Implementation of HIV/AIDS Care and Treatment Services in Tanzania
Transcription
Implementation of HIV/AIDS Care and Treatment Services in Tanzania
THE UNITED REPUBLIC OF TANZANIA MINISTRY OF HEALTH AND SOCIAL WELFARE NATIONAL AIDS CONTROL PROGRAMME Implementation of HIV/AIDS Care and Treatment Services in Tanzania Report Number 3 May 2013 THE UNITED REPUBLIC OF TANZANIA MINISTRY OF HEALTH AND SOCIAL WELFARE NATIONAL AIDS CONTROL PROGRAMME Implementation of HIV/AIDS Care and Treatment Services in Tanzania Report Number 3 May 2013 Implementation of HIV/AIDS Care and Treatment Services in Tanzania Implementation of HIV/AIDS Care and Treatment Services in Tanzania C May 2013 Ministry of Health and Social Welfare National AIDS Control Programme (NACP) P.O. Box 11857 Dar es Salaam - Tanzania Tel: +255 22 2131213, Fax: +255 22 2138282 Email: [email protected] Website: www.nacp.go.tz ISBN No. 978 - 9987 - 650 - 81 - 1 Implementation of HIV/AIDS Care and Treatment Services in Tanzania Contents Contents..............................................................................................................................................i Acknowledgement.............................................................................................................................ii Executive Summary............................................................................................................................v 1. Introduction......................................................................................................................................1 1.1 HIV/AIDS Situation in Tanzania............................................................................................1 1.2 National Response...................................................................................................................1 1.3 National HIV/AIDS Care and Treatment Plan (NCTP)..........................................................2 1.4 Recommendations from the CTC 2 report.................................................................................2 2. Methods...........................................................................................................................................4 2.1 Overview of care and treatment’s Patient Monitoring System....................................................4 2.2 Formats for data reporting...............................................................................................................4 2.3 National Database.........................................................................................................................4 2.4 Analysis of Adults in Pre-ART Care.............................................................................................6 2.5 Analysis of Adults on ART...........................................................................................................6 2.6 Pediatric HIV and AIDS infection...................................................................................................9 2.7 Tracing unreported transfers, lost to follow up and clients picking drugs from multiple facilities....9 3. Results from the National Database..............................................................................................13 3.1 Care and treatment facilities reporting to national database............................................................13 3.2 National Estimates of people Infected with HIV and in need of ART.........................................15 3.3 Estimates of number of people with advanced HIV disease (in need of......................................18 3.4 Estimates of number of facilities per 1000 people infected with HIV.........................................19 4. Results fromAnalysis ofAdults in Pre-ART Care........................................................................21 4.1 Baseline Characteristics..............................................................................................................21 5. Results from analysis of Adults on ART....................................................................................27 5.1 Baseline characteristics...............................................................................................................27 5.2 Mortality in adults who started on ART......................................................................................31 5.3 Hazard rates and ratios..................................................................................................................34 5.4 Analysis of adults No Longer On Treatment (NLOT).................................................................37 5.5 Improvements in CD4 counts........................................................................................................39 5.6 Improvements in BMI................................................................................................................40 5.7 Switches to second-line therapy....................................................................................................42 5.8 Discussion.......................................................................................................................................47 5.9 Limitations.....................................................................................................................................50 5.10Recommendations................................................................................................................................51 6. Results from Analysis of Children on ART...............................................................................53 6.1 Baseline characteristics..............................................................................................................53 6.2.Outcomes of children starting ART...........................................................................................58 6.3. Mortality in children who started on ART..................................................................................59 6.4 Discussion.......................................................................................................................................63 6.5Recommendations.................................................................................................................................64 7. Tracing lost to follow up of HIV clients from multiple facilities in Mwanza region......................65 7.1 Introduction.....................................................................................................................................65 7.2 Results.........................................................................................................................................65 7.3 Discussion.................................................................................................................................................69 7.4 Recommendations....................................................................................................................................70 Annex A.........................................................................................................................................................73 Competing risks analysis................................................................................................................................73 Annex B.........................................................................................................................................................74 PARTICIPANTS TO PREPARATORY AND ANALYSIS WORKSHOPS...............................................74 Implementation of HIV/AIDS Care and Treatment Services in Tanzania i Acknowledgement This report narrates the delivery of HIV Care and Treatment in Tanzania, using the routine data collected at clinics throughout the country. This report builds on the first Care and Treatment report for Tanzania in 2008, and the second Care and Treatment report in 2010. The report comes from a series of workshops and training sessions in Dar-es-Salaam and Mwanza, involving many people and organisations in Tanzania and in other countries. Moreover, the organisation for the 2013 report has been developed by these dedicated groups working on set objectives, and this has been a learning exercise for all. The success of this report demonstrates the hard work and dedication of these groups. The National AIDS Control Programme (NACP), under the jurisdiction of the Ministry of Health and Social Welfare (MoHSW), has provided the direction and leadership for this work, and without their inputs this report would never have been completed. We are indebted to the health workers in all Care and Treatment Centres in Tanzania for providing Care and Treatment to HIV infected people, and for collecting the data, which has enable the monitoring and evaluation of the work they are doing. We thank participants to the preparatory and analysis workshops that has produced this report. We are grateful for the Technical assistance for the analysis of the data which was provided by London School of Hygiene and Tropical Medicine (LSHTM), the National Institution for Medical Research (NIMR Mwanza), Kilimanjaro Christian Medical University College (KCMU College), University Computing Centre (UCC), and AIDS Relief. We are also grateful for the financial support for the two workshops came from WHO Tanzania Country office. Special thanks to Jim Todd for spearheading both write ups (report 2 and 3) and empowering those involved in advanced statistical analysis skills. We also wish to thank the following individuals for their work in preparing and finalising this report: Aifello Sichalwe KCMC MSc Epidemiology and Biostatistics student Anath Rwebembera – NACP HIV Pediatric Programme officer Angela Ramadhani – NACP Program Manager Apaililia Kibona – Clinician Kilwa Road Police Hospital Bonita Kilama- NACP Acting Head, Epidemiology unit Denna Michael- NIMR Mwanza Researcher Elaine Baker- UCC Systems Analyst Emanuel MwendoKCMC MSc Epidemiology and Biostatistics student Filemon Tenu- KCMC MSc Epidemiology and Biostatistics student Fiona Vanobberghen - LSHTM & MITUNIMR MWANZA Medical Statistics Fellow Gretchen Antelman- ICAP Research And Evaluation Director ii Jim Todd – LSHTM and NIMR (Reader in Applied Biostatistics) Joseph Nondi- KCMC MSc Epidemiology and Biostatistics student Robert Josiah –NACP Deputy Program Manager Munda Elias- Applied Epidemiology Field Resident Naimi Mbogo – CSSC M&E Officer Neema Makyao- NACP KP Coordinator Paul Mahunga- TUNAJALI M&E Officer Ramadhani Mwiru- MDH Nutrition Support Coordinator Raphael Isingo – NIMR Mwanza Statistician Sesil Latemba – Pharmaccess M&E officer Tuhuma Tulli – Futures Group SI Advisor Implementation of HIV/AIDS Care and Treatment Services in Tanzania Gretchen Antelman- ICAP Research And Evaluation Director Hilda Mmari- Tanzania Prison Headquarters HIV Services Head James Juma – IeDEA Coordinator Japhet Kamala- UCC PEPFAR Support Jennifer Ward- CDC Research Fellow Tuhuma Tulli – Futures Group SI Advisor Veryeh Sambu- Data Manager- NACP Juma Mwinula – Medical Director Armed Forces –JKT Mariam Ngaeje- AGPHAI M&E Manager Moses Kehengu- Futures Group ICT Specialist The Ministry is committed to use this report as evidence base to strengthen delivery of HIV Care and Treatment services in Tanzania. Dr. Donan W. Mmbando Chief Medical Officer Ministry of Health and Social Welfare Implementation of HIV/AIDS Care and Treatment Services in Tanzania iii Abbreviations AIDS Acquired Immune Deficiency Syndrome ANC Ante-natal Care ART Antiretroviral Therapy BMI Body Mass Index C&T Care and treatment CHMT Council Health Management Team CTC Care and treatment clinics DACC District AIDS Control Coordinator DMO District Medical Officer EWI Early Warning Indicators for HIV Drug Resistance GFATM The Global Fund To Fight AIDS, Tuberculosis and Malaria HBC Home-based care HIV Human Immunodeficiency Virus HSHSP-III Health Sector HIV/AIDS Strategic Plan Number III HTC HIV Testing and Counseling ICAP At Columbia University, (formerly International Centre forAIDS Care and Treatment Programs) MMAM”Mpango wa maendeleo ya afya ya msingi” National Plan for Development of Primary Health Care M&E Monitoring and evaluation MOHSW Ministry of Health and Social Welfare NACP National AIDS Control Programme NBS National Bureau of Statistics NLOT No Longer On Treatment (patients who have not attended CTC for atleast 3 consecutive months) NTCP National Care and Treatment Plan PEPFAR President’s Emergency Plan for AIDS Relief PITC Provider Initiated Testing and Counselling PLHIV People living with HIV/AIDS PMS Patient Monitoring System PMTCT Prevention of Mother-to-Child Transmission RACC Regional AIDS Control Coordinator RHMT Regional Health Management Team RMO Regional Medical Officer TACAIDS Tanzanian Commission for AIDS TB Tuberculosis THIS Tanzania HIV/AIDS Indicator Survey THMIS Tanzania HIV/AIDS and Malaria Indicator Survey UA Universal Access (WHO recommended Indicators for Universal Access to ART) UNAIDS Joint United Nations Programme on HIV/AIDS UNGASS United Nations General Assembly Special Session on AIDS VCT Voluntary counselling and testing (now often referred to as CITC, client initiated testing and counselling) WHO World Health Organisation iv Implementation of HIV/AIDS Care and Treatment Services in Tanzania Executive Summary The National HIV/AIDS Care and Treatment Plan for Tanzania mainland was launched in 2004, with the aim of providing quality care and treatment for all people living with HIV/AIDS (PLHIV) in Tanzania. In 2010, the second report on Care and Treatment Clinic (CTC) services showed the impact of anti-retroviral therapy (ART) on the lives of PLHIV. This report (the third CTC report) builds on the second report and provides an update on the impact of ART on PLHIV, and tackles the further analysis of important data collected by the CTC program. These reports will become a regular part of the monitoring of the CTC program, and help policy makers and funders evaluate the program. Although there are many health facilities approved to deliver care and treatment in Tanzania, data are available on the numbers in care and treatment from 848 health facilities in Tanzania. Using the 2012 Tanzania HIV and Malaria Indicator survey (THMIS) and other resource materials, the projections done estimate 1.2 million PLHIV in the country among which 123,000 are children. This reports estimates that 65% of adults and 35% of children who need ART in Tanzania are receiving ART, which indicates that further efforts are needed to reach all those in need of HIV treatment. However, the need for services is uneven, with 5 regions in Tanzania having 50% of PLHIV (626,000 in total), and further efforts are needed to deliver CTC in regions of greatest need. Since the second CTC report, the data collection and reporting systems for the CTC program have been improved. Harmonisation of the reporting system has also contributed to better records being sent to NACP for this report, and for regular monitoring of the program. For this report, individuallevel patient records have been extracted from the national database and used for the analysis. The analysis has used a database with over 600,000 adult patients, and 8 million visits over 8 years of follow-up, which are spread across 19 regions in Tanzania. Therefore we believe that the results NACP, and utilised the personnel who have been trained in statistical analysis within NACP and partner organisations. These data and the results that can be generated from them are invaluable in in Tanzania, and are an inspiration to many in other countries. In line with the second report on Implementing HIV/AIDS Care and Treatment Services in Tanzania, of 2010, we found that two thirds of adults enrolling in CTC and starting ART are female. In children, the numbers of male and female children presenting for care and starting ART are similar. This analysis of data to the end of 2011 shows that the numbers newly enrolling in CTC and starting ART have remained consistent since 2009, and are no longer increasing as they did for the years 2004 through 2008. With new guidelines for ART eligibility, the median CD4 count for those starting ART has increased over the past 3 years, from 214 cells/µl in 2007 to 245 cells/µl in 2010, however CD4 documentation is poor in both pre and post ART initiation. The survival of PLHIV has improved since the second CTC report in 2010. This report shows that 9.7% in 2008, to 7.3% for those who started ART in 2010. There may be several reasons for the better survival, including the fact that PLHIV are now accessing ART with higher CD4 counts. After per annum. Implementation of HIV/AIDS Care and Treatment Services in Tanzania v This report shows that many of those who start ART continue to drop out of treatment and are classified as No Longer on Treatment (NLOT). The analysis shows that as much as one quarter of those who start ART are no longer receiving treatment from the clinic where they initiated ART. This is of serious concern, and this report devotes a chapter to the analysis of those who are lost to follow up, using the IQTools software developed by AIDS Relief, and applied to data from Mwanza region. Initial analysis shows that only 7% could be traced using the tool on the CTC3 data. Further research on the best way to use IQTools to trace those lost to follow up is being undertaken and hopefully subsequent results can show what has happened to these people. For the first time, we have analyzed data on switching to second line treatment, using competing risks analysis which allows for the different (competing) outcome of death, within a time-to-event model. The results show that less than 1% of those who initiated first line treatment were observed to subsequently switch to second line treatment. The rate of switching was constant and consistent across time, with about 0.35% switching in any year, giving a cumulative hazard rate of 2.4% after 6 years on ART. In this analysis, we only looked at those who actually switched to second line drugs, and we did not assess the need for second line drugs. This will be an aim for future work, but it will require good data on CD4 counts and treatment failures to determine what proportion of PLHIV require a change from first line to second line drugs and thus determine whether this need is being met within the current CTC program. This report has also analyzed patient level data for children and it has revealed children to have as impressive survival than adults. Using data from around 26,500 children who started ART, the mortality in the first year was 9.1% for those who started in 2005, and 8.3% for those who started in 2010. Data on children show smaller effects of sex, and better treatment outcomes. Risk of death was seen to be higher among children without baseline CD4 count, WHO Stage 4 and 3. vi Implementation of HIV/AIDS Care and Treatment Services in Tanzania 1. Introduction 1.1 HIV/AIDS Situation in Tanzania HIV and AIDS continue to be a cause of high morbidity and mortality in the world. More impact is seen in developing countries. WHO reported about 2.5 million children less than 15 years to be living with HIV infection and about 330,000 children were newly infected by HIV in 2011 (UNAIDS Report 2012). More than 90% of the children 0-14 years who acquired HIV infection in 2011 live in Sub-Saharan Africa. Between 2009 and 2011 many Sub-Saharan countries have registered decline in acquisition of new HIV infection in children ranging between 1-59%. In Tanzania the decline in acquiring new HIV infection among children is between 1-19% (UNAIDS Report 2012). Tanzania is facing similar trend in terms of HIV/AIDS burden. The National HIV prevalence is estimated to be 5.1% (THMIS 2012) with highest prevalence in Njombe region (14.8%). Prevalence of HIV among children 0-14 years is calculated from adult prevalence. With this projection, 123,000 children aged 0-14 years are estimated to be living with HIV and AIDS in Tanzania by end of 2011. Provision of pediatric HIV care and treatment in Tanzania is faced with many challenges. The biggest challenge is low enrolment of children into HIV care and ART services. Programmatic data from NACP shows current rate stands at 8% but the national target is 20%. Other challenges include poor retention of children under care and treatment and improper documentation of services offered to those already enrolled into care 1 UNAIDS 2012, 2PMS Review. Furthermore, data analysis, data use and dissemination of results for planning purposes is also a challenge especially at level of health facilities3. Due to this, MOHSW Tanzania has introduced data analysis exercise to ensure that pediatric HIV statistics are available and utilized for policy and decision making in Tanzania. of the analysis results in decision making. 