Considerations in Post Market Surveillance: The Medical Device Epidemiology Network (Mdepinet) Methodology Center
Laura Hatfield, Harvard Medical School  Lauren Kunz, Harvard School of Public Health  *Sharon-Lise T. Normand, Harvard Medical School and Harvard School of Public Health  Fred Resnic, Lahey Clinic Medical Center   Aartik Sarma, Harvard Medical School 

Keywords: bayesian, comparative effectiveness, variaiblity

Clinical outcomes associated with medical devices depend on interactions among (1) device hardware and software, (2) patient anatomy and physiology, (3) concurrent medical therapies, and (4) operator experience. The MDEpiNet Methodology Center is focused on developing methodology for combining information across data sources for inferring post market device safety and effectiveness accounting for these sources of variations. The comparative safety of artificial hips (data across registries and claims data), vascular closure devices (data across operators), and the effectiveness of cardiac resynchronization therapy (CRT) devices (data across published studies) illustrate our approaches. Using fully Bayesian hierarchical models, estimators are developed and the validity of assumptions and limitations of various approaches are discussed.