted a systematic review of 66 studies investigating engagement in and resistance to using eHealth systems amongst healthcare professionals.58 They identified barriers and facilitators to engagement and resistance and argue that implementation and integration issues are key to resistance and lack of professional engagement. Miranda and colleagues report that in this context user feedback may lead to an increased feeling of user involvement rather than a feeling that they were presented with a finished product.54 Therefore, efforts need to focus on user 491 involvement and implementation needs to be guided by consensus wherever possible.117 Here, it is also important that changes to the system are made as quickly as possible so that users can see that their input is incorporated.36 116 In support of the importance of active user engagement in EHR implementations, case studies of successful implementations often describe a substantial amount of user involvement. For example, Dagroso and colleagues describe the implementation of an obstetrics system in US secondary care.116 During the roll-out they uncovered problems with the design and functionality of the system. As a response to complaints from users, the implementation team reviewed the performance of the system, heavily involving users in the implementation process (e.g. on-site trainers were used to promote the system and collect user feedback). A new improved system was launched tailored to the needs of users which increased user acceptance and resulted in successful implementation. Design of the system should to be based on and tailored to individual needs and this can only be achieved through close collaboration between designers, IT-staff and end-users.17 32 36 39 40 42 42 46 50 51 53 66 70 79 83 94 101 104 118 118 119 The underlying assumption here is that users know their environment and the context of use of the system best, whilst developers of systems are often detached from the use of systems.109 121 For example, Erdley and colleagues describe how a new information system in US prenatal care was developed with no collaboration between designers and nurses (the users).78 The result was a ‘technically correct but unwieldy application’. In this context, a lack of customisability resulting in a system that does not meet the individual needs of users is a commonly cited barrier to EHR adoption.22 29 This may be addressed by implementing a combination of standardised core solutions and customised elements, corresponding to particular users and environments (so-called configurational technology).122 This can help designed solutions to be adapted to the particular context of use and help to facilitate ownership.123 However, some have stated that it is 492 exactly the limited local customisation that is one of the biggest risks to the success of the national Programme.93 Tailoring a new EHR system to individual needs also involves an analysis of the cognitive aspects of use, the findings from which then have to be incorporated into design.33 46 If the design is too complex then the cognitive demands on users increase, errors are more likely to happen and usability goes down.35 The difficulty here is that cognitive needs of healthcare professionals are likely to differ and that this needs to be incorporated into design. For example, Jonson and Turley undertook a cognitive task analysis to explore the needs of clinicians and nurses.43 Participants were asked to look at different cases on the EHR and were instructed to think aloud. They found that nurses tended to view the patient in observational terms (e.g. in relation to monitoring and documenting changes), whilst clinicians tended to view the patient in causal terms and in relation to decision-making (e.g. in relation to diagnosis, treatment). The authors conclude that clinicians and nurses have different practice models and that IT design needs to take these differences into account by designing different interfaces tailored to the needs of the professions. However, other healthcare professionals’ input in system design has so far somewhat been neglected. It has been argued that this has resulted in systems not being aligned with, for example, nursing needs.124 125 Moen describes EHR implementations from a nursing perspective.26 Nurses handle large amounts of information and present a large group of EHR users. It is therefore important to assess existing nursing work practices before design and facilitate nursing input so that the system can meet nursing needs.26 Nursing leadership is important in this context in order to identify barriers to implementation specific to nursing practice. Similar issues arise with other professions. No study included in the review has addressed any other members of the healthcare team such as administrative staff and pharmacists. 493 Some have also cautioned that user input in design pre-supposes that users know what they want and that this does not change over time. This is often difficult, especially when being presented solely with an idea of an electronic solution, without understanding how this may be applied in practice (see Chapter 16). It has therefore been argued that ideally design should be evolving and users should be able to try a system making subsequent changes to its design.126 This then also allows the system to respond to changing contextual requirements. However, the different requirements of different users may complicate this idea. In relation to the NPfIT a more participatory approach to development and implementation has repeatedly been advocated.120 Catwell and Sheikh outline that this should be characterised by early user involvement and development of a shared vision, formative evaluations and testing of prototypes to assess whether the system is perceived as usable, potential redesign so that the system fits with user’s needs, summative evaluation and benefits identification once implementation is completed, with incorporation of issues that have been identified along the way. NHS CFH has already made significant efforts to facilitate clinical engagement in the context of the national Programme.21 They have, for example, established a Communications and Engagement Team with the responsibility to facilitate communication between key stakeholders including NHS CFH and the NHS and other stakeholders such as the commercial sector (e.g. LSPs) or the media. This is divided into public engagement (aiming at increasing patient involvement) and clinical engagement (aiming at increasing clinical engagement). NHS CFH has also developed user informed design guidance as part of the NHS Common User Interface to address users’ needs.22 Despite these efforts, many have argued that there is still a lack of clinical engagement in the NPfIT and NHS CFH’s efforts have been criticised as lacking input from other user groups, which leads to the initiative being viewed as being imposed by the government by many.21 22 93 There is of course an important tension here because a national implementation of systems by definition means limited local customisation.22 110 494 Issue 13: Engagement and user input in design is limited in national implementations such as the NHS CRS. Recommendation 13: The key is that different user groups are not only consulted, but feel that they are listened to. The new strategic direction of the Programme with increased local input and choice may help to facilitate this. 18.5.6 Training and support Finding the time to allocate time for training in the busy healthcare environment can be difficult, but intensive training and support is important for the successful implementation of a new IT system as it can facilitate adoption. 