1.2 National Response Tanzania has a generalised HIV epidemic as in most sub-Saharan African countries where the prevalence of HIV among pregnant women is 6.9% 4 The national response is coordinated by the Tanzania Commission for AIDS (TACAIDS), which was established through the enactment of a law, Act No. 22 of 2001 by Parliament. TACAIDS is legally mandated to provide strategic leadership, and In the health sector, the NACP has the responsibility of providing prevention, care, treatment and support services across the entire health care system. The NACP is also responsible for monitoring and evaluating the health sector response to HIV and AIDS. 1 UNAIDS (2012)Global Report: UNAIDS Report on the Global AIDS Epidemic 2012 UNAIDS, Geneva. 2 HIV Care Patient Monitoring System Review (2010) 3 HIV/AIDS/STI Surveillance Report number 22 4 Surveillance of HIV and Syphilis infections among Antenatal Clinic Attendees 2008 Implementation of HIV/AIDS Care and Treatment Services in Tanzania 1 1.3 National HIV/AIDS Care and Treatment Plan (NCTP) The government of the United Republic of Tanzania launched the HIV/AIDS Care and Treatment Plan in October 2003, and since 2004, a rapid roll-out of antiretroviral treatment (ART) has been ongoing. A number of partners have been working with the Government of Tanzania in this undertaking, including the governments of Sweden, Canada, Norway, the Netherlands and Japan, along with the US government through the President’s Emergency Plan for AIDS Relief (PEPFAR). UN Agencies such as UNDP, UNAIDS and WHO have also collaborated with international funding through the Global Fund for AIDS, TB and Malaria (GFATM). The implementation of care and treatment services at the health facility level has been regionalised, with each partner - AIDS Relief, Elizabeth Glaser Pediatric AIDS Foundation (EGPAF), TUNAJALI (Deloitte/Family Health International), International Centre for AIDS Care and Treatment Program (ICAP), Muhimbili Dar City Harvard (MDH) and Walter Reed - supporting one or multiple regions, and PharmAccess International (PAI) supporting the military and the uniformed forces. To contribute further to Tanzania’s development goals in health, the USG has now set transition mechanism to strengthen health systems and create ownership for sustainability of health programs particularly HIV care and treatment programs. Most of the implementing partners have collaborated with local NGO’s or establish one to facilitate the transitions. The University Computing Centre (UCC) supports database systems, including both the CTC2 database used at a large number of HIV care and treatment clinics and the CTC3 Macrodatabase database at NACP. By the end of 2011, Tanzania had more than 1100 health facilities approved to provide care and treatment services, estimated to enable more than 1,000,000 patients to access HIV Care services. Guidance on the implementation of care and treatment services is obtained from the Health Sector HIV/AIDS Strategic Plan (HSHSP-II) that runs from 2008 to 2012 including thematic areas covering care and treatment, and health systems strengthening. The interventions in the thematic areas include provision of services in the health facilities and community, laboratory strengthening, surveillance, operational research and public health evaluations. To facilitate entry to care and treatment services, new HIV testing approaches such as provider-initiated testing and counselling (PITC) and home-based counselling and testing have been introduced in several regions. Tanzania has benefitted from the investment in data collection at the CTCs. Through the generous support of various donors (PharmAccess International, PEPFAR, Global Fund for AIDS, TB and Malaria) a rigorous system of routine data collection was initiated to record the patients attending CTC. Through the years, the systems has been revised to accommodate new or changed recording and reporting obligations from implementing partners, donors and international reporting commitments such as Global AIDS Response Progress Reporting formerly UNGASS and the Universal Access to HIV services . 1.4 Recommendations from the CTC 2 report The second CTC report in 2010 made 9 recommendations about the provision of care and treatment for HIV positive people in Tanzania, and the reporting and analysis of data from the CTC programme (www.nacp.go.tz ). 2 Implementation of HIV/AIDS Care and Treatment Services in Tanzania This report will update the descriptive and analytical tables and results using de-identified patientlevel CTC data, and provide evidence for future recommendations. Following the recommendations of the second report, the data collection tools used in CTC clinics were revised and improved (Recommendation 1). The new reporting procedures are simplified using only one aggregated report (the monthly report) and supplemented by collection of the data extracted from the CTC2 database in clinics collecting electronic data (Recommendation 4). Since 2010 there has been greater recognition of the difficulties in tracking and tracing patients who transfer from one clinic to another, or who are lost to follow up (Recommendations 2 & 3). There is still an urgent need for further improvement in retention of patients in the CTC, but this report includes an extra chapter highlighting efforts to address this problem. The recommendation that regular analyses and reporting of the CTC data is undertaken (Recommendation 6), is taken up by this analysis 3 years later, but we recognise that it would be useful to have update reports annually. It would also be useful to disseminate findings more widely, and to bring the analysis programs to the regional and district level (Recommendation 5), but this will take resources and training, and may take longer to achieve (although it remains the long term vision). From the second CTC report, one peer-reviewed paper has been published 5, and one is in preparation ready for submission (Recommendation 8). This report shows that the distribution of HIV services in Tanzania is not uniform, and appropriate ways to address these inequalities are needed (Recommendation 7). Although much has been done, fully achieving improved access and HIV service coverage can only be achieved through greater local participation, continued investment in the various WHO Building Blocks for Health System Strengthening as well as Community System Strengthening avenues (Recommendation 9). 5 Somi, G., Keogh, S. C., Todd, J., Kilama, B., Wringe, A., van den Hombergh, J., Malima, K., Josiah, R., Urassa, M., Swai, R. and Zaba, B. (2012), Low mortality risk but high loss to follow-up among patients in the Tanzanian national HIV care and treatment programme. Tropical Medicine & International Health, 17: 497–506. doi: 10.1111/j.13653156.2011.02952.x Implementation of HIV/AIDS Care and Treatment Services in Tanzania 3 2. Methods 2.1 Overview of care and treatment’s Patient Monitoring System The patient monitoring system for National AIDS Control Program (NACP) runs through sets of recording and reporting tools. The system is distinctively centralized, with a national database which recognizes each patient by their unique CTC ID number from all over the country. All individuals enrolled in care are supplied with client held CTC-1 cards. Elementary records of patient encounters are captured within facility-held CTC-2 cards, which are the foundation of both the paper and electronic system in use. Health facilities with paper HIV Care Patient monitoring system provide aggregated reports (Quarterly and Cohort) while facilities with electronic system (CTC2 database) has additional reports it can provide including exportation of patient level data. The quarterly report provide cross sectional information of clients receiving Care and Treatment services whereas the Cohort report provides longitudinal details and key treatment outcomes of client on ART. Reports from facilities are aggregated by District AIDS Control Coordinators (DACCs) office then forwarded to Regional AIDS Control Coordinators (RACCs) office for regional compilation. After regional compilation, reports are forwarded to NACP for national compilation and verification. 2.2 Formats for data reporting The de-identified export files (without patient names) from facility-level electronic databases (CTC2 micros) are received at NACP’s central database (CTC2 Macro) via email. The regionalized program’s implementing partners together with regions’ Regional AIDS Control Coordinators facilitate this process. The national database contains applications which enable it to generate aggregate reports from individual level data and to combine individual level data with other aggregate data received at each quarter. The national care and treatment database contains two main sections – a “patient-level” section and an “aggregate-level” section. The “patient-level” section primarily contains data exported electronically from clinics using an electronic database, mostly the CTC-2 micros databases. The “aggregate-level” section contains data received on paper summary reports from clinics and entered into the national database at the NACP. 2.3 National Database Data definitions and Formats The data were abstracted as per regions and facilities reporting on aggregate and individual level and presented independently such that no single facility producing both reports was counted twice. The counts were abstracted for persons currently in care and patients on ART for both adults and children as of 2011 in each region. Total of 348 clinics had reported comprehensive patient level data for 2011, and for these clinics patient level data was used. For a further 379 clinics, data on current numbers of patients for 2011 was available from quarterly aggregate reports only. Patient-level data: Current in care and on ART and New enrolments and on ART overtime For each clinic in the set of 348, patient level data for 2011 was retrieved. For each quarter of 2011, a count was done of unique adult or child (<15 yrs on 1 July 2011) patients visiting the clinic during that quarter to obtain current in care counts. Similarly, counts were done of adult and child patients visiting the clinic in each quarter, who had an ARV status of 2-Start, 3-Continue, 4-Change or 6-Restart at the last visit observed during that quarter to obtain current on ART counts. For each 4 Implementation of HIV/AIDS Care and Treatment Services in Tanzania clinic the 4-quarter average was then calculated for adults and for children or those in care and those on ART. The earliest visit date for each patient was established as the enrolment date, and the clinic at which that first visit took place was set as the clinic where the patient enrolled. All patients enrolled were then counted by year of enrolment, adult/child status and sex. Visits on ART were compiled as a list of all visits with an ARV status of 2-Start, 3-Continue, 4-Change or 6-Restart, and the earliest visit from this list for each patient was provisionally set as the ART start date, with the clinic where the visit took place set as the clinic where ART was started. However, where reported transfer-in records showed an earlier ART start date than the earliest visit on ART for a patient, that patient was excluded from the patient level analysis (as it can be assumed that the patient started ART at a nonelectronic-reporting clinic). Age at ART start was calculated as the year ART was started minus the year of birth. These ART start dates were then counted by year of ART initiation, adult/child status, and sex. Selection of reporting facilities for patient-level analyses The reporting by facilities varies depending on whether they are either paper only or both paper and electronic sites. Whereas the latter provides comprehensive extracts that are suitable for analyses, the former are only capable of producing quarterly, indicators-driven aggregate (containing summary data for estimation of cumulative and current numbers) and patients’ cohort data. At any given time of the year, there are facilities reporting completely, partially, in duplicate, or not at all. Prior to analyses, the final dataset was de-duplicated (by site and quarter) – with the most comprehensive report being chosen for inclusion where there was more than one available. Aggregate Data: Current in Care and on ART and New enrolments and on ART over time For each clinic not already counted among the 348 electronic (patient level) reporting sites, aggregate level data were examined from quarterly reports in both old and new formats. Care was taken not to double-count figures reported as subsets (for example pregnant women in the new report are reported as subsets of the total as well as being included in the relevant female age groups). Age groupings are different in the old and new report formats, but both allowed a breakdown by adult and child less than 15. Total current in care and total current on ART (a subset of those in care) for adults and for children was calculated for each clinic and quarter. Figures were averaged across all 2011 quarters for which data was available. Totals were then calculated across clinics, by region. Data on new patients from both old monthly reports and new quarterly reports were used after deduplication. For 66 clinics, more than 12 months’ data in a calendar year were reported (mostly during 2011). For sites where monthly and quarterly reports covered the same period, a manual review was undertaken of all reports in order to exclude duplication and select the most complete report where there were differences observed in total counts. Age and sex breakdowns different between the old and new format reports, but both allowed breakdown by sex and adult/child. Totals of new in care and new on ART were calculated for each clinic for, by adult/child status and sex. Data Compilations In some cases the above analyses found a number calculated from patient-level data and a number calculated from aggregate-level data for the same clinic for the same year for the same category (care or ART). This occurred when clinics reported both patient-level data and aggregate-level reports Implementation of HIV/AIDS Care and Treatment Services in Tanzania 5 for the same time period. In these cases, the higher number was taken for that clinic for that year, resulting in only one number for each clinic for each year for each category for each adult/child and sex combination. These figures were then summed up across clinics. Table 2.4 represents the de-duplicated totals for all 860 clinics nationwide with exception of ART sites in Rukwa, Ruvuma and Mbeya. 2.4 Analysis of Adults in Pre-ART Care Adults ages 15 years and older from 2005 to 2011 were included in the analysis. Patient and Visits Tables, containing demographic and clinical data, respectively, from the CTC3 macrodatabase were merged, matching on the unique CTC Number. Exclusion criteria included ever having prior exposure to ART (except for PMTCT prophylaxis), and age less than 15 years at time of enrolment. Enrolment date into pre-ART care was defined as the earliest CTC visit date. Enrolment year was then extracted from the enrolment visit date. The dataset was transposed into a ‘flat’ file where all patient’s characteristics were in a single record. Variables specific to ART initiation, including adherence to ART and reason for ART initiation, were dropped, as they were not pertinent to the pre-ART population of interest. Body Mass Index (BMI) at enrolment date was calculated for each client by dividing weight at enrolment by the square of average height while in pre-ART care. As height is not expected to change much in adults (though slightly more among adolescents 15-20), average height was used to take into account actual growth as well as measurement error. BMI was then classified in accordance with current WHO category definitions for underweight (<18.5 kg), normal weight (18.5–25 kg), and overweight (>25 kg). Weight at each visit was also categorized as under 45 kg, 45–54 kg, and 55 kg or over. A binary variable identifying those clients with confirmed TB was created, grouping all those classified as ‘confirmed TB’ from codes such as ‘yes’, ‘confirm TB’, ‘TB RX’, ‘SS+’, ‘CXR+’, ‘START TB’, ‘CTN TB’, ‘RESTART TB’) All other TB categories including suspected TB and negative screening results were grouped into the new variable as ‘unconfirmed TB.’ The binary variable TB screening was also created, grouping all variables related to screening, including chest x-ray results, sputum sample results, and suspected TB into the new variable yes, screened for TB. Lastly, the source of client referrals included the sub-categories PMTCT, VCT, and TB to reflect referrals from programs offering HIV services. Baseline descriptive statistics were carried out for this population at enrolment date for the variables age, sex, marital status, facility level, referral source, BMI, weight, WHO clinical stage, CD4 cell count, pregnancy status, TB status, and functional status. To determine baseline CD4 count trends over time, a new variable for first CD4 count while in preART care was created to also capture those pre-ART clients with missing CD4 counts at enrolment date. All analyses were carried out using Stata version 12 statistical software (StataCorp. 2011. Stata Statistical Software: Release 12. College Station, TX: StataCorp LP). 2.5 Analysis of Adults on ART Data compilation Patient-level data were taken from the 348 clinics identified from the national CTC3 database. The 348 clinics used in this analysis do not represent every region in Tanzania (Rukwa, Mbeya and Ruvuma did not send the extract files to the national database in time for the analysis). Data 6 Implementation of HIV/AIDS Care and Treatment Services in Tanzania included all adult clinic visits from the first visit in 2004 through to 2012. The data were cleaned using Stata version 12, and two data files were generated: one with a single record for each person, and one with a record for each clinic visit (multiple records per person). All data used the CTC patient number as the unique identifier for the individual, which allowed the tracing of patients across clinics, with some patients attending up to 5 different clinics. A preliminary workshop was organised by NACP/MOHSW in June 2012, and identified the objectives and analyses for adults on ART. It was held at NIMR, Ocean Road and facilitated by staff from LSHTM based at NIMR Mwanza and NACP staff. Participants included staff members from implementing partner organisations as well as TACAIDS, MOHSW, and other organisations. Participants were requested to adapt the analyses which were performed for the second report in 2010, but additional longitudinal cohort methodology for this third report was introduced. The preliminary analysis using this new methodology was based on data from Zamzam hospital in Kagera, Meru Hospital in Arusha, Vikindu Dispensary and Mwananyamala Hospital, Sekou Toure hospital in Mwanza and Mirembe Hospital in Dodoma. The data from these sites were provided for the workshop for learning purposes. A follow-up workshop was held from 28th January to 8th February 2013 were all analyses in this report are a result of that workshop. Data definitions Patients were included in the analysis if they had started ART at one of the 348 clinics and were 15 years or over at the time that they initiated treatment. All data were censored at 31 December 2011. The outcomes for ART patients included: i) ii) iii) Current on ART Died Lost to follow up (LTFU) Patients were defined as LTFU if they had not attended any clinic (either the original clinic, or one they had transferred to of the 348 clinics) in the 6 months preceding 31 December 2011 (i.e. had last attended before 1 July 2011), and were not known to have died. Patients were considered to be currently on ART if they were not LTFU and had not died. Patient follow-up was censored at the earliest of: date of death, 6 months after the last clinic visit, or 31 December 2011. Patients were defined as No Longer On Treatment (NLOT) if they had died or were LTFU. If patients had transferred between clinics and records were available from more than one clinic, then the records were cross-checked to ensure that the ART start date used for the analyses corresponded to the earliest reported date of ART start. Data analysis The data analyses were conducted using Stata version 12. Patient characteristics at the time of ART initiation were described by the year of ART initiation, using frequencies and percentages for categorical variables, and the median and interquartile range (IQR) for continuous variables. Categories were combined for variables where some categories had very few participants. We investigated the distributions of some characteristics of interest by sex, by plotting bar graphs. Kaplan-Meier estimation methods were used to calculate the probabilities of patients surviving Implementation