Evolution of a Clinical Research Informatics Group within a Service

Evolution of a Clinical Research Informatics Group within a Service

Evolution of a Clinical Research
Informatics Group within a
Service-oriented Clinical Trials
Data Management Organization
B. McCourt, D. Fasteson-Harris,
S. Chakraborty, C. Bova Hill
AMIA CRI Summit
San Francisco, March 2010
Topics
ƒ Context
ƒ Challenges
ƒ Organizational Design Solution
ƒ 2 Year Experience
ƒ Now and next steps
1
DCRI Context
What is DCRI?
ƒ DCRI is the largest academic
clinical research organization
(ARO) in the world
ƒ A global coordinating center for
multimulti-center clinical trials that
integrates the medical expertise of
Duke University Medical Center
with the operational capabilities of a
fullfull-service CRO
2
DCRI Facts
ƒ Founded in 1969 with the development of the Duke
Databank for Cardiovascular Diseases
ƒ 21 years of experience in coordinating multimulti-center trials
in over 20 therapeutic areas
ƒ 900+ staff and 120 clinical/statistical faculty
ƒ FullFull-Service Capabilities
ƒ 4,600 manuscripts in peerpeer-reviewed journals
ƒ More than 420 projects completed in 64 countries
enrolling more than 579,900 patients
DCRI – Trials Experience by Phase and Size
3
CDM Challenges
History (1997 – 2007)
ƒ History & Organizational Design
ƒ Causal Issues
• Capacity for change
• Industry trends
ƒ Environmental Issues
• Organizational changes
• Qualitative studies
4
Existing Org Structure (1997-2007)
Clinical Data Integration
ƒ Highly nested team units
ƒ Clinical Data Managers
ƒ Clinical Programming
ƒ Data Management Teams
ƒ Duke FollowFollow-up
ƒ Quality Control
ƒ Case Report Form Design
ƒ Medical Coding
10 Year Headcount Trend
5
10 Year Headcount Trend
Large paper trials locked
Adopted new
network
Industry Trends (2007)
ƒ Increasing number of projects with
increasing complexity and decreasing data
processing
ƒ Increased demand for EDC (vs. paper) and
increased demand for CDI consulting with
sponsor EDC tools & implementations
ƒ Increasing infrastructure initiatives
supporting DCRI, DTMI, NIH & Industry to
develop and adopt innovative operational
methods.
6
Industry Trends (cont)
ƒ Increased reliance on standards and
technology to meet increasing variety of
CDM requirements
ƒ Emerging industry recognition of increasing
CDM scope and trends
Environmental Studies
ƒ DCRI wide telephone survey
ƒ Departmental web survey
ƒ Horizontal focus groups
7
What did the studies tell us?
ƒ Opportunities for improvement:
• Inconsistent management decisions
• Career progression
• More effective communications
• Trust
• Collaboration
ƒ Mastery of administrative, technical, training
and project management was difficult to
achieve
Organizational Design Solution
8
Resulting Org Structure (Dec 2007)
Clinical Data Integration
Clinical Data Management
Clinical Research Informatics
Debra Fasteson-Harris
Brian McCourt
Associate Director
Associate Director
ƒ Clinical Data Managers
ƒ Data Management Teams
ƒ Research Informatics
ƒ Quality Control
ƒ Clinical Programming
ƒ Case Report Form Design
ƒ Duke FollowFollow-up
ƒ Medical Coding
Impact of matrixed CDM teams
ƒ Administrative Reporting Relationships
• 90 of 138 Employees switching managers
ƒ Project Teams Intact
• 88 projects total
• 69 have no changes to CDM Lead
• 12 the new lead has been involved already
and lead role being formalized
• 4 leads will change after resources available
• 2 new projects not yet assigned
• 1 project transition is being scheduled
9
DCRI Research Informatics
ƒ Support research data integration projects
with complex research requirements.