17 23 28 29 32 36-38 40-42 44 49 51 53 61 62 64 83 85 94 113 115 118 119 127 Systems that are perceived as easy to learn tend to be adopted more readily and “well-trained” users tend to be more satisfied with a new system.17 66 Conversely, if training is not adequate this can decrease confidence and inhibit adoption.31 65 Some have even gone as far as to state that the lack of training users is one of the most important factors to contribute to large scale IT implementation failure.101 Most studies stated that in hindsight training efforts should have received more attention and therefore it is of primary importance to allocate sufficient time for training in advance.40 42 51 54 Planning training will also involve considering how it will fit in with clinical responsibilities. It is therefore important that the organisation supports training efforts by helping clinical users to free time e.g. by reducing their workload.45 Training needs to begin well before the new system is implemented and continue when it is established so that arising problems can be dealt with effectively and do not compromise care.12 25 69 118 The issue here is also that some functions are not practised and therefore often forgotten.36 For example, this may also involve training staff in using paper processes in case of system failure.76 Training and support should therefore be ongoing as users discover new functions and as software changes. Maximising use and gaining proficiency in using a new system may take years and should start well before actual implementation.12 13 32 35 36 38 42 66 70 128 495 Extensive user support during the initial “go-live” period is most important.25 51 61 127 live. 51 In addition, ongoing help has to be available when needed after goThis means quick turnaround and response times to arising problems.39 One successful implementation project had training coaches available who were reachable by pagers, which facilitated availability and timeliness of support.119 Ideally, special attention should be paid to target ward managers, senior staff and local leaders with thorough training so that they can help to train other staff.42 70 In one case study discussing the introduction of a documentation system at a hospital in Canada, for example, some nurses were trained as change managers and acted as coaches in the initial period after go-live.13 They provided support, but also collected feedback from users, which was then incorporated in the redesign of the system. The authors conclude that coaches were successful as they were peers of the users (nurses) and understood existing work processes. However, it needs to be kept in mind that training by peers means that one learns only what the peer shows and how he/she uses the system.42 A variety of training approaches can be used including web-based formats, classroom-based formats, reading material and one-to-one training formats. Studies indicate that training sessions with the possibility to try out the system and one-to-one training sessions are most effective.12 31 41 65 115 116 119 128 This can be done on the ward itself or through simulated clinical scenarios.32 49 70 91 Short sessions of on the ward training seem most effective, as this integrates best with busy clinical work schedules.36 66 115 Due to the nature of healthcare professionals’ work, a large amount of paper learning resources may not be appropriate.70 However, it has to be kept in mind that not all training formats and training contents are appropriate for all types of users. Users are likely to draw on the system in different ways depending on their role. Training therefore needs to be tailored accordingly in the light of existing levels of experience, existing 496 skills, existing needs and emerging problems.12 16 36 50 66 115 In this context, an initial assessment of current skills before training commences and associated tailoring of training efforts in line with levels of existing computer skills as well as assessments of the context of use is important.29 32 35 51 52 61 Again, training is complex in a national implementation and due to the large scale of the programme, limited resources, and tight implementation deadlines, in-depth training and intensive support are often not possible. Therefore, the responsibility for training in NHS CRS software has been devolved to local organisations. To support training of users, NHS CFH had set up a national “Education, Training and Development Programme”, working with organisations and suppliers to help with effective training by developing standards and providing guidance surrounding delivery and evaluation.129 However, some have cautioned that early implementers of NPfIT software tend to be enthusiasts and followers are likely to need more support. This has to be planned for in terms of resources and can easily be under-estimated.93 Due to some NHS CRS software still being in development (e.g. Lorenzo), dummy training giving users the chance to play around with the system, is complicated and has often not been achieved in relation to the NHS CRS. A Training Messaging Service (TraMS) with a 'Like Live' training environment was planned to be introduced and to enable NHS staff to practice on dummy patients for the NHS CRS, but the majority of users do not seem to be aware of this or have access to it.130 Issue 14: Central guidance in relation to training has to be limited and organisations need to devise their own approaches best suited to their staff and needs. Availability of software to play around with and tight implementation deadlines may compromise the quality of training. Recommendation 14: Organisations need to have increased input into implementation timelines so that they can prepare well for go-live. Dummy software reflecting the system that will ultimately be used needs to be made available. 