of HIV/AIDS Care and Treatment Services in Tanzania 7 up to 7 years after starting ART, stratified by baseline characteristics of interest. Those who were LTFU were censored. Cox proportional hazards models were used to investigate the associations of baseline characteristics with mortality. Since estimating mortality using CTC data is complicated due to the unknown survival status of patients who were LTFU, as a sensitivity analysis we analysed deaths and LTFU combined (NLOT). To define values at ART initiation for time-varying measurements, such as CD4 count and weight, we used the most recent value up to 3 months prior to ART initiation, or if no such value existed then the earliest CD4 within 2 weeks after ART initiation. To define values at switch to secondline therapy, we applied the same rule for CD4 selection (3 months before switch up to 2 weeks after switch). When incorporating time-dependent data into models, we carried forward the last observation for up to 3 months. In order to investigate CD4 changes over time after ART initiation, we plotted the median and IQR of CD4 count in six-monthly intervals, by baseline characteristics of interest. For each sixmonth interval, we allowed a window of -3 months to +2 weeks. For example, CD4 counts from 3 months up to just before 9 months were permitted to be included as the CD4 measurement “at” 6 months. However, we ensured each patient contributed at most one CD4 measurement during each interval; if there was more than one such measurement available, then the value measured closest to the nominal six-month interval was chosen (with preference for earlier). The six-month intervals were chosen to achieve a balance between the ideal interval between CD4 count measurements for patients on ART according to national guidelines (6months) and ensuring sufficient numbers of patients in each interval. A similar strategy was adopted to show BMI changes after ART initiation. Our final aim was to investigate the rate of, and factors associated with, switch to second-line therapy. While outcome for this analysis was switch to second-line therapy, death is a “competing risk” in this analysis because clients who died on first-line therapy could not reach the outcome of switching to second-line and it is highly likely that switch to second-line is associated with mortality (those who did not switch when they needed to are more likely to die). The usual approaches for time-to-event analyses, such as Kaplan-Meier estimation and Cox proportional hazards models (as we used for mortality and NLOT), will be biased in the presence of competing risks. Therefore we used appropriate methods, such as “competing risks regression”, to account for the competing risk of death. LTFU was assumed to be uninformative (not associated with the outcome). However, analogously to the mortality analysis, it is likely that those LTFU are more likely to have died, therefore future analyses should also consider combining LTFU and death (“NLOT”) as a competing risk. We firstly used unadjusted models and then a fully adjusted model. We combined the CD4 categories of 350-499 and ≥500 cells/mm3 into a single category (≥350 cells/mm3) due to few patients with such high CD4 counts. We considered only a subset of the variables which we thought would be most relevant to switching, namely facility level and ownership, year of ART initiation, sex, age, weight, WHO stage, CD4 count and ARV regimen. Individuals with intermittently missing data did not contribute follow-up during that time, but could re-enter the risk set when measurements were available. Intermittent regimens of duration ≤14 days were ignored. Individuals who changed to an unknown ART regimen were censored at that time. Individuals with missing ART information from the date when they were last known to be on firstline therapy until the date they switched to second-line therapy were assumed to have switched to second-line therapy at the mid-point between these dates. 8 Implementation of HIV/AIDS Care and Treatment Services in Tanzania 2.6 Pediatric HIV and AIDS infection Data for pediatric analysis was obtained from the CTC2 database and includes all clients on care and on ART below 15 years. The data used has been extracted from 348 health Facilities who reported to NACP by end of 2012 Data analysis The data analysis in this report was conducted using Stata version 12.0. Children characteristics at the time of ART initiation were described using cross-tabulations, frequencies and percentages for categorical variables, and using means, standard deviations and medians for continuous variables. Baseline characteristics of interest included sex, weight and age at ART initiation and CD4 at baseline. Clinical characteristics at baseline included WHO stage, CD4 % or count, body weight and age. The closest measurements of CD4% or CD4 count were recorded from the time of ART initiation which was taken to be 3 months prior to, or up to 2 weeks after initiating ART. At baseline and subsequent clinic visits, the number of pediatric clients screened for TB were recorded and calculated. The number of Person-years lived following ART initiation were computed, taking into account the date when ART was started, the last date the patient was seen, and the date of the data exported from each CTC 2 database. Kaplan-Meier technique was used to calculate the probabilities of patients surviving up to 5 years after starting ART and stratifying by baseline characteristics of interest. Smoothed hazard rates were also obtained to find the location of the peak period of mortality following treatment initiation and age- specific mortality rates were calculated to assess the effect of current age on mortality. In order to investigate CD4 changes over time after ART initiation, the lowest CD4 (%) or count per patient in each six month period following ART start was identified (where available). The six month periods corresponded to the recommended frequency for CD4 counts for ART patients (ref). For each six-month period from ART start, the median value of CD4 for all patients with CD4 data available was calculated. The median values for each follow up period were then displayed graphically by baseline characteristics of interest. Weight for age was used to identify change in weight gain after ART initiation. 2.7 Tracing unreported transfers, lost to follow up and clients picking drugs from multiple facilities Patient level data from Mwanza 50 CTCs were used to assess unreported transfers, Lost to Follow up and clients picking drugs from multiple facilities. The IQ tool developed by Futures Group was used to assess clients who were LTFU or had unreported transfer. Data of clients who had been on ART and with unknown or non documented treatment outcomes were subjected to IQ tool to see if there are any duplicates using unique CTC Id number. The duplicates generated were then analysed using SPSS version 17 by assessing the socio-demographic characteristics from CTC2 database export files. Data Extraction Patient-level data were retrieved from the CTC2 databases. All data were included from the time the facility began providing services up to September 2012. All patient data were exported to Access Implementation of HIV/AIDS Care and Treatment Services in Tanzania 9 using the ‘export for analysis’ feature from the CTC2 database. This feature excluded personal identifying data such as names, but did include the CTC unique patient identifying number and contact data (address where a client lives, including division, ward, village or street), which were used for patient tracing. Software: IQ Tools IQ Tools is software that deals with data validation, query building and reporting to allow facilities to develop their own analyses and reports. It has been built in the .net framework with an SQL back-end database in order to extract data. IQ Tools can connect to Microsoft SQL server, Microsoft Access, MySQL or any other backend for data analysis. The query merge within IQ Tool has been created purposely for tracing LTFU. In this analysis the IQ Tools was used to trace patients based on the information available in the CTC 2 database. This includes the CTC Number (the ID of the patient), the ward, village, and subvillage of the patient’s home. Inclusion of facilities Of the total 97 facilities registered and active in Mwanza, 29 are fully-computerised, 47 are purely paper-based, and 21 are paper-based but have data clerks visiting the facilities to support data entry. All of the 29 fully-computerised facilities were included in this analysis, except Bugando Paediatrics Department, which is a subset of the Bugando Referral Hospital. Nineteen of the paperbased clinics were included in this analysis; two clinics were excluded because one had problems with the patient identifiers, and the other’s database was corrupted. It was not feasible to include the purely paper-based facilities in this analysis. Therefore, a total of 47 (48% of 97) facilities were included in the analysis. Data Definitions For this analysis, we used the following definitions: HIV Clients “in care” include all those who have received care at the facility. These include both; those who have not started ART and those who have started ART. HIV Clients “on ART” include all those who have started treatment including transfers in while on treatment. LTFU (Lost to follow up): A patient was defined as LTFU if he/she was more than 3 months late for the last recorded appointment with no reports of death or transfer-out. We determined this in two ways: first, on the CTC2 database there is a variable indicating status, which includes report of LTFU from the clinics; and second, if a patient had no visit for over 3 consecutive months, we declared them as LTFU from the time of their last scheduled appointment. Patients meeting either of these definitions were considered LFTU. Of note, patients could be LTFU from one health facility, traced to another health facility, and then LTFU again. Therefore, rather than consider LTFU at the patient level, we refer to “LTFU- episodes”. Each patient can experience more than one LTFU episode. Traced patients: These are patients who were considered LTFU and were subsequently found to be accessing care and treatment in another facility. We aimed to trace all of the LTFU episodes. This process is described in detail below. 10 Implementation of HIV/AIDS Care and Treatment Services in Tanzania Figure 2.1: Tracing LTFU All LTFU episodes Patient ID matched in another facility (with inter date of presentation) Patient ID not matched in any other facility TRACED Contact address (with included division, ward, village or street) matched in another facility with later date of presentation) Date of birth and sex matched TRACED Any one of: date of birth and sex not matched NOT TRACED Tracing LTFU within the same district The following analyses were done for each district in Mwanza region: At the start of the analysis, we compiled one dataset for each facility within the district. This dataset contained information of all clients who were enrolled in care at that facility during the time period of the analysis (up to 30th September 2012). For each patient, information was extracted on the date of enrolment, the date of the last appointment in the facility, and whether he/she was on ART or not. In order to trace LTFU in a particular facility within the district, each facility in turn was considered was used to trace these patients in all other facilities within the same district. This was done using the including patient ID, along with the name of the health facility where they were considered as LTFU was traced. The ID of the traced client was then compared in all health facilities within the same clients were then counted only once in the health facility even if he/she was found active in more than one other facility. If the client was found to have presented at the traced facility prior to the date of presentation at the master facility, this original facility became the patient’s master facility. as “traced”. For example, if a patient attended (master) clinic X, then clinic Y (all within a given district), then he/she was counted as LTFU and traced (once) in relation to the master clinic X. This process allowed us to trace patients after the last date of attendance in the master clinic. However it does not record circular movement by patients whereby a patient may move from one Implementation of HIV/AIDS Care and Treatment Services in Tanzania 11 the most recent record in clinic X, this would be identified as a patient LTFU from clinic Y and traced in clinic X, rather than the opposite. This process was repeated for each facility acting as the master facility. Of note, if a person was LTFU and/or traced in relation to multiple master facilities, then this person was counted multiple times. For example if a person attended clinic X, then clinic Y and then clinic Z, then he/she would be counted as LTFU from clinic X and traced to clinic Y, and then LTFU from clinic Y and traced to clinic Z. Therefore our denominator for this analysis was the number of “LTFU-episodes”, where one person could contribute to the denominator more than once. Our outcome of interest was the number of “traced-episodes”, where one person could contribute to this more than once. The ideal scenario is that 100% of LTFU episodes are traced. These analyses were repeated separately for each of the 8 districts in Mwanza region: Nyamagana, Ilemela, Magu, Kwimba, Missungwi, Sengerema, Geita and Ukerewe. Tracing lost to follow up across districts within Mwanza region From each facility, a dataset containing a list of those remaining LTFU (who had not been traced to facilities within their district) was generated. These datasets were merged to make a district-level dataset that was then used to trace LTFU from other districts. Each district was set as a master, in order to trace LTFU in the particular district within Mwanza. The same algorithm and methods used to identify those traced between multiple facilities was used to trace LTFU between districts. However, since this part of the analysis was performed across districts, it did not yield information about the facilities from which patients were LTFU or traced. Data analysis: Data analysis was done using SPSS version 17. Clients who were LTFU were matched to the export files of the CTC2 database to get more information about their characteristics including their marital status. Cross tabulations, frequencies and percentages for categorical variables were used to display the results. Results are presented for all patients in care (“in care”) and those who were on ART just before being lost to follow up from their master facility (“on ART”; a subset of those in care). 12 Implementation of HIV/AIDS Care and Treatment Services in Tanzania 3. Results from the National Database 3.1 Care and treatment facilities reporting to national database In 2011, the NACP started phasing out a previous cross-sectional reporting format consisting of separate monthly reports (with data on new and cumulative patients) and quarterly reports (with data on current patients) and instead introduced a new quarterly report with combined data on new cumulative and current patients. As a result of this transition of systems and reporting formats in 2011, the national dataset for this past year is composed of a combination of patient-level data, aggregate-level data in old report formats, and aggregate-level data in the new report format. Table 3.1 shows that while 860 clinics reported in any format during 2011, only 727 reported in a way which included data on current numbers of patients. The table below shows the reports received in various formats by region. Table 3.1 Number of clinics submitting CTC reports in 2011 by type of report Report Formats Number of clinics reporting in 2011 Comprehensive patient-level data electronically 348 No comprehensive electronic data, but reported new quarterly report (including data on current patients) 341 No comprehensive electronic data, no new quarterly report, but reported old quarterly report (including data on current patients) 38 No comprehensive electronic data, no quarterly report, but reported opted out monthly reports (with data on new patients but not current patients) 133 Total facilities reporting in any format in 2011 860 The regional distribution of reporting facilities in all formats, ranging from incomplete to complete below highlights the results. Implementation of HIV/AIDS Care and Treatment Services in Tanzania 13 Figure 3.1: Regional distributions of counts of facilities reporting in various formats The national database contained patient-level data from 441 clinics (both electronically imported data and pre-2007 patient level data entered through CTC3 forms). Of these, 363 of these clinics reported recent data, signified by having at least some data (more than 10 visits) during 2011. Of these, 14 clinics did not have any data signifying that they only very recently began services, and were excluded, leaving 349 clinics. One of these clinics was found to have non-continuous data (with two years data missing from a stream of several years), and was excluded, leaving 348 clinics. These 348 clinics have data going back various lengths of time. Figure 3.2 below shows the distribution of the 348 selected clinics across various program times, since 2004. Fig 3.2: Clinics providing patient level data in 2011, by year start of care and treatment records. 14 Implementation of HIV/AIDS Care and Treatment Services in Tanzania Only 348 facilities which electronic data covering the year 2011 in its entirety with a continuous stream of data from other years were included in individual level data analyses. We note that recent electronic data was not submitted in time from Mbeya, Rukwa and Ruvuma regions to be included in this patient-level analysis. Numbers of new clients enrolled is observed to decrease after 2009. Figure 3.3: Number of new patients enrolled in care, by year Figure 3.4: Number of new patients on ART by year 3.2 National Estimates of people Infected with HIV and in need of ART Regional Population Estimates The regional estimates of selected populations groups were obtained from age-specific 2002 census projections (National Bureau of Statistics (NBS) Ref6). The populations were broken into three age groups, 0-14 years, 15-49 years of age, and those over 50 years of age. The NBS-projected population growth of 2.8% was used, and the same growth was applied to all 21 regions of mainland Tanzania and to all age groups. Estimates of Prevalence of HIV/AIDS, and numbers Infected with HIV. The regional HIV prevalence estimates among adults (age 15-49) from the Tanzania HIV/AIDS and Malaria Indicator Survey (THMIS) in 2012 were used for this report. The prevalence of adults aged 6 United Republic of Tanzania: Ministry of Planning, Economy and Empowerment, National Bureau of Statistics. Regional and Districts Projections Vol XII, Dec 2006, Dar-es-Salaam, Tanzania Implementation of HIV/AIDS Care and Treatment Services in Tanzania 15 50+ was estimated as 63% of the prevalence of adults aged 15-49 years, based on the unpublished estimated by dividing the adults’ (age 15-49) regional HIV prevalence by 8, which is close to the estimate from 1987 which showed the ratio of the HIV prevalence in children to adults was 1:7. However recent evidence, from a community study in Uganda where ART has been made available, suggests that the ratio of HIV prevalence between children and adults may be may be closer to 1:15,8 which HIV. The counts of persons infected with HIV were generated by multiplying the regional HIV prevalence estimates for each age group by the respective groups’ regional population estimates. This is totalled over the 21 regions to produce a national total of adults and children living with HIV. The population counts, the prevalence of HIV and the number of people infected with HIV by region and age group are then applied to the number of facilities in each region, enabling the calculation of the number of clinics per 1000 persons infected with HIV. 7 Jain V1, Liegler T, Kabami J, Chamie G , Clark TD, Black D, Geng EH, Kwarisiima D, Wong JK, Abdel-Mohsen M, Sonawane N, Aweeka FT, Thirumurthy H, Petersen ML, Charlebois ED, Kamya MR, Havlir DV; SEARCH Collaboration. Assessment of population-based Clin Infect Dis. 2013 Feb;56(4):598-605. doi: 10.1093/cid/cis881. Epub 2012 Dec 12. 