FDA’s Sentinel Initiative

1. FDA’s Sentinel Initiative Dr. Doug Fridsma Director, The Office of Standards and Interoperability, The Acting Chief Scientist The Office of the National Coordinator for Health Information Technology (ONC) 07 May 2012 1

2. FDA Amendments Act of 2007Section 905: Active Postmarket Risk Identification and Analysis • Establish a postmarket risk identification and analysis system to link and analyze safety data from multiple sources, with the goals of including • at least 25,000,000 patients by July 1, 2010 • at least 100,000,000 patients by July 1, 2012 • Access a variety of sources, including • Federal health-related electronic data (such as data from the Medicare program and the health systems of the Department of Veterans Affairs) • Private sector health-related electronic data (such as pharmaceutical purchase data and health insurance claims data)

3. Sentinel Initiative • System will refine safety signals in near real-time. • This will require the following capabilities: • rapidly defining exposed cohorts; • establishing algorithms to capture health outcomes of interest; • using sophisticated modular programs capable of running assessments with minimal input from epidemiologists and clinicians and limited or no ad hoc programming; and • developing a framework to guide methodological approaches for safety surveillance assessments that include confounding adjustment.  • Collaboration among all stakeholders

4. How does FDA assess postmarket safety signals? Medical literature Premarket development program Spontaneous reports (e.g.,AERS) Microbiology data Foreign postmarket experience Preclinical data Postmarket clinical trials Clinical pharmacology data Engineering data Sentinel evaluation- another tool in the toolbox

5. Signal Generation Refinement Evaluation Stages of Postmarket Risk Assessment

6. Sentinel Initiative: A Collaborative Effort • Collaborating Institutions (Academic and Data Partners) • Private: Mini-Sentinel pilot • Public: Federal Partners Collaboration • Industry • Observational Medical Outcomes Partnership • All Stakeholders • Brookings Institution cooperative agreement on topics in active surveillance

7. Mini-Sentinel www.mini-sentinel.orgHarvard Pilgrim Health Care Institute Develop the scientific operations needed for an active medical product safety surveillance system Create a coordinating center with continuous access to automated healthcare data systems, which would have the following capabilities: Provide a "laboratory" for developing and assessing scientific methodologies that might later be used in a fully-operational Sentinel System. Offer the Agency the opportunity to assess safety issues in existing automated healthcare data system(s) and to learn more about some of the barriers and challenges, both internal and external. 7

8. Mini-Sentinel Principles/Policies • Public health practice, not research • Minimize transfer of protected health information and proprietary data • Data partners participate voluntarily • Maximize transparency • Tools, methods, protocols, computer programs • Findings

9. The Mini-Sentinel Distributed Database • Quality-checked data held by 17 partner organizations • Populations with well-defined person-time for which medically-attended events are known • 126 million individuals* • 345 million person-years of observation time (2000-2011) • 44 million individuals currently enrolled, accumulating new data • 27 million individuals have over 3 years of data *As of 12 December 2011. The potential for double-counting exists if individuals moved between data partner health plans.

10. The Mini-Sentinel Distributed Database *As of 12 December 2011. • 3 billion dispensings • Accumulating 37 million dispensings per month • 2.4 billion unique encounters • 40 million acute inpatient stays • Accumulating 41 million encounters per month including 400,000 hospitalizations • 13 million people with >1 laboratory test result

11. Mini-Sentinel Partner Organizations Institute for Health

12. Mini-Sentinel Common Data Model (MSCDM) v1.0 Describes populations with administrative and claims data Has well-defined person-time for which medically-attended events are known Data areas Enrollment Demographics Outpatient pharmacy dispensing Utilization (encounters, diagnoses, procedures) Mortality (death and cause of death)