ƒ Evaluate and operationalize new ideas for
data management tools and methods (‘data
management pipeline’)
ƒ Develop and implement clinical data
standards
10 Year Headcount Trend
10
Emergence of CRI
ƒ Papers on issues
Clinical Research Informatics context
Duke Translational Medicine Institute (DTMI)
DTRI
DCRU
DCRI
11
DNRI DCCR
GHI
2 Year Experience
BioSignatures \ Biomarker Studies
class Ortel
TEST
ELISA_AV G_DATA
Design
«column»
*PK AVGDATID: NUMBER(10)
SPECMCD: VARCHAR2(40)
*FK PLATEID: NUMBER(6)
*FK TESTID: NUMBER(6)
RPTRESC: VARCHAR2(2048) = NULL
RPTRESN: FLOAT
CALSD: FLOAT
CALCV: FLOAT
RPTRFLG: VARCHAR2(10)
«FK»
+ FK_ELISA_AVG_DATA_ELISA_PLATE(NUMBER)
+ FK_ELISA_AVG_DATA_TEST(NUMBER)
100% eSource
ƒ
Metadata heavy
ƒ
Sample management
ƒ
New workflow
ELISA_RAW_ DATA
«column»
«column»
*PK TESTID: NUMBER(6)
*PK RAWDATID: NUMBER(10)
TSTCD: VARCHAR2(20)
*FK PLATEID: NUMBER(6)
TSTNAM: VARCHAR2(100)
SPECMCD: VARCHAR2(40)
+PK_ELISA_TEST
RPTU: VARCHAR2(20)
*FK TESTID: NUMBER(6)
(TESTID = TESTID)
LBTEST: VARCHAR2(100) 1
WELLNUM: VARCHAR2(10)
+FK_ELISA_RAW_DATA_TEST
LBTESTCD: VARCHAR2(20)
«FK»
RPTRESC: VARCHAR2(2048) = NULL
0..*
RAWVALU: FLOA T
«PK»
DILUTION: NUMBER(6) = NULL
+PK_ELISA_TEST
+ PK_ELISA_TEST(NUMBER)
CALCVALU: FLOAT
RPTRFLG: VARCHAR2(10)
1
+PK_ELISA_TEST
1
*
TRANSMID: NUMBER(6)
+FK_ELISA_RAW_DATA_ELISA_PLATE «FK»
+FK_ELISA_AVG_DATA_ELISA_PLATE
(TESTID = TESTID)
+ FK_ELISA_RAW_DATA_ELISA_PLATE(NUMBER)
0..*
+ FK_ELISA_RAW_DATA_TEST(NUMBER)
ELISA_ PLATE
1..*
(PLATEID = PLATEID)
«PK»
«FK»
(PLATEID = PLATEID)
«column»
+ PK_ELISA_RAW_DATA(NUMBER)
«FK»
+PK_ELISA_PLATE *PK PLATEID: NUMBER(6)
«FK»
+FK_ELISA_STD_DATA_TEST
*
TRANSMID: NUMBER(6)
1
PLATENAM: VARCHAR2(40)
+PK_ELISA_PLATE
ELISA_ STD_ DATA
TSTDTM: DATE
+sample 1
1
TSTTYP: VARCHAR2(22) = S
+commentCollection
«column»
TSTDESC: VARCHAR2(40)
SAMPLES_4488
0..* *PK STDDATID: NUMBER(6)
TSTSTAT:
VARCHAR2(13)
=
D
0..*
*FK TESTID: NUMBER(6)
«column»
RPTRTYP: VARCHAR2(12) = N
+elisaPlate
STDNAM: VARCHAR2(40)
SPECMCD: VARCHAR2(40)
BATTRNAM: VARCHAR2(40) = Elisa Quantific...