16 Implementation of HIV/AIDS Care and Treatment Services in Tanzania Implementation of HIV/AIDS Care and Treatment Services in Tanzania 17 43,169,325 Total 19,167,877 729,497 1,182,221 953,519 684,533 1,306,312 969,781 598,584 356,404 680,141 928,000 1,260,005 868,306 530,650 1,680,154 442,600 761,655 588,518 2,013,719 610,135 1,218,102 805,041 0-14 years 20,106,422 840,823 1,742,551 1,004,777 906,355 1,152,255 780,688 853,591 463,639 645,035 815,914 1,246,606 1,068,973 644,917 1,680,535 517,848 683,409 688,503 1,698,058 651,248 1,028,479 992,218 15-49 years 0.4 0.9 0.4 1.1 0.6 0.4 0.5 0.4 0.2 0.6 1.1 0.5 0.5 0.5 0.7 0.8 0.9 0.9 0.4 0.7 0.3 0.6 3,895,026 0-14 yrs + Col 6/15.6 141,304 270,131 205,521 173,397 203,104 140,724 216,999 121,841 116,595 147,993 236,473 224,918 173,668 308,691 126,210 113,136 135,063 286,887 142,682 196,468 213,221 50+ years 5.1 3.2 6.9 2.9 9.1 4.8 3.4 3.8 2.9 1.5 4.5 9.0 3.8 4.1 4.2 5.9 6.2 7.0 7.4 3.3 5.5 2.4 15-49 yrs 3.2 2.0 4.3 1.8 5.7 3.0 2.1 2.4 1.8 0.9 2.8 5.6 2.4 2.6 2.6 3.7 3.9 4.4 4.6 2.1 3.4 1.5 123,037 2,918 10,197 3,457 7,787 7,838 4,122 2,843 1,292 1,275 5,220 14,175 4,124 2,720 8,821 3,264 5,903 5,150 18,627 2,517 8,374 2,415 1,031,371 26,906 120,236 29,139 82,478 55,308 26,543 32,436 13,446 9,676 36,716 112,195 40,621 26,442 70,582 30,553 42,371 48,195 125,656 21,491 56,566 23,813 15-49 yrs Col3*Col6 122,071 2,826 11,649 3,725 9,862 6,093 2,990 5,154 2,208 1,093 4,162 13,302 5,342 4,450 8,103 4,654 4,384 5,909 13,269 2,943 6,754 3,198 50+yrs Col4*Col7 8 9 10 Estimated number HIV positive 50+yrs 0-14 yrs Col6/1.6 Col2*col5 National population estimates 5 6 7 HIV prevalence from 2012 AIS ** -Estimated population from 2002 census, with projected annual increase of 2.8% per annum (Tanzania Population Projections for 2011) -HIV prevalence in adults aged 15-49 years taken from 2012 THMIS -HIV prevalence in children aged 0-14 years taken as the HIV prevalence in adults aged 15-49 divide by 8 -HIV prevalence in adults of age 50 years or more taken as the HIV prevalence in adults aged 15-49 divide by 1.6 (TAZAMA project, Mwanza: unpublished data). 1,711,624 3,194,903 2,163,817 1,764,285 2,661,671 1,891,193 1,669,174 941,884 1,441,771 1,891,907 2,743,084 2,162,197 1,349,235 3,669,380 1,086,658 1,558,200 1,412,084 3,998,664 1,404,065 2,443,049 2,010,480 Total Arusha Dar es Salaam Dodoma Iringa Kagera Kigoma Kilimanjaro Lindi Manyara Mara Mbeya Morogoro Mtwara Mwanza Pwani Rukwa Ruvuma Shinyanga Singida Tabora Tanga Region 2 1 3 4 Estimated Mainland Tanzania Population in 2011* Table 3.2: Population by age and region: Total population, estimated HIV prevalence, and number infected with HIV 1,276,479 32,650 142,082 36,320 100,127 69,239 33,655 40,433 16,946 12,044 46,098 139,671 50,087 33,611 87,506 38,471 52,658 59,254 157,552 26,951 71,694 29,427 Total Col8+Col9+Col10 11 3.3 Estimates of number of people with advanced HIV disease (in need of ART) The new WHO guidelines recommend that ART eligibility begins when CD4 counts reach 350 cells/ mm3 or below. The Tanzania NACP has begun implementing the proposed eligibility threshold 2011. We have obtained this by adding the following: - All those (both adults and children) who were on ART in 2010 multiplied by 0.9, which is an estimate of the 12 month survival rate for those ON ART All HIV-infected children under 2 years of age. From the 2010 data, 22% of children aged 0-14 years were under the age of 2 years. We have applied this to the HIV positive children in each region. Of the remaining HIV positive persons (both adults and children) in 2010, we estimated that 20% would become eligible for ART during the following 12 months. By adding these three components, we obtained the counts of persons with advanced HIV disease, who should be receiving ART in 2011 by region, for adults and children separately. It should be noted that these estimates cover only people who are already in care. 18 Implementation of HIV/AIDS Care and Treatment Services in Tanzania Table 3.3 Population of HIV positive adults and children, the number with advanced HIV disease and the number and proportion on ART in 2011, by region HIV Positive Region Col A HIV positive Col B Adults (15+ yrs) Col C Children (0-14 yrs) Patients in need of ART Numbers with advanced HIV disease Col D Col H Adults Children (15+ yrs) (0-14 yrs) Patients on ART Proportion of those with Patients on ART advanced disease on Col L Col I Col J Col K Adults Children Adults Children (15+ yrs) (0-14 yrs) (15+ yrs) (0-14 yrs) Arusha 29,732 2,918 10,022 1482 3720 486 37% 33% Dar es Salaam 10,197 Tanga 131,885 32,864 92,340 61,401 29,534 37,590 15,654 10,769 40,878 125,496 45,963 30,892 78,686 35,207 46,755 54,104 138,925 24,434 63,320 27,012 2,415 53,485 10,447 31,842 17,676 7,005 12,791 5,233 3,386 12,514 59,399 14,703 7,951 25,658 9,783 11,895 15,287 34,070 6,691 17,824 12,259 5521 1576 3856 3261 1633 1615 630 572 2128 8605 1874 1129 3832 1409 2323 2183 7377 1064 3469 1478 43758 7574 32423 4298 2027 7089 3925 1516 6474 47847 9468 4431 20825 6176 10615 11486 7155 3350 12653 1897 3353 788 2709 368 194 968 360 124 343 3228 724 377 1527 906 408 811 472 264 1037 250 82% 72% 102% 24% 29% 55% 75% 45% 52% 81% 64% 56% 81% 63% 89% 75% 21% 50% 71% 15% 61% 50% 70% 11% 12% 60% 57% 22% 16% 38% 39% 33% 40% 64% 18% 37% 6% 25% 30% 17% Total 1,153,441 123,037 379,921 57,015 248,707 19697 65% 35% Dodoma Iringa Kagera Kigoma Kilimanjaro Lindi Manyara Mara Mbeya Morogoro Mtwara Mwanza Pwani Rukwa Ruvuma Shinyanga Singida Tabora 3,457 7,787 7,838 4,122 2,843 1,292 1,275 5,220 14,175 4,124 2,720 8,821 3,264 5,903 5,150 18,627 2,517 8,374 3.4 Estimates of number of facilities per 1000 people infected with HIV This indicator measures both, the burden of care and service provision and program coverage in the currently reporting clinics and in each region respectively. It should be noted that, although the information are extracted for reporting facilities, the indicator is statistically comparable against total facilities ever authorized to provide care and treatment services. Table 3.5 below summarizes the counts of clinics available in each region per 1000 individuals infected with HIV. Implementation of HIV/AIDS Care and Treatment Services in Tanzania 19 Table 3.4 Number of persons infected with HIV, care and treatment clinics, and numbers of clinics per 1000 Persons infected with HIV by region in 2011. Arusha (a) Dar es Salam 26,349 184,806 Health facilities approved for CTC CTC clinics reporting Clinics with Clinics with Total Clinics with Patient level Aggregate current figures data** data only** 28 1 29 52 13 65 Dodoma 39,414 28 12 40 1.01 Iringa 166,202 37 18 55 0.33 Kagera 46,340 23 9 32 0.69 Kigoma 16,755 7 13 20 1.19 Kilimanjaro 19,524 17 9 26 1.33 Lindi 21,380 8 52 60 2.81 Manyara 11,423 12 9 21 1.84 Mara 74,528 10 3 13 0.17 Mbeya 135,716 0 49 49 0.36 Morogoro 64,526 26 14 40 0.62 Mtwara 28,349 1 50 51 1.80 Mwanza 110,945 27 8 35 0.32 Pwani 41,882 14 19 33 0.79 Rukwa 39,344 0 21 21 0.53 Ruvuma 47,828 0 41 41 0.86 Shinyanga 148,477 15 6 21 0.14 Singida 21,047 8 0 8 0.38 Tabora 78,679 8 27 35 0.44 Tanga 56,500 27 5 32 0.57 Total 1,380,013 348 379 727 0.53 Region Number of people infected with HIV * * Taken from Table 2.2 20 Implementation of HIV/AIDS Care and Treatment Services in Tanzania Number of facilities per 1000 infected with HIV 1.10 0.35 4. Results from Analysis of Adults in Pre-ART Care 4.1 Baseline Characteristics A total of 440,880 adult pre-ART clients ages 15 and above were enrolled in CTCs from 2005 – 2011. Their characteristics are shown in Tables 4.1 and 4.2. A total of 144,401 (32.7%) males and 296,471 (67.3%) females and 8 adults with unknown sex were enrolled in pre-ART care by December 31, 2011. The female to male ratio (2:1) was similar across all calendar years. The overall age structure of those who never started ART was similar for all years. About 40% of both male and female clients were between 30-39 years old at pre-ART enrolment. Overall, however, the age pattern differed by sex, with older males (45% age 40 and over) and younger females (35% under age 29). The percent of clients screened for TB increased over time from 87% from 2005-2007 to 94% in 2010, with a slight decrease in 2011 to 93%. Confirmed cases of TB decreased over time, from 3% in 2007 or before to approximately 0% in 2011. The number of pre-ART HIV positive clients enrolled into CTCs increased from 2003 to 2009, and decreased from 2009 to 2011. From 2005 or before to 2011, mean first CD4 cell count increased from 243 to 323 cells/µL. Implementation of HIV/AIDS Care and Treatment Services in Tanzania 21 Table 4.1 Baseline demographic characteristics of 440,880 pre-ART clients at CTC enrollment 2007 or before 2008 Category n % n % n % n % Total Sex Male Female Age (years) 121,671 27.6 80,036 18.2 86,831 19.7 83,672 19.0 68,670 15.6 39,886 81,781 32.8 67.2 26,273 53,759 32.8 67.2 28,184 58,647 32.5 67.5 27,269 56,403 32.6 67.4 22,789 45,881 33.2 66.8 32,414 26.6 23,241 29.4 26,056 30.0 24,653 29.5 19,692 28.7 49,584 27,198 12,468 7 40.8 22.4 10.3 0.0 31,856 16,756 8,182 1 39.8 20.9 10.2 0.0 34,145 17,742 8,886 2 39.3 20.4 10.2 0.0 32,665 17,312 9,036 6 39.0 20.7 10.8 0.0 26,911 14,393 7,658 16 39.2 21.0 11.2 0.0 15–29 30–39 40–49 ≥50 Missing Median (IQR) Marital status Single Married Cohabiting Divorced Widowed Missing Facility level Hosp Health C. Dispensary Facility type Private Government Referred from PMTCT VCT TB Missing 2009 2010 2011 n % 35.3 (29.6-42.5) 34.9 (29.1-42.2) 34.8 (28.9-41.9) 35.0 (28.9-42.2) 35.2 (29.0-42.5) 23,204 51,114 1,039 8,144 10,019 28,151 24.8 54.7 1.1 8.7 10.7 23.1 17,252 42,887 1,153 8,040 8,026 2,678 21.6 53.6 1.4 10.1 10.0 3.4 18,508 47,684 1,300 8,793 7,702 2,844 21.3 54.9 1.5 10.1 8.9 3.3 18,502 46,154 1,405 8,480 6,977 2,154 22.1 55.2 1.7 10.1 8.3 2.6 15,543 36,114 1,319 7,021 5,503 3,170 22.6 52.6 1.9 10.2 8.0 4.6 105,461 4,515 8,441 86.8 3.7 7.0 62,320 9,151 6,783 78.2 11.5 8.5 55,596 17,629 12,065 64.5 20.5 14.0 48,184 19,136 14,879 58.1 23.1 17.9 38,095 16,507 13,042 55.9 24.2 19.1 8,418 10.3 5,507 10.1 6,391 10.3 6,170 10.3 4,601 9.4 73,192 89.7 49,457 89.9 55,846 89.7 54,046 89.7 44,165 90.6 13 50 87 0.0 0.0 0.1 1 53 40 0.0 0.1 0.1 8 36 31 0.0 0.0 0.0 1 109 42 0.0 0.1 0.1 2 48 8 0.0 0.1 0.0 121,521 99.9 79,935 99.9 86,742 99.9 83,502 99.8 68,606 99.9 *The high number in Missing category may be due to importation of data and version differences of CTC 2 cards used 22 Implementation of HIV/AIDS Care and Treatment Services in Tanzania Table 4.2 Baseline clinical characteristics of 440,880 pre-ART clients at CTC Enrolment 2007 or before 2008 2009 2010 2011 Category n % n % n % N % n % Total Weight <45kg 45–54kg 55+kg Missing BMI Underweight Normal Overweight Missing WHO stage 1 121,671 27.6 80,036 18.2 86,831 19.7 83,672 19.0 68,670 15.6 23,912 43,666 47,473 6,620 19.7 35.9 39.0 5.4 14,773 29,438 31,704 4,121 18.5 36.8 39.6 5.2 15,263 31,603 35,135 4,830 17.6 36.4 40.5 5.6 14,596 30,114 34,641 4,321 17.4 36.0 41.4 5.2 12,092 25,114 29,033 2,431 17.6 36.6 42.3 3.5 5,892 13,053 4,118 98,608 4.8 10.7 3.4 81.0 5,516 11,804 3,470 59,246 6.9 14.8 4.3 74.0 5,564 12,463 3,502 65,302 6.4 14.4 4.0 75.2 5,073 11,837 3,595 63,167 6.1 14.2 4.3 75.5 6,641 14,005 4,246 43,778 9.7 20.4 6.2 63.8 20,398 16.8 18,960 23.7 23,445 27.0 24,557 29.4 22,068 32.1 2 22,838 18.8 19,875 24.8 23,221 26.7 24,493 29.3 19,733 28.7 3 32,500 26.7 25,855 32.3 25,102 28.9 22,568 27.0 18,410 26.8 13,496 11.1 9,541 11.9 8,554 9.9 7,965 9.5 6,845 10.0 32,439 26.7 5,805 7.3 6,509 7.5 4,089 4.9 1,614 2.4 5,077 4.2 3,333 4.2 3,518 4.1 3,258 3.9 3,021 4.4 50-199 10,230 8.4 6,615 8.3 6,591 7.6 5,369 6.4 5,423 7.9 ≥200 Missing 17,114 14.1 12,943 16.2 13,900 16.0 11,461 13.7 9,961 14.5 89,250 73.4 57,145 71.4 62,822 72.4 63,584 76.0 50,265 73.2 4 Missing CD4 cell count <50 Median (IQR) Functional status Working 214 (88-395) 240 (97-438) 247 (101-438) 245 (95-439) 225 (89-418) 80,223 65.9 65,805 82.2 76,468 88.1 76,191 91.1 62,616 91.2 Ambulatory 14,958 12.3 11,600 14.5 8,469 9.8 6,008 7.2 4,916 7.2 Bed-ridden 1,955 1.6 2,032 2.5 1,711 2.0 1,300 1.6 922 1.3 Missing Pregnancy status (among responding women only) Yes 24,535 20.2 599 0.8 183 0.2 173 0.2 216 0.3 4,045 5.1 5,959 11.2 7,196 12.4 8,088 14.5 7,106 15.7 No TB status 74,811 94.9 47,290 88.8 51,032 87.6 47,776 85.5 38,132 84.3 3,341 2.8 2,390 3.0 1,743 2.0 380 0.5 1 0.0 118,330 97.3 77,646 97.0 85,088 98.0 83,292 99.6 68,669 100.0 Screened 106,274 87.4 74,957 93.7 81,487 93.9 78,974 94.4 63,742 92.8 Not screened 15,397 12.7 5,079 6.4 5,344 6.2 4,698 5.6 4,928 7.2 Unconfirmed TB TB screening Implementation of HIV/AIDS Care and Treatment Services in Tanzania 23 Figure 4.1 Age structure of pre-ART clients at Enrolment, by sex Figure 4.2 Enrolment into pre-ART care from 2003 – 2011 24 Implementation of HIV/AIDS Care and Treatment Services in Tanzania Figure 4.3 Mean first CD4 count from 2003 – 2011 Discussion Potential explanations for higher enrolment of females into pre-ART care may be that women have more opportunities/access to health care delivery services or that females have more proactive health-seeking behaviors. The age structure by sex is reflective of the epidemic in Tanzania whereby women with HIV are younger than men with HIV. Women of reproductive age have access to PMTCT services, resulting in earlier diagnosis and subsequently earlier entry into pre-ART care. National guidelines state that 100% of clients should be screened for TB upon enrolment into CTC. We see that screening is generally above 90%; however this is not completely compliant with the guidelines. The decrease in confirmed TB cases over time may be due to changes in reporting indicators in the CTC2 form, decrease in overall risk of acquiring TB as more PLHIVs enrol early for Care and Treatment services. The low number of new clients enrolled into care in 2003 and 2004 may be due to the early stages of the program, lack of awareness and availability of CTC services, stigma, discrimination, or poor linkages between counselling and testing intervention with care and treatment services. The increase in enrolment of clients from until 2009 is likely attributed to national scale-up of care and treatment services in Tanzania. The decrease in enrolment from 2009 to 2011 may be due to stabilization of the epidemic and the positive effect of ART program as for treatment for prevention. The trend of increasing first CD4 cell count over time indicates that clients are enrolling in care earlier. This could be due to increased access to services and increased knowledge regarding seeking care. We see that mean CD4 count is above the eligibility criteria for initiating ART (CD4 of 250 cells/µL), showing that the majority of clients are receiving pre-ART care before reaching eligible status. Implementation of HIV/AIDS Care and Treatment Services in Tanzania 25 Recommendations • Implement strategies that promote male involvement in health-seeking and enhancing men’s access to health care services. There is need to increase the proportion of men who test for HIV and improve linkage to HIV care for those who test HIV-positive. • Promote strategies that will increase male engagement in reproductive child health , testing and counselling services. • Support more intervention that increase education and dialogue about intergenerational sex. • Improve TB screening documentation by having targeted supportive supervision and mentoring to health providers in Care and Treatment sites as a component in HIV and TB collaborative activities. • Consider developing strategies to enrol more people in care and treatment services using innovative testing approaches. 26 Implementation of HIV/AIDS Care and Treatment Services in Tanzania 5. Results from analysis of Adults on ART 5.1 Baseline characteristics A total of 255,143 patients aged ≥15 years were observed to initiate ART in the 348 electronic clinics by 31st December 2011. Their characteristics are shown in Table 5.1 by year of ART initiation. Over time, care of adults initiating ART has devolved from hospitals to health centres and dispensaries, with 91.1% of patients initiating ART in 2005 or before doing so in hospitals, compared to 58.7% in 2011. The ownership of facilities in which people initiated ART remained fairly stable over time, with 68.3% being government, 24.4% faith-based/voluntary and 7.3% private in 2011. Overall, 88,530 (35%) of the included patients were male and 166,607 (65%) were female. The male to female ratio was similar across all calendar years of ART initiation. The age distribution was similar across all years, with 20% aged 15-29 years, 41% aged 30-39 years, 26% aged 40-49 years and 13% aged 50 years or older. The age pattern differed between the two sexes with males tending to be older than females at the time of starting ART (Figure 5.1). For example, only 12% of males were under the age of 30 years at ART initiation, compared to 25% of females, and 18% of males were aged 50 years or older, compared to 10% of females. Over 50% of patients initiating ART were married or cohabiting, and around 25% single, with the distribution of marital status remaining relatively stable for the past 4 years. Functional status improved over time, with more people working, and fewer bed-ridden at ART initiation. Similarly, more people initiated ART at WHO stages 1 and 2 in later years, compared to more initiating at stages 3 and 4 in previous years. Further, fewer people initiated at CD4 <200 cells/mm3 in recent years compared to previously. Of note, 70.1% of all patient-visits were missing BMI, mainly due to height not ever being recorded for that patient. The most common regime used initially was d4T, 3TC and NVP but, since the elimination of d4T since 2010, the use of ZDV, 3TC and EFV has increased substantially, from 3.9% in 2007 to 41.0% in 2011. Figure 5.2 shows that, at the time of starting ART, 16,389 (27%) males had CD4 count <50 cells/mm3 compared to 24,243 (21%) females. Further, 27% of males were WHO stage 4 at ART initiation, compared to 22% of females. This indicates that men were initiating ART at later stages of disease compared to women (Fig 5.3) Implementation of HIV/AIDS Care and Treatment Services in Tanzania 27 28 Implementation of HIV/AIDS Care and Treatment Services in Tanzania Category BMI, kg/m2 Weight, kg Divorced/ separated 533 Widowed 951 Working 6109 Ambulatory 883 Bed-ridden 120 <45 2301 45-<55 3824 55+ 4802 <18.5 715 1060 1093 20,544 680 7.0 12.5 86.0 12.4 1.7 21.1 35.0 44.0 25.8 2007 (%) 1205 7.5 1757 11.0 13,951 84.4 2238 13.5 351 2.1 5553 24.0 9041 39.1 8533 36.9 1592 29.7 n 2008 (%) n 3100 3574 23,753 7023 871 8039 12,939 12,444 2611 8795 20,122 2395 11,598 22,101 6629 14,330 8649 4084 37 6835 10.6 12.2 75.1 22.2 2.8 24.1 38.7 37.2 29.3 4714 5376 35,632 7161 1030 9750 17,440 16,495 4083 11.2 12.7 81.3 16.3 2.4 22.3 39.9 37.8 29.5 28.1 10,218 25.5 64.3 26,669 66.6 7.7 3188 8.0 34.4 15,594 35.3 65.6 2 8,565 64.7 19.7 9015 20.4 42.5 18,347 41.6 25.7 11,246 25.5 12.1 5553 12.6 31-44 36 31-44 23.4 9611 22.8 5191 5174 41,574 5111 924 10,320 18,377 18,353 4198 10,244 28,744 3816 16,682 30,958 9883 19,592 12,030 6136 36 10,170 (%) 11.3 11.3 87.3 10.7 1.9 21.9 39.1 39.0 29.1 23.9 67.2 8.9 35.0 65.0 20.7 41.1 25.3 12.9 30-44 22.1 13.2 17.2 68.1 1.5 18.7 2009 6.8 3212 7.3 6220 6.8 4415 10.1 8120 84.1 35,389 80.6 32,187 2.3 915 2.1 714 33,699 13.2 44,161 17.3 47,642 n 4.5 2282 4.7 2287 87.9 28,234 2.9 771 23,418 9.2 30.7 5881 26.5 64.6 14,997 67.5 4.8 1336 6.0 34.0 8164 34.9 66.0 15,253 65.1 17.8 4513 19.3 42.5 9890 42.3 28.2 6248 26.7 11.6 2749 11.8 31-44 37 31-44 25.4 3953 24.7 4.1 1.5 91.1 3.3 11,277 4.4 2005 or before 2006 N (%) n (%) Dispensary 460 Health centre 174 Hospital 10,274 Other 369 Facility ownership Faith-based/ voluntary 3376 Government 7101 Private 524 Sex Male 3837 Female 7439 Age, years 15-29 2004 30-39 4783 40-49 3171 ≥50 1304 Median (IQR) 37 Marital status Single 1928 Married/ Facility level Total (row %) Variable 2010 (%) 5254 4871 44,311 3621 792 10,169 18,836 19,228 3988 10,653 30,510 3003 16,954 31,796 10,020 19,890 12,218 6620 37 10,758 7704 10,300 29,729 576 11.1 10.3 90.9 7.4 1.6 21.1 39.1 39.9 27.5 24.1 69.1 6.8 34.8 65.2 20.6 40.8 25.1 13.6 30-44 22.7 16.0 21.3 61.5 1.2 48,750 19.1 n 2011 4940 4302 42,345 3231 587 9350 17,585 18,710 4851 10,258 28,684 3062 15,701 30,495 9462 18,792 11,539 6394 37 31-44 10,073 8009 10,448 26,874 451 46,196 n 11.2 9.7 91.7 7.0 1.3 20.5 38.5 41.0 29.2 56.3 22.8 24.4 68.3 7.3 33.9 66.0 20.5 40.7 25.0 13.9 17.5 22.8 58.7 1.0 18.1 (%) Table 5.1. Baseline characteristics of the 255,143 patients aged ≥15 years who were observed to initiate ART in one of the 348 clinics by 31 December 2011. Implementation of HIV/AIDS Care and Treatment Services in Tanzania 29 18.5-<25 1568 25+ 493 1 712 2 1153 3 3100 4 1883 <50 2113 50-199 4119 200-349 1197 350-499 158 500+ 73 Median (IQR) 105 d4T, 3TC, NVP 9044 ZDV, 3TC, NVP 861 ZDV, 3TC,EFV 421 Other first line 333 Second Line 86 56.5 17.8 10.4 16.8 45.3 27.5 27.6 53.8 15.6 2.1 1.0 43-177 84.2 8.0 3.9 3.1 1.0 3028 56.6 5056 56.8 7984 57.7 733 13.7 1238 13.9 1769 12.8 1116 7.1 1755 5.9 3235 7.7 2957 18.8 4801 16.1 8037 19.1 7171 45.5 14,013 47.0 20,309 48.4 4525 28.7 9264 31.1 10402 24.8 4400 27.0 6066 23.9 7872 22.9 8730 53.5 13,699 54.0 18,322 53.4 2645 16.2 4675 18.4 6819 19.9 361 2.2 600 2.4 817 2.4 197 1.2 333 1.3 482 1.4 105 45-182 119 52-191 123 55-195 18,937 83.3 26,291 79.3 34,112 78.0 991 4.4 1748 5.3 3339 7.6 1181 5.2 3118 9.4 4131 9.4 1194 5.3 1751 5.3 2059 4.7 443 2.0 265 1.0 100 <1 8402 58.2 1826 12.7 4142 9.1 10,318 22.7 21,107 46.3 9978 22.0 7849 22.4 18,267 52.1 7569 21.6 930 2.7 483 1.4 130 57-201 21,599 45.9 6885 14.6 15,030 32.0 1749 3. 