13. Expanding the MSCDM • Clinical Information • Height • Weight • Systolic and Diastolic Blood Pressure • Tobacco Usage • Laboratory Information • Absolute neutrophil count (ANC) • Alanine aminotransferase (SGPT) • Alkaline phosphatase (ALP) • Bilirubin (total) • Creatinine • Fibrin d-dimer • Glucose • Glycosylated hemoglobin (HbA1c) • Hemoglobin • International Normalized Ratio (INR) • Lipase

14. Expanding the MSCDM • Laboratory Information • CK-MB fraction • Creatine Kinase (CK) • Influenza results • Platelets • Pregnancy test (qualitative) either urine or blood. • Troponin I • Troponin T

16. Distributed Querying Approach Three ways to query data: Pre-tabulated summary tables Reusable, modular SAS programs that run against person level Mini-Sentinel Distributed Database Custom SAS programs for in-depth analysis

17. Federal Partners Collaboration An active surveillance initiative via intra-agency agreements with CMS, VA, DoD Small distributed system Each Partner has unique data infrastructure No common data model being utilized FDA proposes medical product – AE pairs to assess Develop a shared protocol Assess interpretability of query findings resulting from a decentralized analytic approach and different patient populations 17

18. Observational Medical Outcomes Partnership (OMOP) Public-Private Research Partnership established to inform the appropriate use of observational healthcare databases for studying the effects of medical products: • Conducting methodological research to empirically evaluate the performance of alternative methods on their ability to identify true associations • Developing tools and capabilities for transforming, characterizing, and analyzing disparate data sources across the health care delivery spectrum • Establishing a shared resource so that the broader research community can collaboratively advance the science

19. Distributed Population Queries (Query Health) • Query Health: ONC-sponsored project establishing standards for distributed population queries • Aggregate population measures from clinical systems • Sends questions to the data sources; protects PHI • ONC & FDA partnering together on a pilot • What questions of interest to the FDA can be answered by a clinical system? • i2b2 Query Composer and an i2b2 data source will be integrated with Mini-Sentinel (PopMedNet) using Query Health Standards

20. Looking Forward • Long-term, complex initiative implemented in stages as scientific methodologies and data infrastructure evolves • Continue to: • Ensure maintenance of privacy and security within the distributed system • Address the concerns of stakeholders including patients and the public • Develop a strategy to allow the eventual Sentinel System to function as a national resource and complement other HHS initiatives using distributed systems for comparative effectiveness and quality assurance

21. Medical Product Safety Quality of Care Sponsors* Sponsors* Queries Queries Queries Queries Coordinating Center(s)† Coordinating Center(s)† Results Results Results Results Common Data Model Coordinating Center(s)† Coordinating Center(s)† Sponsors* Sponsors* Biomedical Research Public Health Surveillance Results Queries Coordinating Center(s)† Sponsors* Comparative Effectiveness Research Distributed Data and Analytic Partner Network • Payers • Public • Private • Providers • Hospitals • Physicians • Integrated Systems • Registries • Disease-specific • Product-specific *Sponsors initiate and pay for queries and may include government agencies, medical product manufacturers, data and analytic partners, and academic institutions. †Coordinating Centers are responsible for the following: operations policies and procedures, developing protocols, distributing queries, and receiving and aggregating results.

22. Backup

23. OMOP Common Framework Accommodating Disparate Observational Data Sources Standardized Terminologies Common Data Model Drugs Conditions Page 23

24. Common Framework applied to data from 194M patients across 11 data sources OMOP Extended Consortium OMOP Research Core Research Lab & Coordinating Center Centralized data Distributed Network Page 24

25. OMOP Profile System and process in place to empirically develop and test the performance of new methods Development and deployment of a database simulator for which signal strength can be directly controlled Extensive mapping of terminologies and vocabularies Growing portfolio of tested and deployed analysis methods within the OMOP Research Lab and other data environments Secure research computing laboratory and network of data partners with access to observational data representing over 100 million patients Robust governance model with broad stakeholder representation across two advisory boards and an executive board Established public-private partnership and diverse research community Open and transparent research culture