+commentCollection
FK PLATEID: NUMBER(6)
FK TRANSMID: NUMBER(6)
1
BATTRID: VARCHAR2(20) = AIM2
WELLNUM: VARCHAR2(10)
SPECMNUM: VARCHAR2(20)
+PK_ELISA_PLATE
SOFTWARE: VARCHAR2(4 0)
0..*
+sample
CONC: FLOAT
SPECSEQ: VARCHAR2(10)
INSTRMT: VARCHAR2(40)
(PLATEID = PLATEID)
CALCVALU: FLOAT
1
SUBJID: VARCHAR2(20)
INSTMSER: VARCHAR2(40)
«FK»
1
+FK_ELISA_STD_DATA_ELISA_PLATE
RPTRESN: FLOAT = NULL
«FK»
(TESTID = TESTID)
VISIT: VARCHAR2(40)
RPTRSTAT: VARCHAR2(11) = F
+ FK_COMMENT_4488_TRANSMISSION(NUMBER) +FK_COMMENT_4488_TRANSMISSION
CALCV: FLOAT
1..*
VISITNUM: VARCHAR2(20) = NULL
FILENAME: VARCHAR2(40)
«FK»
0..*
CALSD: FLOAT
+commentCollection
0..1
VISITTYP: VARCHAR2(11) = S
FILCRDTM: DATE
RPTRESC: VARCHAR2(2048) = NULL
LBDTM: VARCHAR2(25) = NULL
STUDYID: VARCHAR2(20) = 4488
RAWVALU: FLOA T
STUDNAM: VARCHAR2(200)
RAWAVG: FLOA T
STUDYID: VARCHAR2(20) = 4488
«PK»
RAWSD: FLOAT = NULL
LBSPEC: VARCHAR2(40) = PLA S
+ PK_ELISA_PLATE(NUMBER)
RAWCV: FLOAT = NULL
PROCNAM: VARCHAR2(100)
*
TRANSMID: NUMBER(6)
SHPPLBDT: DATE
SHPTO: VARCHAR2(50)
«FK»
+FK_SAMPLES_4488_TRANSMISSION
+FK_ELISA_PLATE_TRANSMISSION 0..*
+ FK_ELISA_STD_DATA_ELISA_PLATE(NUMBER)
(TRANSMID = TRANSMID)
«FK»
+testDat e 1
+ FK_ELISA_STD_DATA_TEST(NUMBER)
0..*
«FK»
+sample + FK_SAMPLES_4488_TRANSMISSION(NUMBER)
«PK»
«FK»
CLOTTING_ DATA
«unique»
[SPECMCD = SPECMCD]
+ PK_ELISA_STD_DATA(NUMBER)
(TRANSMID = TRANSMID) +PK_TRANSMISSION 1
1 + UQ_4488_SAMPLES_SPECMCD(VARCHAR2)
+clottingDataCollection
«column»
«FK»
0..*
CLDATID: NUMBER(6)
TRANSMISSION«flow»
«table» CDISC_UNKNOWNS
*FK TESTID: NUMBER(6)
+PK_TRANSMISSION
*FK TRANSMID: NUMBER(6)
«column»
CDISC_UNKNOWNS
SPECMCD: VARCHAR2(40)
1 *PK TRANSMID: NUMBER(6)
RPTRESC: VARCHAR2(2048)
*FK LABID: NUMBER(5)
«column»
RPTRESN: FLOAT
VERSION: VARCHAR2(7) = V1.0.01
SITEID: VARCHAR2(20)
+PK_TRANSMISSION
+FK_CLOTTING_DATA_TEST
RPTU: VARCHAR2(20)
TRMSRNUM: VARCHAR2(20) = ORTEL
INVID: VARCHAR(20)
TSTDTM: DATE
1
TRMSRNAM: VARCHAR2(40) = Duke Hemostasis...