7 1785 3.8 8385 2106 5239 12,173 20,733 9896 6760 14,939 8058 988 533 140 9862 11,424 21,544 1663 3864 57.9 14.6 10.9 25.3 43.2 20.6 21.6 47.8 25.8 3.2 1.7 60-221 20.4 23.6 44.6 3.4 8.0 9237 2528 5686 11,446 19,475 9152 5572 12,593 7596 1073 535 148 6907 13,876 18,831 4795 1498 55.6 15.2 12.4 25.0 42.6 20.0 20.4 46.0 27.8 3.9 2.0 65-234 5.1 30.2 41.0 10.5 3.3 IQR = interquartile range; BMI=body mass index; d4T=stavudine; 3TC=lamivudine; NVP=nevirapine; ZDV=zidovudine; EFV=efavirenz. Notes: 1. Percentages are column percentages, except for the first row where the percentages are row percentages. 2. The totals by characteristic may not sum to the total number of people due to missing data as follows: 1652 (1%) facility level; 21,567 (8%) facility ownership; 6 (<1%) sex; 52 (<1%) age; 22,590 (8.9%) marital status; 13,525 (5.3%) functional status; 3054 (1.2%) weight; 178,752 (70.1%) BMI; 21,366 (8.4%) WHO stage; 77,720 (30.5%) CD4 count; and 3426 (1.3%) ARV regime WHO stage CD4 cell count ARV regimen Figure 5.1. Age at ART initiation, by sex. Figure 5.2. CD4 count at ART initiation, by sex. 30 Implementation of HIV/AIDS Care and Treatment Services in Tanzania Figure 5.3. WHO stage at ART initiation, by sex. 5.2 Mortality in adults who started on ART Among the 255,143 patients who were observed to initiate ART by 31 December 2011, 255,141 had available follow-up data (after permitting 6 months until considered LTFU) and were included in the survival analyses; the remaining two initiated treatment on 31 December 2011 and therefore did not contribute any follow-up time. Table 5.2 shows the numbers of patients who started ART in each year, and the mortality within each calendar year cohort. Deaths and mortality rates are shown by the year in which the patient started ART, and by time since ART initiation. Table 5.2. Number of deaths and cumulative mortality risk, by year of ART initiation. 2005 or before 11,277 2006 2007 2008 2009 2010 2011 23,418 33,699 44,161 47,642 48,750 46,196 Within 6 months In 6 months-<1 year In 1-<2 years 1166 2420 2609 3438 3327 2993 2165 272 502 651 726 631 503 105 307 607 761 819 672 298 - In 2-<5 years 447 819 795 551 198 - - In 5-<7 years 123 76 - - - - - 2315 4424 4816 5534 4828 3794 2270 45,871 81,112 99,804 107,970 88,485 59,502 20,922 Individuals on ART Number of deaths Total Person-years Cumulative risk of death per 100 persons initiating treatment Within 6 months 10.3 10.3 7.7 7.8 7.0 6.1 5.5 By 1 year 13.0 12.8 9.9 9.7 8.5 7.3 - By 2 years 16.2 15.9 12.8 12.1 10.4 - - By 5 years 21.5 20.9 - - - - - By 7 years 24.2 - - - - - - Implementation of HIV/AIDS Care and Treatment Services in Tanzania 31 A total of 27,981 deaths were reported to the CTC clinics and included in this analysis, which represents 11.0% of all the patients observed to initiate ART. The risk of dying within 6 months of in 2005 or before, to 5.5% (5.3 to 5.7) for those who started in 2011. The risk of dying before completing 1 year on ART fell from 13.0% (12.4 to 13.6) in those who started in 2005 or before, to 7.3% (7.1 to 7.6) for those who started in 2010 (could not be assessed for those who started in 2011, as they had less than 1 year of follow up at the time of analysis). Figure 5.4. Cumulative probability of death after ART initiation. Figure 5.4 shows cumulative mortality risk over time following initiation on ART. The table in the Figure 5.5. Cumulative probability of death after ART initiation, by sex. 32 Implementation of HIV/AIDS Care and Treatment Services in Tanzania Figure 5.5 shows that the cumulative probability of death over time was higher among males than females. For example, the probability of death among men was 9.3% by 6 months and 24.0% by 7 years, compared to 6.2% and 16.0%, respectively, among women. This may be due to severity of disease at the time of ART initiation, as 27% of males initiated with CD4 count <50 cells/mm3, compared to only 21% of females, and 27% of men were WHO stage 4 at ART initiation, compared to 22% of women (see Figures 5.2 and 5.3). Figure 5.6a. Cumulative probability of death after ART initiation for males, by weight at initiation. kg <45 (%) 45-54 (%) 55 + (%) 0.50 6 months 18.4 10.6 5.7 1 year 22.3 13.3 7.4 2 years 26.9 16.6 9.8 5 years 35.1 23.2 15.1 7 years 38.4 27.7 18.3 0.40 0.30 0.20 0.10 0.00 0 1 2 3 4 5 Time since ART initiation, years Weight at ART initiation, kg <45 45-<55 6 7 55+ Figures 5.6a and 5.6b show that the cumulative probability of death over time for both males and females was higher among those with weight less than 45kg at ART initiation, compared to those with higher body weight. For example, the probability of death among males with weight <45kg at ART initiation by 6 months was 18.4% compared to 5.7% among those with weight ≥55kg. For Figure 5.6b. Cumulative probability of death after ART initiation for females, by weight at initiation. 0.50 Kg <45 (%) 45-54 (%) 55 + (%) 0.40 6 months 11.7 5.0 3.0 1 year 14.0 6.3 4.0 2 years 17.1 8.3 5.4 5 years 21.4 11.8 8.6 2 3 4 5 Time since ART initiation, years 6 7 years 23.3 14.5 11.1 0.30 0.20 0.10 0.00 0 1 Weight at ART initiation, kg <45 45-<55 7 55+ Implementation of HIV/AIDS Care and Treatment Services in Tanzania 33 Figure 5.7. Cumulative probability of death after ART initiation, by CD4 count at initiation. Cells/mm3 <50 (%) 50-199 (%) 200-349 (%) 350-499 (%) >=500 (%) Missing 0.50 0.40 6 months 12.8 5.6 3.8 5.8 6.0 8.3 1 year 15.2 7.2 5.1 7.7 7.7 10.3 2 years 18.2 9.2 6.8 10.0 10.4 13.2 5 years 22.6 13.1 10.7 13.3 15.0 18.8 7 years 25.5 17.2 14.0 15.4 16.0 21.0 0.30 0.20 0.10 0.00 0 1 2 3 4 5 Time since ART initiation, years CD4 at ART initiation, cells/mm <50 350-499 50-199 500+ 6 7 3 200-349 Missing Figure 5.7 shows that the cumulative probability of death over time was higher among those with CD4 counts of less than 50 cells/mm3 at ART initiation. For example, the probability of death by 6 months was 12.8% for those who initiated with CD4 <50 cells/mm3, compared to 5.8% among those with CD4 350-499 cells/mm3. We can also see that those with no CD4 count recorded at ART initiation were at somewhat higher risk of death, with probability of death by 6 months of 8.3%. 5.3 Hazard rates and ratios Mortality rates in the survival analysis can best be described as hazard rates, which estimate the probability of death at any instant in time (see 2010 report www.nacp.go.tz). Hazard rates, estimated in Cox regression models, were calculated from the time patients started ART and used in statistical models to formally compare the hazards experienced by sub-groups within the cohort, represented by hazard ratios. Table 5.3 compares the hazard rates between groups with different baseline characteristics. The baseline group for each factor was chosen as the most frequently reported characteristic. Of note, Cox regression model. Encouragingly, our results indicate a lower risk of death in later years, which may be related to presentation earlier in infection, improved disease management, and, most likely, increased The results show a higher risk of death in patients enrolled in dispensaries and health centres compared to hospitals (adjusted hazard ratio = 1.12, 95% CI 1.05-1.20; and 1.14, 95% CI 1.09-1.21, respectively), in faith-based compared to government facilities (adjusted hazard ratio = 1.10, 95% 34 Implementation of HIV/AIDS Care and Treatment Services in Tanzania CI 1.05-1.14) and those who were single compared to married/cohabiting patients (adjusted hazard ratio= 1.22, 95% CI 1.17-1.27). For the different age groups, there is a significant increased risk of death for older age groups, but this is likely to be due to higher underlying mortality rates in older people. The results show clearly a substantially lower risk of death in females compared to males, which persists after adjustment for other baseline characteristics (adjusted hazard ratio = 0.63, 95% CI 0.60-0.65). Regarding functional status, patients who were bedridden or ambulatory at the time of ART initiation had higher risk of death compared to patients who were working (adjusted hazard ratio = 2.94, 95% CI 2.71-3.20; and 1.72, 95% CI 1.64-1.79, respectively). The results show an increased risk of mortality with lower weight and higher (worse) WHO stage at ART initiation, as anticipated. Similarly, lower CD4 count at ART initiation was associated with higher mortality risk, although there was also an increased risk of death for those who initiated with CD4 counts above 500 cells/μl, compared to 200-349 cells/μl, even after adjustment for other covariates. Initiating at such high CD4 counts is unusual, and it may be that these patients were particularly sick, which may not be completely captured by the variables measured. Implementation of HIV/AIDS Care and Treatment Services in Tanzania 35 Table 5.3. Mortality hazard ratios for baseline characteristics of adults who initiated ART Characteristic Facility level Facility ownership Year of starting ART Sex Age, years Marital status Functional status Weight, kg BMI, kg/m2 WHO stage CD4 count, cells/mm3 ARV regimen Category Dispensary Health centre Hospital Other 0.59 HR 0.82 0.92 1 Crude 95% CI 0.78-0.85 0.89-0.95 (reference) 0.53-0.66 p-value <0.001 <0.001 HR Adjusted* 95% CI 1.12 1.14 1 0.79 1.05-1.20 1.09-1.21 (reference) 0.68-0.92 1.10 1 0.97 1.05-1.14 (reference) 0.90-1.06 1.36 1.15 0.96 1 0.97 0.92 0.88 1.25-1.48 1.07-1.23 0.91-1.01 (reference) 0.92-1.02 0.87-0.98 0.81-0.95 1 0.63 (reference) 0.60-0.65 0.95 1 1.07 1.35 0.90-1.00 (reference) 1.03-1.12 1.28-1.42 1.22 1.17-1.27 Faith-based/ voluntary Government Private 1.09 1 0.84 1.06-1.12 (reference) 0.79-0.88 2005 or before 2006 2007 2008 2009 2010 2011 1.42 1.38 1.08 1 0.87 0.75 0.70 1.35-1.49 1.33-1.44 1.04-1.12 (reference) 0.84-0.91 0.72-0.78 0.67-0.74 Male Female 1 0.64 (reference) 0.63-0.66 15-29 30-39 40-49 ≥50 1.00 1 1.06 1.33 0.97-1.03 (reference) 1.03-1.10 1.28-1.37 Single Married/ cohabiting Divorced/ separated Widowed 1.18 1.14-1.21 1 (reference) 1 (reference) 1.09 0.91 1.05-1.14 0.87-0.95 1.05 0.96 0.99-1.11 0.09-1.01 Working Amburatory Bedridden 1 2.36 4.60 (reference) 2.29-2.43 4.37-4.85 1 1.72 2.94 (reference) 1.64-1.79 2.71-3.20 <45 45-54 55+ 2.71 1.52 1 2.63-2.80 1.48-1.57 (reference) 2.24 1.37 1 2.13-2.34 1.32-1.43 (reference) Underweight Normal Overweight 2.22 1 0.69 2.10-2.36 (reference) 0.61-0.76 1 2 3 4 0.42 0.66 1 2.03 0.39-0.45 0.63-0.68 (reference) 1.97-2.08 <50 50-199 200-349 350-499 500+ 2.79 1.35 1 1.44 1.51 2.67-2.92 1.29-1.41 (reference) 1.30-1.59 1.32-1.72 d4T, 3TC, NVP 1 ZDV, 3TC, NVP 0.60 ZDV, 3TC, EFV 0.72 1.03 0.87 (reference) 0.58-0.63 0.70-0.74 0.98-1.08 0.81-0.93 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 0.58 0.78 1 1.46 0.53-0.63 0.74-0.82 (reference) 1.40-1.52 2.10 1.29 1 1.12 1.19 1.99-2.22 1.22-1.36 (reference) 0.99-1.27 1.02-1.39 1 0.87 0.84 0.96 1.09 (reference) 0.81-0.92 0.80-0.88 0.89-1.02 0.98-1.21 p-value <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 * Adjusted for all variables indicated, namely facility level, facility ownership, year of starting ART, sex, age, marital status, functional status, weight, WHO stage, CD4 count and ARV regimen. BMI was omitted from this multivariable model due to many missing values. Note: p-values are from Wald tests. 36 Implementation of HIV/AIDS Care and Treatment Services in Tanzania 5.4 Analysis of adults No Longer On Treatment (NLOT) The results from the analysis of time to NLOT (death and LTFU combined) are shown in Table 5.4. Numbers NLOT and cumulative risks are shown by the year in which the patient initiated ART, and by time since ART initiation. Table 5.4. Number NLOT and cumulative risk of NLOT, by year of ART initiation. 2005 or before 11,277 2006 2007 2008 2009 2010 2011 23,418 33,699 44,161 47,642 48,750 46,196 Within 6 months In 6 months-<1 year 1166 2420 2609 3438 3327 2993 2165 1422 3727 5703 7917 9059 9894 3476 In 1-<2 years 1078 2024 3472 4369 5174 3644 - In 2-<5 years 1705 3814 5314 4825 2434 - - In 5-<7 years 673 605 - - - - - 7+ years 1 - - - - - - Total 6045 12,590 17,098 20,549 19,994 16,531 5641 Person-years 45,870 81,112 99,803 107,970 88,485 59,502 20,922 Individuals on ART Number NLOT Cumulative risk of NLOT per 100 persons initiating treatment Within 6 months 13.6 14.2 12.1 11.8 11.6 11.1 10.7 By 1 year 23.0 26.3 24.7 25.7 26.0 26.4 - By 2 years 32.5 34.9 35.0 35.6 36.9 - - By 5 years 47.6 51.2 - - - - - By 7 years 56.2 - - - - - - A total of 98,448 adults were recorded as being no longer on treatment (NLOT), representing 38.6% from 13.6% for those who started ART in 2005 or before to 10.7% among those who started in 2011, the NLOT rates by 1 year increased from 23.0% for those who started ART in 2005 or before to 26.4% for those who started in 2010. as LTFU if more than 6 months has passed since ART initiation, which explains the sudden rise in the cumulative probability of NLOT at 6 months of follow-up. initiation patients were only included within the NLOT group if they were reported as having died. The cumulative risk of NLOT rose rapidly to 11.8% by 6 months, 25.7% by 1 year, and up to 61.9% by 7 years. Implementation of HIV/AIDS Care and Treatment Services in Tanzania 37 Figure 5.8. Cumulative probability of NLOT after starting ART. Figure 5.9. Cumulative probability of NLOT after starting ART, by sex. Figure 5.9 shows that males were consistently around 5-9% more likely to be NLOT than females over time. For example, the risk of NLOT by 6 months was 14.4% among men, compared to 10.3% among women, and by 7 years was 67.8% among men compared to 58.8% among women. 38 Implementation of HIV/AIDS Care and Treatment Services in Tanzania 5.5 Improvements in CD4 counts Improvement in CD4 counts after starting ART is an important indicator of the restoration of the just less than 100 cells/mm3. The overall picture is of sustained improvement up to 6 years on ART, . Figure 5.10. Improvements in CD4 count after initiating ART. Figure 5.11. `Improvements in CD4 counts after initiating ART, by sex. The median CD4 count in females initiating ART was slightly higher than for males (Figure 5.11), Implementation of HIV/AIDS Care and Treatment Services in Tanzania 39 and over time the rate of improvement in CD4 counts appeared to be better for females compared to males. Of note, the graph is only plotted to 5 years, due to less data when we divide into sub-groups. 800 600 3 CD4 count cells/mm (Median) Figure 5.12. Improvements in CD4 counts after initiating ART, by CD4 at ART initiation. 400 200 0 0 1 2 3 Time since ART initiation, years CD4 count at ART initiation, cells/mm <50 350-499 50-199 500+ 4 5 3 200-349 As expected, patients who started ART with a lower CD4 count showed better improvements in the (Figure 5.12). However, even after 5 years there remained some differences in the median CD4 counts between groups, with those who initiated at lower CD4 counts having lower CD4 counts up to 5 years later. Of note, the “bumpy” line for those who initiated with CD4 counts ≥500 cells/mm3 is due to few patients included in that group and therefore greater uncertainty and more variable estimates. One limitation to this analysis is that CD4 counts can only be measured in those who survive and come to the CTC clinic regularly. Thus if people with lower CD4 counts were more likely to measurement of survivors. This means that in reality, the improvements in CD4 after ART initiation may not be as good as indicated here. 5.6 Improvements in BMI A similar analysis was done for BMI and showed that the median BMI increased after ART initiation, and then plateaued (Figure 5.13). At ART initiation, the median BMI was around 20 kg/m2, and after 2 years this had increased to around 22.5 kg/m2 and remained stable out to 5 years. 40 Implementation of HIV/AIDS Care and Treatment Services in Tanzania Figure 5.13. Increase in BMI following ART initiation. BMI, kg/m 2 25 20 53094 15 0 27873 1 8599 572 2 3 4 5 Time since ART initiation, years Median 6 7 Upper and lower quartiles Values shown are the number contributing a measurement at each time-point Figure 5.14. Increase in BMI following ART initiation, by sex. 22 2 BMI, kg/m (median) 23 21 20 0 1 2 3 Time since ART initiation, years Male 4 5 Female Figure 5.14 shows that, at the time of ART initiation, women had higher BMI than men. Following ART initiation, both sexes had similar improvements in BMI, although over time women remained with higher average BMI compared to men. Implementation of HIV/AIDS Care and Treatment Services in Tanzania 41 These BMI analyses suffer from the same limitation as the analysis of improvements in CD4 counts, in that BMI can only be measured in those who survive and come to the CTC clinic regularly. Further, BMI could not be calculate for the majority of patients due to height often not being recorded. 5.7 Switches to second-line therapy A total of 243,676 adults were observed to initiate first-line ART and are included in this analysis. Table 5.5 shows the patient characteristics at first-line ART initiation, by whether they later switched, died, or were censored (using the first event observed; for example, if a patient switched to second-line and then later died, they are counted as switched only and not died). The distributions of characteristics at ART initiation were similar to those for the whole dataset (see Table 5.1). The proportions of patients observed to switch to second-line therapy were low across all subgroups of patients, at around 1%. 42 Implementation of HIV/AIDS Care and Treatment Services in Tanzania line therapy, death or censored. Outcome (row %) Category Total (column %) (N=243,676) Switched (N=16668) Dispensary 26,552 (11%) 152 (0.6%) Facility level Health centre 35,870 (15%) 92 (0.3%) Hospital 175,582 (73%) 1283 (0.7%) Other 4177 (2%) 141 (3%) Facility ownership Faith-based/ voluntary 56,258 (25%) 535 (1%) Government 150,834 (68%) 967 (0.6%) Private 16,235 (7%) 112 (0.7%) 2005 or before 10,659 (4%) 315 (3%) Year started ART 2006 22,303 (9%) 515 (2%) 2007 32,908 (14%) 372 (1%) 2008 43,641 (18%) 234 (0.5%) 2009 45,263 (19%) 121 (0.3%) 2010 44,493 (18%) 67 (0.2%) 2011 44,409 (18%) 44 (0.1%) Sex Male 84,622 (35%) 603 (0.7%) Female 159,050 (65%) 1065 (0.7%) Age, years 15-29 49,126 (20%) 301 (0.6%) 30-39 100,910 (41%) 724 (0.7%) 40-49 62,232 (26%) 445 (0.7%) ≥50 31,361 (13%) 197 (0.6%) 50,868 (23%) 362 (0.7%) Marital status Single Married/cohabiting 122,531 (55%) 766 (0.6%) Divorced/separated 23,920 (11%) 123 (0.5%) Widowed 25,119 (11%) 186 (0.7%) Working 198,448 (86%) 1267 (0.6%) Functional status Ambulatory 28,395 (12%) 177 (0.6%) Bed-ridden 4484 (2%) 13 (0.3%) Weight, kg <45 53,032 (22%) 303 (0.6%) 45-<55 94,074 (39%) 576 (0.6%) 55+ 93,801 (39%) 767 (0.8%) BMI, kg/m2 <18.5 19,826 (29%) 249 (1%) 18.5-<25 39,715 (58%) 461 (1%) 25+ 9485 (14%) 131 (1%) WHO stage 1 20,538 (9%) 157 (0.8%) 2 48,846 (22%) 306 (0.6%) 3 101,993 (45%) 624 (0.6%) 53,000 (24%) 275 (0.5%) <50 38,851 (23%) 431 (1%) CD4 cell count 50-199 87,056 (51%) 657 (0.8%) 200-349 36,952 (22%) 195 (0.5%) 350-499 4617 (32%) 27 (0.6%) 500+ 2379 (1%) 9 (0.4%) d4T, 3TC,NVP 126,752 (52%) 1151 (1%) ARV regimen ZDV, 3TC, NVP 39,124 (16%) 171 (0.4%) 64,256 (26%) 268 (0.4%) ZDV, 3TC, EFV 13,544 (6%) Other first line 78 (0.6%) Variable Died (N=25,892) 2203 (8%) 3303 (9%) 19,893 (11%) 296 (7%) 6400 (11%) 15,614 (10%) 1488 (9%) 2084 (20%) 4095 (18%) 4605 (14%) 5367 (12%) 4428 (10%) 3243 (7%) 2070 (5%) 11,402 (13%) 14,490 (9%) 4879 (10%) 10,236 (10%) 6714 (11%) 4056 (13%) 5778 (11%) 11,874 (10%) 2471 (10%) 2298 (9%) 16,857 (8%) 5564 (20%) 1405 (31%) 9112 (17%) 9729 (10%) 6482 (7%) 2126 (10%) 2010 (5%) 320 (3%) 809 (4%) 3070 (6%) 9757 (10%) 9590 (18%) 6548 (17%) 7730 (9%) 2339 (6%) 412 (6%) 230 (10%) 16,994 (13%) 2501 (6%) 4882 (8%) 1515 (11%) Censored (N=216,116) 24,197 (91%) 32,475 (91%) 154,406 (88%) 3740 (90%) 49,323 (88%) 134,253 (89%) 14,635 (90%) 8260 (78%) 17,693 (79%) 27,931 (85%) 38,040 (87%) 40,714 (90%) 41,183 (93%) 42,295 (95%) 72,6170 (86%) 143,495 (90%) 43,946 (89%) 89,50 (89%) 55,073 (89%) 27,108 (86%) 44,728 (88%) 109,891 (90%) 21,326 (89%) 22,635 (90%) 180,324 (91%) 22,654 (80%) 3066 (68%) 43,617 (82%) 83,769 (89%) 86,552 (92%) 18, 655 (89%) 39,363 (94%) 9535 (95%) 19,572 (95%) 45,470 (93%) 91,612 (90%) 43,135 (81%) 31,872 (82%) 78,669 (90%) 34,418 (93%) 4178 (90%) 2140 (90%) 108,607 (86%) 36,452 (93%) 59,106 (92%) 11,951 (88%) BMI=body mass index; d4T=stavudine; 3TC=lamivudine; NVP=nevirapine; ZDV=zidovudine; EFV=efavirenz. Note: totals by characteristic may not add up to the overall totals shown due to missing data, as follows: 1495 (0.6%) facility level; 20,349 (8%) facility ownership; 4 (<0.1%) sex; 47 (<0.1%) age; 21,238 (9%) marital status; 12,349 (5%) functional status; 2769 (1%) weight; 174,650 (72%) BMI; 19,299 (9%) WHO stage; and 73,821 (30%) CD4 count. Implementation of HIV/AIDS Care and Treatment Services in Tanzania 43 The second-line regimens to which patients switched are shown in Table 5.6, with the most common regimen being ABC, ddI and LPV/r (82%). The reasons for switch were not reported for 857 (51%) of individuals; reasons given are shown in Table 5.7. At switch to second line ART, CD4 results were available for 975 (58% of 1668) participants, with 234 (24%), 494 (51%), 156 (16%), 53 (5%) and 38 (4%) <50, 50-199, 200-349, 350-499 and ≥500 cells/mm3, respectively. Table 5.6. Second-line regimens used when switching from first-line therapy. N (total=1668) (%) ABC, ddI, LPV/r 1336 (82%) ABC, ddI, SQV/r 44 (3%) ABC, ddI, NFV 52 (3%) ABC, ddI, ATV/r 4 (0.2%) TDF, 3TC, LPV/r 20 (1%) Second-line regimen TDF, FTC, LPV/r 145 (9%) ABC, 3TC, LPV/r 17 (1%) Other, not specified 20 (1%) ABC=abacavir, ddI=didanosine, LPV=lopinavir, SQV=saquinavir, NFV=nelfinavir, ATV=atazanivir, TDF=tenofovir, 3TC=lamivudine, FTC=emtricitabine. Note: “/r” indicates the protease inhibitor was boosted with low-dose ritonavir. Table 5.7. Reasons for switch to second-line therapy. N (total=1683) 6 (0.4%) Peripheral neuropathy 15 (0.9%) Other adverse event 75 (5%) Clinical treatment failure 106 (6%) Immunological treatment failure 464 (28%) Poor adherence 5 (0.3%) Patient decision 2 (0.1%) Pregnancy 5 (0.3%) Stock out 35 (2%) Other, not specified 98 (6%) Missing 857 (51%) Reason Start tuberculosis treatment (%) Two patients had no follow-up after ART initiation (both initiated treatment on 31 December 2011) and are excluded from further analyses. 