497 18.5.7 Champions The value of key individuals in facilitating the adoption of information systems in healthcare has repeatedly been highlighted in facilitating adoption and in helping to embed systems in the organisation.131 Key individuals may include IT and clinical leaders, local champions, respected peers, opinion leaders, so called “super users” and “boundary spanners” (i.e. those that have dual roles).12 17 23 25 27-31 45 49-51 65 69 70 118 123 132 The role of these key individuals may involve winning over others, feeding back emerging problems and recommendations, facilitating user engagement, and communicating the vision.17 31 118 The underlying assumption is that if these key individuals start “modelling change” then others are likely to follow.103 As organic spread of benefits seems most effective, key individuals are ideally peers of the user group.42 Therefore, it has been argued that support of senior clinical staff is crucial in getting other clinical staff on board.115 Due to the complexity of the healthcare environment, however, one may need to consider involving departmental leaders from different clinical specialties.132 For example, Bossen describes how in a Danish EHR implementation, several groups were formed to facilitate cooperation between different stakeholders.28 Influence and range of influence of key individuals was found to be particularly important in this context. Similarly, further illustrating the variety of roles these individuals can play, Sicotte and colleagues describe two cases of EHR implementations in Canada. In one case, the impact of a selected project champion remained local, whilst in the other case there was a local clinical champion who was heavily involved in system design at each site.14 Conversely, if champions themselves become negative, then this can compromise implementation and inhibit adoption through impacting on attitudes of others.15 31 83 Greenhalgh and colleagues refer to “negative champions” in this context.31 In support of the important role of key individuals in facilitating implementation, Nagle reports on a Canadian government initiative to 498 promote EHR use and to promote healthcare professional engagement.124 In this context, bringing different groups of healthcare professionals together and establishing national and international peer networks with a view to discuss lessons learned was viewed as crucial amongst key stakeholders. Furthermore, highlighting the importance of key individuals in relation to training, De Mul and colleagues, describe the successful implementation of a Dutch intensive care unit system.118 They had “super-users” who helped others with learning how to use the system. It went live initially but due to a decrease in performance the ward returned to using paper systems. After more testing, refining the system, increased training efforts and incorporating user feedback, the system was implemented without further problems. Other studies have confirmed that an increased use of power users that can help out when needed can facilitate implementation.32 However, in the context of the national Programme, despite NHS CFH’s efforts to appoint and involve clinical champions on a national level, some have argued that their influence has not been sufficiently harnessed as they have constantly changed during the Programme.93 In addition, national champions can only influence peers to a limited extent as local champions seem more important to facilitate adoption.98 Also, some non-clinical staff groups have so far been neglected such as, for example, administrative staff. Issue 15: National champions only have limited influence locally and there is an increasing threat that champions become negative champions if NHS CRS software doe not align with their expectations or remains stagnant in terms of development. Recommendation 15: Despite the usefulness of national clinical leads, it seems that local clinical leads are most successful due to their increased contact to the target group. There is an urgent need to target other user groups and carefully assess the impact and address concerns of negative champions. 499 18.5.8 Integration with existing work processes New IT systems are often viewed as an opportunity to review existing work practices and increase organisational efficiency, getting users to adopt new ways of working desired by the organisation. It is important that the new system during this development is effectively integrated with existing work processes. Otherwise, the technology is likely to get rejected by users as healthcare staff need to retain patient safety as the top priority.14 22 26 27 33 35 36 38 42 47 51 65 68 80 133 If staff find it hard to integrate a new system into their day- to-day work practices they may adopt either “partial use” (i.e. only use the parts that are useful to them) or “workarounds” (avoid using the system, as discussed above).81 100 110 Some have argued that this is particularly likely in a situation where a new system is perceived as being imposed on users as is the case of the NPfIT.110 A new system is likely to result in new practices and can also lead to changes to existing professional responsibilities.28 30 47 104 110 For example, in one study clinicians felt that the introduction of the EHR resulted in them having more of a data entry role than was thought appropriate.47 This again illustrates the important role of user input in design as well as adaptation and customisation, in order to integrate systems with existing work practices and facilitate implementation.80 Even if effective planning and work process mapping occur, one will never foresee all changes in work practices that will be brought about by a new system in different settings. There is therefore always a need to try out the system in practice and then keep re-configuring it based on local experiences.81 Not having to make radical changes to work processes can, on the other hand, be a facilitator for adoption and implementation.40 Stewart and Williams have referred to the notion of “domestication” in this context.122 Here it is assumed that users adapt to and integrate the new technology with existing practices (this may involve getting around faults through workarounds), and eventually the technology becomes part of the everyday work routine. 500 Integration can also be facilitated by a thorough analysis of existing individual work practices before the system is introduced and a recognition that these vary across individuals and settings.12 17 25 33 35 44 45 50 67 71 118 132 134 System designs then need to be configured accordingly, so that they fit in with existing routines and the roles of different users.31 36 43 This requires the implementation team and system suppliers to understand the clinical context of use a top priority and to allocate sufficient time so healthcare staff can modify existing practices.40 132 Healthcare staff, on the other hand, need to recognise that new routines may have to be developed and existing work practices may need to change.36 This also means a recognition that new practices should not be mirrored on paper systems as this may limit potential effectiveness of the new EHR.54 In line with this, Lium and colleagues argue that users are often still trapped in “paper-thinking” and that this may inhibit adoption (paper is still used and viewed as valuable for certain tasks). According to these authors, paper routines need to be dropped and new routines n