INVNAM: VARCHAR2(80)
0..*
TSTTYP: VARCHAR2(22) = S
TRMTYP: VARCHAR2(11) = I
SCRNNUM: VARCHAR2(20)
+FK_CLOTTING_DATA_TRANSMISSION
TSTDESC: VARCHAR2(40)
FILENAME:
VARCHAR2(40)
SUBJSID:
VARCHAR2(20)
+PK_TRANSMISSION
(TRANSMID = TRANSMID)
TSTSTAT: VARCHAR2(13) = D
LACTDTM: VARCHAR2(25)
SUBJNIT: VARCHAR2(4)
0..*
«FK»
SOFTWARE: VARCHAR2(40) = ACL TOP 2.8.7
1
RECEXTYP: VARCHAR2(25) = BASE
SEX: VARCHAR2(1)
INSTRMT: VARCHAR2(40) = ACL TOP
LOADDTM: DATE
SEXCD: VARCHAR2(40)
INSTMSER: VARCHAR2(40) = 04090184
STUDYID: VARCHAR2(20) = 4488
BRTHDTM: DATE
BATTRNAM: VARCHAR2(40) = Functional Assay
FILCRDTM: DATE
RACE: VARCHAR2(20)
+FK_TRANSMISSION_LABORATORY
BATTRID: VARCHAR2(20) = AIM1
TRANCOMM: VARCHAR2(200)
RACECD: VARCHAR2(4 0)
RPTRTYP: VARCHAR2(12) = N
VISITMOD: VARCHAR2(20)
0..*
RPTRSTAT: VARCHAR2(11) = F
(LABID = LABID) «FK»
ACCSNNUM: VARCHAR2(20)
SPECNUM: VARCHAR2(10)
+ FK_TRANSMISSION_LABORATORY(NUMBER)
«FK»
+PK_LABORATORY
LABORATORY
PTMEL: VARCHAR2(9)
«FK»
«PK»
«flow»
PTMELTX: VARCHAR2(40)
+ FK_CLOTTING_DATA_TEST(NUMBER)
1
+ PK_TRANSMISSION(NUMBER)
«column»
COLENDTM: DATE
+ FK_CLOTTING_DATA_TRANSMISSION(NUMBER)
*PK LABID: NUMBER(5)
RCVDTM: VARCHAR2(25)
TRMSRNUM: VARCHAR2(20) = ORTEL
SPECICOM: VARCHAR2(2048)
TRMSRNAM: VARCHAR2(40) = Duke Hemostasis...
AGEATCOL: NUMBER(3)
PLBNAM: VARCHAR2(40) = Duke Hemostasis...
AGEU: VARCHAR2(6)
«table» CDISC_UNKNOWNS
PLBNUM: VARCHAR2(20) = ORTEL
FASTSTAT: VARCHAR2(7)
LBNAM: VARCHAR2(40) = Duke Hemostasis...
LBLOINC: VARCHAR2(10)
LBNUM: VARCHAR2(20) = ORTEL
LOINCCD: VARCHAR2(40)
RPTRESCD: VARCHAR2(40)
RPTRESNP: VARCHAR2(5)
«PK»
RPTNRLO: VARCHAR2(40)
+ PK_LABORATORY(NUMBER)
RPTNRHI: VARCHAR2(40)
RPTUCD: VARCHAR2(40)
CNVRESC: VARCHAR2(204 8)
CNVRESCD: VARCHAR2(4 0)
CNVRESN: VARCHAR2(20)
CNVRESNP: VARCHAR2(5)
CNVU: VARCHAR2(20)
CNVUCD: VARCHAR2(4 0)
SIRESC: VARCHAR2(204 8)
SIRESCD: VARCHAR2(40)
SIRESN: FLOAT
SIRESNP: VARCHAR2(5)
SINRLO: VARCHAR2(4 0)
SINRHI: VARCHAR2(40)
SIU: VARCHAR2(20)
SIUCD: VARCHAR2(4 0)
ALRTFL: VARCHAR(14)
DELTFL: VARCHAR2(2)
TOXGR: VARCHAR2(1)
TOXGRCD: VARCHAR2(40)
EXCLFL: VARCHAR2(14)
BLNDFL: VARCHAR2(24)
RPTDTM: VARCHAR2(25)
COMMENT_4488
«column»
SPECMCD: VARCHAR2(40)
FK TRANSMID: NUMBER(6)
SPECCOM: VARCHAR2(2048)
SPECCND: VARCHAR2(2048)
PLATENAM: VARCHAR2(40)
TSTCOM: VARCHAR2(2048)
TECHNAME: VARCHAR2(40)
COMMDATE: DATE
DILUTION: NUMBER(6)
PLBNUM: VARCHAR2(20) = ORTEL
STUDYID: VARCHAR2(20) = 4488
BATTRID: VARCHAR2(20)
ƒ
+PK_ELISA_TEST
(TESTID = TESTID)
+FK_ELISA_AVG_DATA_TEST1
«FK»
0..*
12
«PK»
+ PK_ELISA_AVG_DATA(NUMBER)
+elisaDataCollection
0..*
T1 Challenges in CDM Services Context
ƒ Evolving science causes evolving data
requirements
• Scope, pricing, workflow, bio and information
science skills
ƒ Discovery based work now within scope of
FDA regs and pharma traditions
• Mismatch of burden/benefit and common
understanding of regulations
ƒ Data Management systems are immature or
don’t exist.