25,892 (11%) died while on first-line therapy. Figure 5.15 shows the cumulative probabilities of switch to second-line therapy and death over time, analysed as competing risks. For example, the probability of death was 9.6% by 1 year after first-line ART initiation and 17.4% by 6 years, while the probability of switching was 0.21% by 1 year and 2.61% by 6 years. Of note, the death probabilities in this analysis are different from those presented in the earlier mortality analysis, due to the incorporation of switching as a competing risk (in this analysis, we are looking at the first event of death or switching only, and we do not count deaths which occur after switching to second-line therapy). Also, the graph is only plotted to 6 years compared to 7 years in the mortality analysis, since we have less follow-up here (only to the first event of death or switching). Figure 5.15 shows that the probability of switching is very low compared to the risk of death, with the risk of death approximately 7 times the probability of switching by 6 years. 44 Implementation of HIV/AIDS Care and Treatment Services in Tanzania From the unadjusted models presented in Table 5.8, we see that there were differences in switching rates by facility level and ownership, with those who initiated ART in health centres less likely to switch, those in other facilities more likely to switch, and no difference for those in dispensaries, compared to hospitals. Those who initiated in faith-based/voluntary facilities were more likely to switch compared to those in government facilities. There was an overall trend towards less frequent switching with later year of ART initiation. Females were less likely to switch than men. There was a difference in the switching rates by weight at ART initiation, with those weighing <55kg less likely to switch than those ≥55kg. We found that those with lower CD4 counts at ART initiation, and lower (better) WHO stage, were more likely to switch. For example, those with CD4 count <50 or 50-199 cells/mm3 at ART initiation were 93% and 24%, respectively, times more likely to switch than those who initiated ART with CD4 counts of 200-349 cells/mm3, but with no difference for those who initiated with CD4 ≥350 cells/mm3. For current (time-updated) CD4 count, there was a clear trend with greater probability of switching at lower CD4 counts, as we would expect. There weight (p=0.35). The results of the full multivariable model were similar to the univariable results, except after adjustment for other variables there was no longer evidence of a difference in switching rates by sex (p=0.74) and there was evidence of more frequent switching among those who initiated on non-d4Tsub-HR 1.60, 95% CI 1.32-1.93). In addition, the direction of effect for CD4 at ART initiation reversed, with those with lowest CD4 at ART initiation less likely to switch; this is likely to be due to the adjustment of current (time-updated) CD4, where the association between lower current CD4 count and switch became even stronger. Subsequent work will investigate the need for treatment based on immunological failure, and switch to second line treatment following immunological failure. Implementation of HIV/AIDS Care and Treatment Services in Tanzania 45 Table 5.8. Factors associated with switch to second-line therapy. Variable Univariable models Sub-HR (95% CI) Characteristics at ART initiation Facility level Dispensary Health centre Hospital Other Facility ownership Faith-based/ voluntary Government Private Year of ART start 2005 or before 2006 2007 2008 2009 2010 2011 Sex Male Female Age, years 15-29 30-39 40-49 ≥50 Weight, kg <45 45-<55 55+ WHO stage I II III IV CD4, cells/mm3 <50 50-199 200-349 ≥350 First regimen d4T-based 1.11 0.53 1 3.66 (0.94,1.31) (0.43,0.66) (reference) (3.07,4.35) 1.36 1 0.98 (1.22,1.51) (reference) (0.81,1.19) 3.32 3.06 1.76 1 0.69 0.66 0.97 (2.79,3.96) (2.61,3.58) (1.49,2.07) (reference) (0.56,0.86) (0.50,0.87) (0.70,1.35) 1 0.90 (reference) (0.81,0.99) 0.94 1 0.97 0.92 (0.82,1.08) (reference) (0.86,1.09) (0.79,1.08) 0.77 0.78 1 (0.67,0.88) (0.70,0.87) (reference) 1.42 1.14 1 0.89 (1.19,1.70) (1.00,1.31) (reference) (0.77,1.03) 1.93 1.24 1 1.02 (1.63,2.29) (1.05,1.45) (reference) (0.72,1.46) 1 1.02 Time-dependent characteristics Current weight, kg <45 1.10 45-<55 0.98 55+ 1 Current CD4, cells/mm3 <50 4.62 50-199 3.65 200-349 1 ≥350 0.35 (reference) (0.91,1.14) (0.95,1.28) (0.88,1.10) (reference) (3.72,5.75) (3.03,4.39) (reference) (0.27,0.45) P Final multivariable model Sub-HR (95% CI) p <0.001 <0.001 <0.001 <0.001 0.03 0.68 <0.001 <0.001 <0.001 0.73 0.35 <0.001 0.94 0.55 1 2.83 (0.66,1.33) (0.39,0.78) (reference) (2.20,3.64) 1.95 1 1.03 (1.65,2.30) (reference) (0.65,1.63) 2.47 3.55 1.96 1 0.69 0.59 1.17 (1.83,3.33) (2.76,4.59) (1.51,2.55) (reference) (0.47,1.00) (0.37,0.93) (0.75,1.83) 1 0.97 (reference) (0.82,1.15) 1.01 0.88 1 1.03 (0.81,1.25) (reference) (0.72,1.06) (0.81,1.32) 0.72 0.74 1 (0.49,1.04) (0.59,0.94) (reference) 1.37 1.31 1 1.10 (1.07,1.74) (1.08,1.58) (reference) (0.88,1.36) 0.50 0.48 1 1.24 (0.37,0.66) (0.39,0.62) (reference) (0.76,1.74) 1 1.60 (reference) (1.32,1.93) 1.15 1.22 1 (0.76,1.74) (0.96,1.56) (reference) 6.17 4.04 1 0.26 (4.15,8.43) (3.24,5.04) (reference) (0.19,0.36) Note: p-values are from Wald tests. 46 Implementation of HIV/AIDS Care and Treatment Services in Tanzania <0.001 <0.001 <0.001 <0.001 0.74 0.50 0.04 0.01 <0.001 <0.001 0.26 <0.001 5.8 Discussion Baseline characteristics of adults starting ART The baseline characteristics of adults starting ART show that males consistently represent only around a third of people initiating ART. Females were initiating ART at a younger age compared to males, and were on average in a less-advanced stage of disease compared to males. This is likely to be related to women being enrolled into care at earlier stages of disease, as discussed in Chapter 4 Throughout the years, the greatest proportion of patients initiating ART were married or cohabiting. We observed a trend towards higher CD4 count, lower (better) WHO stage, and better functional status among those initiating ART in recent, compared to earlier, years. However, in 2011, over 60% of adults initiating ART were in WHO clinical stages 3 or 4 and similarly over 60% had CD4 counts of less than 200 cells/mm3. This indicates that even further improvements in identifying HIVpositive individuals for enrolment into care and initiation of ART are needed, particularly for males. The use of ARV regimens has changed over the years, with zidovudine-based combinations becoming more common compared to stavudine-based combinations. The ARV regimen of d4T, 3TC and NVP was prescribed in 15% of patients starting ART in 2011. We found that BMI was missing for around 70% of patients, primarily due to height often not being recorded. BMI is an important factor in survival, and improved quality of life through CD4 count restoration. BMI can only be assessed if height and weight are measured and recorded in the CTC2 card, and included in the submission to the national CTC3 database. Further, around 30% of patients had missing CD4 counts at the time of ART start. Incomplete information on CD4 count hinders the assessment of the improvement in health, and subsequent clinical treatment. While the proportion of patients with missing BMI dropped from 70% among those who initiated in each of 2009 and 2010 to 64% in 2011, the proportion with missing CD4 count increased from 26% in 2009 to 36% in 2010 and 41% in 2011. Actions need to be taken to improve reporting of all these clinical factors. Outcomes of adults starting ART Mortality analysis Results showed that, overall, 27,981 patients (11.0%) died out of the 255,143 patients initiating ART by 31 December 2011, representing a rise from the mortality rate of 9.5% reported in the previous report 10. Survival analysis showed that the probability of death was highest in the first 6 months on ART. However, the risk of dying within the first 6 months of ART initiation fell from 10.3% for those who started ART in 2005 or before, to 5.5% for those who started ART in 2011. The improvement in survival compared to previous report could be due to patients starting ART at a better clinical and immunological status compared to earlier years, as shown in our baseline results. The improvement in survival may also be due to better management of patients, or greater availability of ARVs in more primary health care facilities. Similar to the previous report, mortality in females was lower than in males, which could be due to the greater severity of disease in males at ART initiation. The results also showed that mortality was highest among patients weighing less than 45 kg, and with CD4 counts less than 50 cells/mm3, at ART initiation. This reinforces our recommendations that greater efforts need to be made to enrol 10 Implementation of HIV Care and Treatment Services in Tanzania (2011) Implementation of HIV/AIDS Care and Treatment Services in Tanzania 47 patients, and particularly men, into care and initiate ART before substantial damage to the immune system occurs in order to optimise their prognosis. Mortality over time (from 1+ year after ART initiation) was higher for men compared to women, and the same picture was observed for increases in CD4 counts over time by sex. This suggests that men may have poorer adherence to ART compared to women as well as being more likely to initiate ART at more immune-compromised states compared to women. We found that patients initiating ART at lower level facilities had poorer survival compared to those initiating in hospitals. Lastly, patients who initiated on non-stavudine containing ART regimen had better survival. Patients No Longer On Treatment While the cumulative number of patients no longer on treatment (NLOT) has gradually increased over the past 7 years from 2005, the drop-out rate within 6 months after ART initiation has been decreasing gradually since 2005. For example, the drop-out rate by 6 months was 13.6 per 100 person-years for those who started ART in 2005 or before compared to 10.7 per 100 person-years for those who started in 2011. However, the drop-out rate by one year was 23.0 per 100 person-years for those started ART in 2005 or before and increased to 26.4 per 100 person-years for those who started in 2010. These data show that the retention of patients on ART is a big challenge especially over long-term follow-up. Primarily, facilities need to ensure that counselling is effective to ensure retention and drug compliance. In addition, methods to ensure adequate tracking mechanisms are needed for patients who do not keep appointments. Information available on patients NLOT show that of these, male patients are more likely to be NLOT compared to females. This higher discontinuation among males could be due to the higher mobility patterns that men typically exhibit compared to women. Alternatively, it could perhaps be at ART initiation, with men typically at later stages of disease. It may be that there are differences in health seeking behaviour between men and women with respect to HIV care, or there may be other behavioural reasons. This report highlights the need to explore the problems affecting those who become NLOT. The reason for conducting this analysis of NLOT was to obtain a “worst-case” scenario for mortality, since there are concerns that often the reason for patients becoming LTFU is due to death. We have shown that the NLOT rates are around 40% higher than the mortality rates by 7 years after ART initiation. This means either that we are underestimating mortality rates of patients initiating on ART, or clinics are performing poorly in retaining patients on ART in care. It is likely that a combination of these factors contributes. In order to better understand this issue, there is a need to strengthen death reports at clinic level, while at the same time encouraging clinics to improve retention rates. It may be that many of those who have dropped out of care or treatment may have registered at another clinic, without the proper transfer forms. This would imply that the cumulative numbers than once. Such patients would also be counted as LTFU in our analysis, yet may be under care in 48 Implementation of HIV/AIDS Care and Treatment Services in Tanzania another clinic under a different identifier. Further, some of those which we have considered LTFU may have transferred care to a clinic not captured in the 348 clinics which are included in this analysis. Special analysis done using IQ tools on available electronic data (narrated in chapter 7) from Mwanza shows there is need to improve recording of transfers out of the clinic, and matching these to records of patients transferring in to another clinic. Since publication of the last report, the NACP has developed a tracking mechanism in the health facilities to improve retention and linkages with community based services. The tracking mechanism is foreseen to improve documentation of those not attending CTC for proper patient and programme monitoring. However, challenges still exist the tracking mechanism developed is used by personnel in the community who is yet to be recognised as an established cadre in the health system. The community health providers work as volunteers in many settings. There is need to assess their role and how they can be better used in community health services including retention and linkages of PLHIV clients. Further analysis using IQ tool or other means for remaining regions is crucial to understand regional differences in retention rates. CD4 counts and BMI improvement after ART initiation While we found substantial improvements in CD4 counts following ART initiation, the median increased by less than 100 cells/mm3 over the first year11. As expected, we observed larger increases in CD4 count among those who initiated with lower CD4 counts, although those who had initiated with the lowest CD4 counts remained with lower values up to 5 years after ART initiation. We observed a similar pattern comparing men and women, in that on average men initiated with lower CD4 counts and remained with lower CD4 counts up to 5 years later, compared to women. There appeared to be some divergence in the CD4 curves, perhaps related to poorer adherence among men. Following ART initiation, improvements in BMI were observed over the first year, and then stabilised. Comparing men and women, similar patterns were observed as for CD4 counts, in that men initiated ART at lower BMI and remained at lower BMI up to 5 years later. ARV use and switching to second-line treatment Overall, the numbers of patients switching to second-line therapy were low. By 6 years after ART initiation, the risk of death was approximately 7 times the probability of switching (approximately 17% and 3%, respectively). We found that reasons for changes to ARV regimens were in general poorly recorded, with reasons missing from the time of switch for 51% of the patients who were observed to switch. Further, CD4 count at the time of switch was missing for 42% of patients. We found differences in switching rates by health facility type and ownership, perhaps indicating that standardisation of care with respect to switching to second-line treatment is needed nationally. We found that switching to second-line ART was less common in later years of ART initiation, 11 Geng EH, Bwana MB, Muyindike W, Glidden DV, Bangsberg DR, Neilands TB, Bernheimer I, Musinguzi N, Yiannoutsos CT, Martin JN. Failure to initiate antiretroviral therapy, loss to follow-up and mortality among HIVinfected patients during the pre-ART period in Uganda. J Acquir Immune Defic Syndr. 2013 Jun 1;63(2):e64-71. doi: 10.1097/QAI.0b013e31828af5a6. Implementation of HIV/AIDS Care and Treatment Services in Tanzania 49 probably in part reflecting improved ARV regimes and management. As expected, and reflecting national guidelines for management of patients with immunological failure,those with lower current CD4 count were much more likely to switch. After adjustment for confounders, those weighing more, with lower (better) WHO stage at ART initiation, with higher CD4 at ART initiation and who initiated on non-d4T-based regimens were more likely to switch. This analysis does not assess the need for switch to second line, nor whether this need was met. It would be interesting and informative to use these data to explore when patients should be switching to second-line therapy in order to optimise their future prognosis, but this analysis would be hampered by the large missingness of longitudinal data, particularly CD4. Therefore, we recommend strengthening the recording of key data such as reasons for changes of ARV regimens and CD4 counts over time in order to facilitate such analyses in the future. 5.9 Limitations Despite the large amount of data available for this report, there are several limitations to our findings. There are limitations in the data that are reported to CTC by the patients, and that can be collected in the CTC database. It is unlikely that all deaths are reported to the clinic for several reasons, including lack of disclosure, and subsequent household dissolution. This indicates that the mortality on ART treatment in this report may be a lower estimate of the actual mortality experienced after ART initiation. The data are also unable to explain the high numbers who are no longer in treatment (NLOT) shown in this report. This seems to be consistent across the country, and deserves more research to find out whether such patients have died or moved to another clinic. The analysis of these data did not take into account the socio-economic status or other demographic characteristics of the patients enrolled in HIV care and treatment. Nothing at all is known about clinics that did not return data extract forms, so their role in ART outcomes could not be assessed. The attrition rate of patients from care and treatment programmes in Tanzania is high, highlighting the need for concerted efforts to improve tracing mechanisms to document the true outcomes for these patients, and to encourage them to return to care in the case of default. Specific studies are also necessary to shed light on the likely proportion of patients among those lost to follow up who have died as opposed to those who moved clinic without adequate documentation or who defaulted from care. The barriers to accessing CTC services are largely not known, although research has shown that transport, financial, and logistical barriers, including unfamiliarity with large hospital clinics, play a role 12,13,14. Whilst the roll-out of treatment is helping to overcome many of these obstacles, the provision of ART within smaller dispensaries and health centres brings its own challenges in terms of data collection and patient care, and needs to be closely monitored. Survival analysis, using hazard rates to calculate lifetable and to show Kaplan-Meier graphs of cumulative probability of death, give robust results and are far more flexible than traditional cohort 12 Roura M, Busza J, Wringe A, Mbata D, Urassa M, Zaba B. Barriers to sustaining antiretroviral treatment in Kisesa, Tanzania: a socio-ecological approach to understanding attrition from the ART program. AIDS Patient Care and STDs. PMID:19203295, Feb 2009. 