Building chronic disease registries from EMR and administrative data Elodie Portales-Casamar Clinical Assistant Professor, Dept of pediatrics, UBC Clinical Research Informatics Lead, CFRI
Outline Background Proof-of-concept Cross-Canada Opportunities Context/Challenges/Needs for Registries Solution: i2b2 & SHRINE Proof-of-concept Pediatric Diabetes Registry Two partner sites: Vancouver, BC & Edmonton, AB Cross-Canada Opportunities
Context Disease registries are a fundamental tool for surveillance, monitoring and outcome measurement along with many other research applications from epidemiology to biomarker discovery.
Challenges Quantity and quality of data from multiple sources No standard platforms Inconsistent vocabulary Provincial, national and international boundaries Limited integration with the clinical systems
Needs Data integration technology Interoperability and data sharing technology Consistent vocabularies Address privacy regulation across jurisdictions
Solution Informatics Integrating Biology & the Bedside (i2b2) data warehousing framework NIH-funded initiative scalable open-source software enables secondary usage and sharing of clinical data for discovery research SHRINE Data Sharing Network
Build on Successes e.g. ImproveCareNow US Pediatric IBD Network Modular pediatric chronic disease registry for Quality Improvement and Clinical Effectiveness research
Build on Successes e.g. Childhood Arthritis & Rheumatology Research Alliance (CARRA) Registry North American Network of pediatric rheumatology research centers Observational registry to assess therapeutics used to treat pediatric rheumatic diseases
Enable high performance collection of medical record data for querying and distribution Enterprise web client Create patient cohorts for further investigation Enable discovery within data on enterprise wide scale
Data remains in local Clinical Data Warehouse (e.g. i2b2 repository) Broadcast queries and aggregate responses No central database Hospital Autonomy: each site remains in control over all disclosures Patient privacy: no attempts to re-identify patients
Proof-of-Concept Time-frame: 1 year Focus on one disease: Diabetes Two partner sites: CFRI, BC: BC Children’s Hospital and UBC WCHRI, AB: Stollery Children’s Hospital and U of A Use mock-up data modeled from the real clinical data Deliverable: Assessment of technical implementation feasibility
Collaborators Informatics Clinical Partners Lawrence Richer, WCHRI Elodie Portales-Casamar, CFRI Clinical Elizabeth Rosolowsky, U of A Shazhan Amed, UBC Partners Anne Junker, MICYRN Wyeth Wasserman, CFRI
Model EMR EMR i2b2 i2b2 SHRINE Authorized User I2b2 User Query Interface Webserver EMR Clinical Data Warehouse EMR Clinical Data Warehouse i2b2 Research Data Warehouse i2b2 Research Data Warehouse SHRINE Federated Queries Data Aggregation Administrative Health Data Administrative Health Data BC Children’s Hospital Stollery Children’s Hospital
EMR Data Demographic Data Clinical Data* DOB Personal health number Contact information Geographic location Diagnosis Anthropometric data Vital signs (Blood pressure,…) Lab results (A1C, TSH,…) Medication (Insulin regimen,…) *for every patient visit
Administrative Data BC Ministry of Health (via PopData BC) or Alberta Health Services Physician Billing (Medical Benefits) Hospitalizations (DAD) Rx dispensations Demographic data Vital Statistics
PoC Objectives Demonstrate the utility and feasibility of a Canadian translational research informatics infrastructure that can: Address key questions, such as: epidemiology within and across provinces prevalence of co-morbidities adherence to recommended course of treatment Facilitate research study development through easier and consolidated data access
PoC Technical Considerations EMRs are poorly standardized Vocabularies/terminologies are different and need to be mapped to one another Disease definition varies across jurisdiction Different algorithms have been developed to define diabetes cases To be able to compare cases across province, the same case definition needs to be used
Broader Challenges Privacy/Legal framework Risk Aversion Unconnected data stewardship and governance Cross-jurisdiction Risk Aversion Multi-step data access processes Highly qualified personnel
Opportunities in BC Clinical & Systems Transformation Population Data BC SPOR Support Unit Provincial Data Platform
Fractal Model for BC Health Data Data is organized, managed and accessed locally To share, we need standardization and coordination across the pieces Coming together is a real opportunity but we can each assemble and manage the infrastructure we need http://www.rgbstock.com/cache1vE3Lv/users/x/xy/xymonau/300/omq8J1C.jpg
Fractal Model for BC Health Data How to make it work: Individual infrastructure Qualifying security, privacy, analytic characteristics Central location(s) for linking data Minimal essential data standards http://www.rgbstock.com/cache1vE3Lv/users/x/xy/xymonau/300/omq8J1C.jpg
Opportunities across Canada SPOR SUPPPORT Units Emerging Informatics Efforts MICYRN Clinical Research Informatics working group
Thank You.