• Bioinformatics tools lack data management
workflow
• IT -> Research IT\Computer Science
Decision support methodology:
implementation on a clinical trial
Site contact
information
Web-based
Study Database
Enrollment and
randomization
information
Lab results
Treatment arm
(as needed)
Decision
Support
Tools
(21CFR Part 11
compliant)
Computer-assisted treatment
recommendations
Web-based reports
Study participants
(patients)
Site
investigators
MD reviewer
evaluates
and enters a
Treatment
Decision
Clinical Operations
reviews and
manages site
communications
Wilgus,
Wilgus, et al. Poster presented at AMIA Annual Meeting, 2009.
13
T2 Challenges in CDM Services Context
ƒ New risk profile of research tasks within
patient care process.
ƒ EHR’s -> Disease cohorts -> Trials -> EHR
• Complex governance
• Consent; research vs quality improvement;
future use; Identification.
• Data collection design
ƒ Interoperability
• CDISC : HL7; WHO Drug : RxNorm;
MedDRA : … ; Metadata!
14
Duke CDR & Knowledge Repository
Study
Meta Data
Consent
Geospatial/
Environmt.
Sample
Data
CRF/
Clinical
Data
Omics
Data
CDR &
Knowledge
Repository
Operational
Data
Imaging
Data
Electronic
Health
Records
Decision
support
Discovery
External
sources
Cohort
selection
Key Issues
ƒ Unanticipated acceleration of trends
• CTSA, ARRA, Duke Center for Health
Informatics, DCRI BioSignatures Program
ƒ What is informatics?
• Depends on who you ask
• Functional bounds of interdisciplinary domain
ƒ Field is still immature
• Too few examples
• Talent gap
ƒ Cost constraints & allocation
• Project vs infrastructure
15
Where Next?
ƒ Practical need for CRI will meet vision
(bottom up meets top down)
ƒ Grow as research partner, beyond service
provider
• CRI Faculty
ƒ Data Infrastructure
• Adopt HL7 Development Framework as
methodology
• Heavily use & contribute to standards
• Leadership, data governance
Conclusions
ƒ The challenges discussed at this meeting
have corresponding business challenges
ƒ The research trends driving the growth of
CRI exist
• Not only as CTSA phenomena
• Can be expected to impact services market
ƒ CRI is not yet well defined or established
ƒ Need to demonstrate value
16
Acknowledgements
ƒ Swati Chakraborty,
Chakraborty, M.Eng.
M.Eng.
ƒ Connor Blakeney
ƒ Carol Hill, PhD
ƒ Robert Harrington, MD
ƒ Cindy Kluchar,
Kluchar, MS
ƒ Meredith Nahm
ƒ James Topping, MS
ƒ James Tcheng,
Tcheng, MD
ƒ Becky Wilgus,
Wilgus, RN, MSN
Evolution of a Clinical Research
Informatics Group within a
Service-oriented Clinical Trials
Data Management Organization
B. McCourt, D. Fasteson-Harris,
S. Chakraborty, C. Bova Hill
AMIA CRI Summit
San Francisco, March 2010
17