13 Wringe A, Roura M, Urassa M, Busza J, Athanas V, Zaba B. Doubts, denial and divine intervention: Understanding delayed attendance and poor retention rates at a HIV treatment programme in rural Tanzania. AIDS Care, 2009; 21(5):632-7 14 Mshana G, Wamoyi J, Busza J, Zaba B, Urassa M. Barriers to accessing antiretroviral therapy in Kisesa, Tanzania: a qualitative study of early rural referrals to the national programme. AIDS Patient Care STDS. 2006 Sep;20(9):649-57 50 Implementation of HIV/AIDS Care and Treatment Services in Tanzania analysis. The analysis in this report only included baseline characteristics of patients at the time of ART initiation, as explanatory factors. This analysis did not consider the evolving effect on welfare of variables that change over time, such as CD4 counts or weight gain. The analysis of improvements in CD4 counts, and body mass index (BMI) in this analysis use the simple calculation of medians at each six monthly interval, which suffers from bias due to the drop out of those who died or dropped out of treatment. More advanced multi level statistical models will provide better estimates for these outcomes, but the analysis programs need to be developed. The wealth of data collected and collated in the CTC3 database deserves further analysis. Extra support to the epidemiology section of NACP, along with some resources, and personnel, dedicated to analysis would assist in the preparation of the data and performing more complex analyses. The further analysis would enhance our understanding of the outcomes of patients who enrol in care, and initiate treatment in Tanzania. This would also inform many other countries, in similar phase of the HIV epidemic, about the successes and problems experienced in the delivery of care and treatment in Tanzania. Further analysis of the data would also provide insights that could inform data collection, and reporting, in the future, ensuring that more relevant and useful data can be analysed in the future. This analysis of the clinic visit data is based on all clinic visits, which did not distinguish between regular scheduled visits, and extra visits for clinical treatment. Further analysis is needed to look at clinic visits in more detail and to find out what proportion of clinic visits were unscheduled. Some patients were given scheduled appointments for longer periods than 1 month, and this needs to be included in any analysis on clinic visits. 5.10 Recommendations • In order to improve quality of data, we recommend that Implementation partners should focus to improve quality of data in their respective regions through continued support and training to healthcare workers and data managers. It is also recommended that , Regional , District and implementing partners should ensure conducting regular data quality assessment preferably in every quarter • Currently some data are not adequately recorded eg height, CD4 results. We recommend that facilities/healthcare workers should be educated on the importance of recording all necessary information. This will minimise the number of missing values in some variables which are very important in the analysis. • There is a need to strengthen linkage between care and treatment clinics and home – based care services in order to minimise LTF , increase retention and improve documentation of vital status for NLOTs patients • A further development, improvement and adoption of the system for the recording of patients who transfer from one clinic to another ( IQ tool) is needed. This will enable more accurate analysis, and also better ways of monitoring the transfers, and outcome of patients. • More effort should be made to enable district and regional authorities to analyse and use these data to improve their services. The role of the implementing partners in working with the district and regional authorities, and helping to develop capacity to analyse and understand these data needs to be made clear. The levels of support for regional and district authorities should be agreed, so that analysis programs and best practice can be shared across the country. Implementation of HIV/AIDS Care and Treatment Services in Tanzania 51 • Implement strategies that promote male involvement in health-seeking and enhancing men’s access to health care services. There is need to increase the proportion of men who test for HIV and improve linkage to HIV care for those who test HIV-positive. This will assist men to have better ART treatment outcomes. • Consider developing strategies to enrol more people in care and treatment services using innovative testing approaches. • Conduct a second review on the HIV Care Patient Monitoring System to document gaps and functionalities. The data collection tools need to be revised and updated, and this process needs to be continually evaluated to ensure the data collected are necessary and useful. • Links with other health services should be developed. In addition sharing of data between different services should be facilitated. 52 Implementation of HIV/AIDS Care and Treatment Services in Tanzania 6. Results from Analysis of Children on ART 6.1 Baseline characteristics A total of 26,527 patients aged less than 15 years were observed to initiate ART in the 348 electronic clinics by 31st December 2011. Their characteristics are shown in Table 6.1 by year of ART initiation. Overall, 13,071 (49.3%) of the included patients were male and 13,456 (50.7%) were female. The male to female ratio was similar across all calendar years of ART initiation. The age distribution was similar across all years, with 9.2% aged <1 year, 18.6% aged 1-2 years, 21.6% aged 3-5 years and 50.8% aged 6 -14 years. About half (47.8%) of children initiating ART had weight between 10 and 20 kilograms and 7.5 % of the children’s weight was unrecorded. More than half (57.6%) of children initiated ART at WHO stages 2 and 3. With CD4, more than 50% of children had unrecorded CD4 baseline measure. Implementation of HIV/AIDS Care and Treatment Services in Tanzania 53 54 Implementation of HIV/AIDS Care and Treatment Services in Tanzania No baseline CD4 Count 200+ 50-199 Cell 1,395 Not recorded 23.0 53.1 1,345 14.6 9.3 55.0 5.2 21.6 10.8 7.3 2.2 12.8 51.5 23.3 10.3 62.7 22.4 10.5 4.4 50.4 49.6 9.6 (%) 582 371 236 131 4 <50 548 3 CD4 274 2 Stage 186 1 56 323 31+ Not recorded WHO 1,304 591 10-20 21-30 4-10 260 1,589 6-14 Median (IQR) Weight < 10 567 265 3-5 1-2 (years) 113 1,278 Female <1 1,256 Male Sex Age 2,534 n Total Category 2005 or before 1,857 801 514 332 1,528 272 950 504 250 76 281 1,778 797 3-10 572 2,032 760 508 204 1,783 1,721 3,504 n 53.0 22.9 14.7 9.5 43.6 7.8 27.1 14.4 7.1 2.2 8.0 50.7 22.8 16.3 57.9 21.7 14.6 5.8 50.9 49.1 13.2 (%) 2006 2,077 1,165 618 305 672 580 1,569 901 444 71 270 2,147 886 3-10 791 2264 959 714 228 2,109 2,056 4,165 n 49.9 24.0 14.8 7.3 16.1 13.9 37.7 21.6 10.6 1.7 6.5 51.6 21.3 19.0 54.3 23.0 17.2 5.5 50.6 49.4 15.7 (%) 2007 2,284 1,275 601 347 318 593 1,820 1,171 605 74 334 2,244 852 2-9 1,003 2,301 1,034 840 332 2,264 50.7 28.3 13.3 7.7 7.1 13.2 40.4 26.0 13.4 1.6 7.4 49.8 18.9 22.3 51.1 22.9 18.6 7.4 50.2 (%) 4,507 17.0 2,243 49.8 n 2008 2,417 1,198 582 440 373 576 1,665 1,218 805 85 322 2,201 840 2-9 1,189 2,169 1038 927 503 2,373 2,264 4,637 n 52.1 25.8 12.6 9.5 8.0 12.4 35.9 26.3 17.4 1.8 6.9 47.5 18.1 25.6 46.7 22.4 20.0 10.9 51.2 48.8 17.5 (%) 2009 Table 6.1 Baseline characteristics of 26,527 children started on ART by December 2011 2,339 888 383 343 166 544 1,461 1,091 691 86 243 1,747 644 2-9 1233 1,701 826 891 535 2,034 1,919 3,953 n 59.2 22.5 9.7 8.7 4.2 13.8 37.0 27.6 17.5 2.2 6.2 44.2 16.3 31.2 43.0 20.9 22.5 13.5 51.4 48.6 14.9 (%) 2010 2,037 592 320 278 103 503 1,298 808 515 60 211 1,249 570 2-9 1,137 1,418 532 749 528 1,615 1,612 3,227 n 63.1 18.4 9.9 8.6 3.2 15.6 40.2 25.0 16.0 1.9 6.5 38.7 17.7 35.2 43.9 16.5 23.2 16.4 50.1 49.9. 12.2 (%) 2011 14356 6501 3389 2281 4555 3199 9311 5967 3495 508 1984 1267 5180 6185 13474 5716 4894 2443 13456 13071 26,527 Total 54.1 24.5 12.8 8.6 17.2 12.1 35.1 22.5 13.2 1.9 7.5 47.8 19.5 23.3 50.8 21.6 18.6 9.2 50.7 49.3 100 (%) A total of 13,071 (49.3%) males and 13,456 (50.7%) females were observed to initiate ART by the end of December 2011. The male to female ratio (1:1) was similar across all calendar years over time there’s a trend towards younger patients being enrolled. E.g. 4.4% were less than 1 year old in 2005 or before, compared to 16.4% in 2011. The age pattern between the two sexes was similar at time of starting ART (Figure 6.1). Figure 6.1 Age of children starting ART by sex The weight structure between the two sexes was similar at the time of starting ART. However, majority of children starting ART weighed between 10-20 Kg (52.5% males and 50.8% females) as shown in Figure 6.2. Figure 6.2 Weight of children starting ART by sex Implementation of HIV/AIDS Care and Treatment Services in Tanzania 55 The weight structure by age show majority of under 2 year olds start with less than 10 kgs. For those aged between 3 and 14 years majority of children started ART weighing between 10-20 as shown in Figure 6.3. Figure 6.3 Weight of children starting ART by age. The patten of CD4 counts at time of initiation of ART was similar for both sexes, although about 54% of the children initiated ART without a reported baseline CD4 count. About 21% of those who were initated ART had CD4 count of less than 200 cells/μl. However, CD4 count is not a good measure of immunity in children; CD4 percentage for children younger than 6 years would be a better indicator of level of immunity, yet most clinics did not take CD4% for children below 6 years and for the few clinics who documented CD4% this was reported as no baseline CD4 count. Figure 6.4. CD4 counts of children starting ART by sex. 56 Implementation of HIV/AIDS Care and Treatment Services in Tanzania Figure 6.5. CD4 counts of children starting ART by age. Majority of children (69.3% males and 68.9% females- Fig 6.6a) were started on ART with WHO stage 3 or 4. For children below one year 60.3% started ART with WHO stage 3 and 4 and children aged 1-2 years 73.2% started ART with stage 3 and 4. This is late stage disease; the ART protocol urges all children below 2 years to start ART regardless of CD4 count or WHO stage. (Figure 6.6) Figure 6.6. Baseline WHO stage of children starting ART by age at initiation Implementation of HIV/AIDS Care and Treatment Services in Tanzania 57 Figure 6.6a. Baseline WHO stage of children starting ART by sex 6.2. Outcomes of children starting ART Overall, about 60% of children were observed to be still on follow up after ART initiation. About 30% were lost to follow up and remaining 10% (1% transferred out and 9% had died). About 40% of children initiated on ART were no longer on treatment.(Fig 6.7) Figure 6.7. The outcome of children who were observed to initiate ART at last visit by sex 58 Implementation of HIV/AIDS Care and Treatment Services in Tanzania 6.3. Mortality in children who started on ART Among the 26,527 patients who were observed to initiate ART by 31 December 2011, 25,890 had available follow-up data (after permitting 6 months until considered LTFU) and were included in the survival analyses. Table 6.2 shows the numbers of patients who started ART in each year, and the mortality within each calendar year cohort. Deaths and mortality rates are shown by the year in which the patient started ART, and by time since ART initiation. A total of 2,204 deaths were reported to the CTC clinic in children who had started ART from 2005 or before to December 2011. This represents 8.5% of the 25,890 children who started ART and had follow up data. Table 6.2. Mortality rates of children by year of ART initiation from 2005 or before to 2011. 2005 or before 2006 2007 2008 2009 2010 2011 2,847 3,482 4,102 4,771 4,615 4,402 235 263 306 285 303 194 58 31 20 344 55 43 26 387 61 36 12 415 52 19 356 27 - - Total deaths 29 25 18 178 330 194 Person-years 7,288 10,494 10,857 10,357 8,911 5,574 2,111 By 1 year By 2 years By 3 years 6.9 9.1 11.0 8.7 11.3 12.7 8.0 10.0 11.7 8.1 9.9 11.2 6.6 7.8 9.3 7.2 6.1 8.3 - By 4 years 12.5 13.7 12.8 11.7 - - - Individuals 1,671 started on ART Deaths Within 1 year 106 In 1-2 years In 2-3 years Mortality rate Figure 6.8 shows how cumulated mortality risk changes over time following initiation on ART. The of death of 0.05 at 6 years. Implementation of HIV/AIDS Care and Treatment Services in Tanzania 59 Figure 6.8. Cumulative probablity of death of children after ART initiation. Figure 6.9 shows that the cumulative probability of death over time is slightly higher among males than females. Log rank test result show no statistical significance (p- value 0.9288 χ2 =0.01) between female and male children. Figure 6.9. Cumulative probablity of death of children after ART initiation by sex. Figure 6.10 shows that lower CD4 counts at ART initiation were associated with higher mortality. Of note, those without a baseline CD4 count had similar mortality over the first 2 years after ART initiation as those who initiated with CD4 counts <50 cells/μl. Children without baseline CD4 test results may have been of same characteristics with those with CD4 counts <50 cells/μl. 60 Implementation of HIV/AIDS Care and Treatment Services in Tanzania Figure 6.10. Cumulative probablity of death of children after ART initiation by CD4 count. Figure 6.11 shows that children with WHO stage 3 or 4 disease at ART initiation had higher mortality compared to those who initiated ART with stage 1 or 2 disease. This scenario is also seen in adult ART patients. Figure 6.11. Cumulative probablity of death of children after ART initiation by WHO stage. Implementation of HIV/AIDS Care and Treatment Services in Tanzania 61 Hazard rates and ratios Table 6.3 compares the hazard rates between groups with different baseline characteristics. The baseline group for each factor was chosen as the most frequent reported characteristic. The results ratio = 0.84, 95% CI 0.77-0.92). For the different age groups after adjustment there was apparent less than 1 year against 1-2 years. The hazard ratios show the increased hazard in those initiating ART with lower weight, higher (worse) WHO stage, and with lower CD4 count. After adjusting associated with mortality and survival. Important to note the higher risk of death among those with no baseline CD4 count even after adjusting for other covariates. Table 6.3. Hazard ratios for baseline characteristics of paediatric patients who initiated ART from 2005 or before to 2011. Crude Characteristic Category Hazard ratio 95% CI Hazard ratio 95% CI Sex Male 0.82 0.76-0.89 0.84 0.77-0.92 Female 1 <1 3.25 2.82-3.75 1.05 0.85-1.30 1-2 2.07 1.82-2.36 0.88 0.74-1.06 3-5 1 6-14 1.09 0.97-1.23 1.27 1.10-1.47 <10 3.28 2.90-3.72 3.88 3.09-4.86 10-20 1.15 1.01-1.30 1.43 1.24-1.66 21-30 1 Age (years) Weight (kg) 31+ WHO stage CD4 cell count 62 Adjusted 1.24 1 1 1 0.89-1.73 1.24 0.86-1.79 1 1 1 2 1.12 0.89-1.42 1.42 1.12-1.80 3 1.71 1.39-2.11 2.08 1.68-2.57 4 3.98 3.21-4.92 3.87 3.12-4.81 <50 2.70 2.32-3.14 2.55 2.16-3.01 50-199 1.34 1.14-1.57 1.54 1.30-1.84 ≥200+ No baseline CD4 1 2.41 2.14-2.71 1 1.75 1.54-1.98 Implementation of HIV/AIDS Care and Treatment Services in Tanzania Fig 6.12.Smoothed hazard rates for children patients following initiation on ART Figure 6.12 shows the hazard rates over time from ART initiation. This confirms the high mortality (hazard) experienced in the first few months on ART, which is around 6% per person-year, or 0.6% per person-month. This high hazard comes down over the first 12 months, and at 12 months the hazard is around 3% per person-year. The hazard reduces even further to an underlying rate of 2% per person-year 6.4 Discussion Baseline characteristics of children starting ART The baseline characteristics of children starting ART show both male and female children of all ages have equal opportunities in starting ART. Throughout the years, there is no gender and age differences in children across years as what is observed in adults in HIV Care Provision. Looking at weight children initiate ART, we observed about half of children initiating ART had weight between 10 and 20 kilograms and only 7.5 % of the children’s weight was unrecorded. We also observe, more than half ( 57.6%) of children initiating ART at WHO stages 2 and 3 and more than 50% of children had unrecorded CD4 baseline measure. This indicates a need of intervention to have children tested for CD4 for eligibility assessment and initiated on ART promptly. Outcomes of children starting ART Mortality Analysis Among the 26,527 patients who were observed to initiate ART by 31 December 2011, 25,890 had available follow-up data (after permitting 6 months until considered LTFU). Survival analysis showed that the probability of death was highest in the first 6 months on ART. A total of 2,204 deaths were reported to the CTC clinic in children who had started ART from 2005 or before to December 2011. This represents 8.5% of the 25,890 children who started ART and had follow up data. This is the first analysis for children on ART using routinely collected data. Through the years the mortality rate in year one ranged from 6.1 to 8.7 deaths per 1000 person years and for year 4 it varied from 11.7 to 13.7 deaths per 1000 person years. More needs to be done to have better management of children patients and have availability of ARVs for children. Implementation of HIV/AIDS Care and Treatment Services in Tanzania 63 Looking at the hazard ratios, key findings are seen for children on ART that need attention. There is significant difference in risk of death between male and female children as well as by age groups higher in children aged 6-14 years, those initiating with lower weight, higher WHO stage ( 3 and 4) and with lower CD4 count. Even on adjusting for baseline factors, the mentioned factors still remained to be significantly associated with mortality and survival. Children with no baseline CD4 counts had higher risk of death even after adjusting for other variables. This suggest importance of having an initiative to have every child before ART initiation gets a CD4 test done. 6.5 Recommendations • There is need to improve quality of data, we recommend that Implementing entities ( Partners, RHMT and CHMT ) empower health care providers in data quality principles through training, supervision and mentoring. Regular data verification at health facilities with HIV Care services should use or establish quality improvement teams towards having quality data. • Key data for monitoring progress of children in ART is not adequately recorded e.g height, CD4 results, HIV adherence session. We recommend that facilities/healthcare workers should be educated on the importance of recording all necessary information. This will minimise the number of missing values in some variables which are very important in patient level analysis. • There is a need to strengthen linkage between care and treatment clinics, Reproductive Child Health services, and home – based care services in order to minimise LTF , increase retention and improve documentation of vital status for NLOTs patients as analysis reveal children still on ART was only 60% and 40% were no longer in ART services. We recommend special back to HIV Care services initiatives specifically focused on paediatrics. • Carry out further analysis for existing paediatric data for better programming and evidence decision making. E.g. Analyse paediatric patient level using anthropometric measures, ART progress by implementing partner and region, adherence to national guidelines, pre ART characteristics of HIV children. 64 Implementation of HIV/AIDS Care and Treatment Services in Tanzania 7. Tracing lost to follow up of HIV clients from multiple facilities in Mwanza region 7.1 Introduction Lost to follow up (LTFU) is commonly reported in care and treatment health facilities in Tanzania. It is estimated that 18% and 13% of adults and children, respectively, who started ART in 2007 were LTFU one year later (NACP HIV Care Report, 2010). The LTFU reported is due to unreported deaths, self-transfer to other care and treatment clinics without documentation, and disengagement from care. Patient level data from Mwanza 50 CTCs were used to assess unreported transfers, Lost to Follow up and clients picking drugs from multiple facilities. The IQ tool developed by Futures Group was used to assess clients who were LTFU or had unreported transfer. Data of clients who had been on ART and with unknown or non documented treatment outcomes were subjected to IQ tool to see if there are any duplicates using unique CTC Id number. The duplicates generated were then analysed using SPSS version 17 by assessing the socio-demographic characteristics from CTC2 database export files. The main objective of this detailed analysis is to describe magnitude and pattern of traced and non traced LTFU in multiple health facilities in Mwanza region, Tanzania. Specifically tracing the LTFU if receiving services in other facilities in the region, compare characteristics of traced and non traced LTFU and document movement pattern of traced clients 7.2 Results Information from Table 1 reveal there were 24,790 episodes of LTFU “in care” from the 47 health facilities in Mwanza region, of which 1601 (6.5%) were traced. Of the total 7,900 episodes of LTFU of those “on ART”, 620 (8%) were traced. Within districts, Kwimba had the highest proportion of LTFU episodes who were traced, with 10% and 15% of LTFU traced within that district among those “in care” and “on ART” respectively. Ilemela district had the lowest proportion of traced LTFU episodes within that district: 1% and 0.2% who were “in care” and “on ART”, respectively. Among those traced outside the districts, Ilemela district had the highest proportion of LTFU episodes traced: 12% and 14% among those “in care’ and “on ART” respectively. Implementation of HIV/AIDS Care and Treatment Services in Tanzania 65 Table 7.1: Total Lost to follow up and patients traced in care and treatment in Mwanza region as of 30th September 2012. District name Total Lost to Total LTFU follow-up (LTFU) traced within the district Care ART 2698 408 1697 1078 396 182 732 709 Total LTFU to be Total LTFU traced outside Traced outside the district the district Total LTFU traced Care % ART Traced % Care Traced ART Care % ART Traced % Care Traced % ART % Traced 137 13 144 175 87 110 38 72 2% 1% 3% 4% 10% 6% 3% 2% 1% 0% 4% 9% 15% 1% 3% 1% 2663 407 1625 984 335 180 713 703 190 192 127 91 39 71 31 84 3% 12% 3% 2% 5% 4% 2% 2% 3% 14% 4% 5% 7% 6% 3% 3% 5% 13% 6% 6% 14% 10% 5% 4% 105 60 144 148 83 13 41 26 4% 15% 8% 14% 21% 7% 6% 4% NYAMAGANA ILEMELA GEITA MAGU KWIMBA MISUNGWI UKEREWE SENGEREMA 6004 1636 4925 4159 914 1894 1441 3817 35 1 72 94 61 2 19 6 5867 1623 4781 3984 827 1784 1403 3745 TOTAL 24790 7900 776 3% 290 4% 24014 7610 70 59 72 54 22 11 22 20 327 205 271 266 126 181 69 156 825 3% 330 4% 1601 6% 620 8% From a total of 24,790 patients who were lost to follow up while in care (including those on ART) 1601 (6%) were traced while out of 7,900 patients who were lost to follow up while on treatment 620 (8%) were traced and were active within Mwanza region. Kwimba has highest 10% and 15% patients on care and on ART respectively for patients who were traced within Kwimba district while Ilemela districts have 12% and 14% on care and on ART respectively patients who were traced outside Ilemela Districts. Figure 7.1: Age distribution of traced patients who were considered lost to follow up Out of 1601 patients traced 1483(92.6%) were adult while 118 (7.4%) were children. 66 Implementation of HIV/AIDS Care and Treatment Services in Tanzania Table 7.2: Distribution of marital status by sex among patients traced M A R I TA L STATUS SEX TOTAL Female % Male % 8 53% 7 47% 15 1% 146 464 313 70% 61% 72% 64 298 123 30% 39% 28% 210 762 436 103 78% 29 22% 132 13% 48% 27% 8% 30 1064 65% 66% 16 537 35% 34% 46 3% Percentage COHABITING DIVORCED MARRIED SINGLE WIDOW MISSING DATA TOTAL 1601 Person’s Chi Square 26.15 and p-value is <0.01 Out of Total lost to follow up traced, married were more traced as compared to unmarried people while female who were married were traced more than married male. Table 7.3: Distribution of lost to follow up traced by facility type Movement by Level of Facility Health Center to Hospital Hospital to Hospital Hospital to Health Centre Hospital to Dispensary Dispensary to Health Centre Dispensary to Hospital Dispensary to Dispensary Health Centre To Health Centre Number Patients 74 235 276 45 21 11 5 72 Health Centre to Dispensary 37 of Percent 10% 30% 36% 6% 3% 1% 1% 9% 5% Of the 741 patients whose movement were traced, majority of patients moved from Hospital to Health centre 276(36%) and from Hospital to Hospital (30%). More analysis was done to clients who were found lost to follow up at hospital and were traced in other hospitals. The results shows majority 61% were found lost to follow up at district level hospital and moved to other district hospital mostly Faith based organization Hospital (DDH) Implementation of HIV/AIDS Care and Treatment Services in Tanzania 67 Figure 7.2: Movement of Patients from Hospital to Hospital Level DH = District Hospital , RH = Regional Hospital Thirty seven percent of patients lost to follow up at district hospital were traced at Bugando referral hospital. Figure 7.3: Missed duration by number of patients after being traced in HIV patients who were considered LTFU while on ART The median duration is 680.5 and IQR is 417-680. Majority of patients traced have been lost to follow up for more than one year. 68 Implementation of HIV/AIDS Care and Treatment Services in Tanzania 7.3 Discussion This is the first analysis of its kind in Tanzania to trace clients who are considered LTFU. Understanding LTFU patterns may help to develop strategies to improve retention or tracing mechanisms of HIV clients. The 6.5% of LTFU who were traced after analyzing only 48% of facilities indicates an even higher proportion would be traced if all health facilities in this region were analyzed, as well as tracing LTFU outside the region. In the future, we hope to expand the analysis to include all facilities and all regions in Tanzania to get the true picture of LTFU. The IQTools used to trace LTFU HIV clients compared only unique ID and some contact information such as village, ward, and division, and excluded other important criteria of comparison of patient identification such as name and treatment supporter. These are the data available to the CTC 2 database ‘export for analysis’ feature, and they represent the minimum tracing that can be done with anonymous CTC2 data. This is likely to have reduced the list of LTFU traced as there may be some clients using the same name but different unique IDs, or they register as new clients in the other clinics. Using more patient information could have increased the number of LTFU traced. However, we would not expect to be able to trace all of those LTFU as some might have died without being reported, moved out of the region which is not covered here, or moved to a clinic not captured in this analysis. There might also be some patients who presented to a clinic with different contact information and date of birth (DOB) and obtained a new patient ID for which the IQTools cannot capture. Different methods should be identified to capture those patients who are not using unique identification numbers provided by the Ministry of Health. The tracing can include the use of additional identifiers within available CTC 2 data such as names and telephone numbers. Few LTFU episodes were traced at Ilemela region within the district. Instead, the majority of LTFU episodes from Ilemela district were traced outside of the district. Ilemela is the most urban district and home to Sekou Toure Regional Hospital, Bugando Referral Hospital and other private hospitals that are better equipped than other districts’ hospitals. These might have resulted in the new registration or enrolment of clients at Ilemela health facilities from other districts when they were sick in order to receive better HIV care and management, who then returned to their home district after their status improved without notifying their mother facilities located at Ilemela district. The movement pattern among traced LTFU showed multiple trends. A high proportion of LTFU were traced in health centers. The movement from hospitals to health centers could be due to new health facilities opening within the district and patients sought to access services nearer their homes. For patients who moved from dispensary to health centers or from health centers to hospital, this may be due to patients’ preference for better services that may not be found in the lower health facilities where they were accessing services before. Among the LTFU episodes traced, those who were on ART were traced more often than those who were not on ART; this may be due to health seeking behavior for those who were on ART as they had already experienced health deterioration before. An HIV client on ART might have decided to seek more medical attention in other facilities if he/she finds the need for more care; it may also be due to service availability within the same facility. An HIV client may decide to stay in a facility where s/he gets all services like CD4 testing and other investigations. A linkage analysis between CTCs that provide all services and those not providing all services can be done to confirm the Implementation of HIV/AIDS Care and Treatment Services in Tanzania 69 LTFU patients’ movement across CTC facilities. During data analysis, some LTFU episodes were seen in multiple facilities within the same period. Although the number was very small, there is a need to look at patients attending and picking ARVs in multiple clinics. 7.4 Recommendations • Most of patients found to be missing in most Health facilities and hospitals were found in referral hospital or other facilities which seem to provide more services like CD4 test and advance clinical management services. There is a need therefore to upgrade Health Facilities to have more advanced services. • It was noted that some clients move within district level facilities. Patients Health Education on appropriate transfer should be done at facility level and Council Health management teams (CHMT) should lead the exercise of tracing Lost to follow up within the district. • More married people were LTFU and traced compared with unmarried; a need for a special study about disclosure among married is important to know the reason for most married patients to be lost to follow up. The Family information in the CTC2 forms was very important variables which used to show the HIV status at family level. The variable needs to be re-included to capture this important information. • Some patients were observed to attend multiple facilities and were found lost again within the facility they moved into. The social and behaviour characteristics of these people need to be done. • A self transfer without proper documentation and referrals should be discouraged at any facility and patients should be encouraged to request for transfer out note. • A need for proper feedback across facilities is crucial as it will reduce patients who have moved from one facility and are accessing treatment in the other facilities. • The analysis showed 8% of people who were lost to follow up were traced within Mwanza region. We recommend more analysis done in all regions to get the pattern of patients lost and overall patients lost at National level. 70 Implementation of HIV/AIDS Care and Treatment Services in Tanzania 8. REPORTING ON NATIONAL AND INTERNATIONAL INDICATORS to report on the following national and international indicators. The indicators reported here are Universal Access (UA), United Nations General Assembly Session on AIDS (UNGASS) and HIV Drug Resistance Early Warning Indicators (EWI) as well as national care and treatment indicators. Table 9.1 Indicators for Care and Treatment in Tanzania. INDICATOR Percentage of health facilities that offer ART (UA) EXPLANATION 2 Percentage of adults and children with advanced HIV infection receiving antiretroviral therapy(UNGASS) Estimates for adults children in 2011 3 Percentage of adults and children with HIV known to be on treatment 12 months after initiation antiretroviral therapy(UNGASS) Estimated for 3 yearly cohorts (2008, 2009 and 2010). 2008-74% 2009-74% 2010- 74% 2008-81% 2009-80% 2010-78% 4 Percentage of adults and children with HIV known to be on treatment 24 months after initiation antiretroviral therapy (UA) Estimated for three yearly cohorts (2007, 2008 and 2009) 2007-65% 2008-65% 2009- 63% 2007-72% 2008-72% 2009-70% 5 Percentage of adults and HIV known to be on months after initiation therapy(UA) Percentage of adults and HIV known to be on months after initiation therapy(UA) Percentage of adults and HIV known to be on months after initiation therapy (UA) children with treatment 36 antiretroviral Estimated for three yearly cohorts (2006, 2007, and 2008) 2006-59% 2007-58% 2008-57% 2006-66% 2007-65% 2008-65% children with treatment 48 antiretroviral Estimated for three yearly cohorts (2005, 2006, and 2007) 2005-57% 2006-54% 2007-52% 2005-68% 2006-62% 2007-61% children with treatment 60 antiretroviral Estimated for two yearly cohorts (2005, and 2006) 2005-52% 2006-49% N/A 8 Percentage of individuals starting ART who are prescribed a standard regimen (EWI) Estimated for 4 cohorts in 2008, 2009, 2010 and 2011 2008-99.7% 2009-96% 2010-92% 2011-97% 2008-99.7% 2009-98% 2010-96% 2011-97% 9 Percentage lost to follow-up during the 12 months after starting ART (EWI) Estimated for 3 yearly cohorts (2008, 2009 & 2010). 2008-16.0% 2009-17.5% 2010-19.1% 2008-11.1% 2009-13.8% 2010-14.8% Estimated for 3 yearly cohorts (2008, 2009 & 2010). 2008-99% 2009-94% 2010-93% 2011-98% 2008-99% 2009-96% 2010-97% 2011-98% 1 6 7 10 months later(EWI) From National database and RESULTS UPDATE: 1156 out of 6216 = 18.6% Adults Children 54.5% 31.3% Implementation of HIV/AIDS Care and Treatment Services in Tanzania 71 These indicators use the traditional cohort analyses, based on a life-table with the assessment of survival and loss to follow up at 12, 24 and 36 months. This is similar to the cohort analyses performed on monthly cohorts at clinic level in Tanzania, and does not allow for censoring during the period analysed. For indicators 3,4,5,9 and 10, the numbers in the analysis are the same as those used in main report Tables 5.4 and 6.2, but the analyses in Chapter 5 are based on the cumulated hazard function, which allows for censoring over the period of analysis and the analysis in chapter 6 does not show lost to follow up in children. This explains the slight differences between the estimates in the 2 tables. For the Early Warning Indicator (EWI) on the percentage of patients starting a standard ART regime line treatments at 12 months were used, with around 11% of adults, coded as not being on first line drug regimes after 12 months, although some of these may be on new regimes that had not been given codes on the CTC cards. For all EWI the denominators for the analysis are all patients starting ART in the calendar year. 72 Implementation of HIV/AIDS Care and Treatment Services in Tanzania Annex A Competing risks analysis This report has used survival analysis to show the risk of death among those who initiate ART. Survival analysis is the analysis of the time to an event. With this type of analysis, usually the event is called a failure (such as death), and lack of the event is survival (or continued existence). Thus survival and death in these analyses are mutually exclusive events, and if one happens it is at the expense of the other (if you die, then you do not survive, and if you survive, then you do not die). The other important aspect to this analysis is that everyone starts at time zero, and we need to define what time zero is. In the analyses in this report time zero is the date when the patient started ART – before they start ART the time is not used in the analysis, after starting ART, each day counts as time at risk for the survival analysis (see Annex to the previous Care and Treatment report for more details of survival analysis). In survival analyses people be lost-to-follow-up, if they do not die, and yet they stop coming to clinic. The technical term for this is censoring, and this was also explained in the Annex to the previous Care and Treatment report. With the traditional survival analysis, it is assumed that becoming lostto-follow-up is not associated with the event. In other words those who are censored are at the same risk of death as those who are still being followed up. This may not be true, but we do not have any further information to estimate mortality if the person does not come to clinic, and so this is the only assumption we can make. For some analyses in this report then survival and death are not mutually exclusive. Another event may happen that takes the person out of the risk set (or cohort) in the analysis. This happens for HIV positive people in care, who are not yet eligible for ART. In these analyses, we would like to know how long they wait before getting onto ART. The choice is between getting onto ART, and not getting onto ART. However there is a third event that could happen which upsets the simple survival analysis, and this is death. While waiting to get onto ART, some of those will die, and hence can never initiate ART. Unlike those who are lost-to-follow-up, who could at a later date initiate ART, those who die have no chance of initiating ART, and hence we would be wrong to treat them in the same way as those who are lost-to-follow-up. To compound this problem it is usually the case that those who die are much more likely to be in need of ART. Hence if we censor them at their date of death, treating them exactly the same way as those who were lost-to-follow-up, then we would be biasing the estimate of the time people wait to initiate ART. The solution we have used in this report is to treat death as a competing risk, and to simultaneously calculate the cumulative probability of both accessing ART (the event of interest) and death (the competing risk). We used Stata 12 for our analyses, and Stata has a command stcrreg to enable the sub-hazard functions of these two events to be calculated simultaneously. Stata used the method of Fine and Gray 1999 15 to calculate these sub-hazards. 15 Fine, J. P., and R. J. Gray. 1999. A proportional hazards model for the subdistribution of a competing risk. Journal of the American Statistical Association 94: 496-509. Implementation of HIV/AIDS Care and Treatment Services in Tanzania 73 Annex B PARTICIPANTS TO PREPARATORY AND ANALYSIS WORKSHOPS NAME INSTITUTION TITLE Aifello Sichalwe (Dr) KCMC & MoHSW MSc Epidemiology and Applied Biostatistics student Anath Rwebembera MOHSW- NACP Pediatric services program officer Annette Almeida (Dr) ICAP M&E Director Apaililia Kibona POLICE Medical Officer Aska Hasegawa NACP/JICA Project Coordinator Bernard Rabiel MOHSW-NACP M&E officer Bhavin Jani (Dr) CSSC M&E officer Bonita Kilama (Dr) MOHSW-NACP Acting Head, Epidemiology Unit Cyprian Makwaya (Dr) MUHAS Senior Lecturer and Biostatistician Deborah Kajoka (Dr) MOHSW - PMTCT National Coordinator Denna Michael (Dr) NIMR Mwanza MSc Epidemiology and Applied Biostatistics student Elaine Baker- Guni UCC Software Developer Elia Mbaga (Dr) MUHAS Epidemiologist and lecturer KCMC & NIMR, Mwanza MSc Epidemiology and Applied Emanuel Martin (Dr) Biostatistics student Essau Amenye AIDSRELIEF Strategic Information officer Ester Mungure MDH Data Manager ( MDH ) Filemoni Tenu KCMC & NIMR Tanga MSc Epidemiology and Applied Biostatistics student LSHTM & MITU/NIMR Research Fellow in Medical Statistics Fiona Ewings (Dr) MWANZA Florence Ndaturu MOHSW-NACP Program officer George Laizer MOHSW NACP ICT officer Germana Leyna (Dr) MUHAS Epidemiologist and lecturer Godwin Mnuo (Dr ) CDC TB/HIV officer Gretchen Antelman ICAP Research Director Helena Haule DELOITTE M&E officer Heriana Wilbert PASADA M&E officer Herilinda Temba (Dr) MUHAS Hilda Mmari (Dr) Prisons Head Quarters Medical Officer Humphrey Mkali MDH Data Manager ( MDH) Humphrey Shao (Dr ) ICAP Research Officer James Juma MOHSW-NACP IeDEA Coordinator Japhet Kamala UCC PEPFAR Support Jennifer Ward CDC Fellow Jenny Tiberio UCSF Program analyst Jerry Gibson CDC Senior Clinical Advisor Jim Todd LSHTM& NIMR- TAZAMA Reader in Applied Biostatistics MOHSW-NACP & KCMC MSc Epidemiology and Applied Joseph Nondi Biostatistics student 74 Implementation of HIV/AIDS Care and Treatment Services in Tanzania Julius Mboya Juma Mwinula ( Dr ) Karugira Rweyemamu (Dr) Levina Lema Mariam Ngaeje (Dr) Mohamed Mfaume (Dr) Moses Kehengu Mpoki Kajigili Munda Elias (Dr) Naimi Mbogo Neema Makyao Nobuhiro Kadoi Paul Mahunga Ramadhani Mwiru (Dr) Raphael Isingo Rita Mkama Sesil Latemba Stephanie Kovacs Theopista Komba Touma Ngwanakilala Tuhuma Tulli (Dr) Tumaini Francis Veryeh Sambu Victor Katemana Werner Maokola (Dr) Yasin Shangari Zebedayo Sekirasa DOD TPDF FELTP resident MOHSW - PMTCT AGPHAI CDC AIDSRELIEF PHARMACCESS MUHAS M&E CSSC MOHSW-NACP NACP/JICA DELOITTE MDH NIMR Mwanza DELOITTE PAI CDC EGPAF EGPAF AIDSRELIEF ICAP M&E MOHSW-NACP AIDSRELIEF MOHSW-NACP MOHSW NACP MOHSW-NACP M&E officer Medical Services Director MUHAS Strategic Information officer M&E officer Care and Treatment officer IT programmer M&E officer FELTP resident M&E officer KP coordinator Advisor M&E officer Nutrition Support Coordinator Statistician M&E officer M&E officer Fellow M&E officer M&E officer Strategic Information Advisor officer Data Manager ICT officer TB/HIV officer ICT officer Program officer – Care and Treatment Implementation of HIV/AIDS Care and Treatment Services in Tanzania 75