Composite Kaplan-Meier and Commensurate Bayesian Models for Combining Historical and Progressively Accruing Survival Information

• Activity Number: 408
• Type: Topic Contributed
• Date/Time: Tuesday, August 6, 2013 : 2:00 PM to 3:50 PM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Ted Lystig*+ and Thomas Murray and Brian Hobbs and Bradley P. Carlin
• Companies: Medtronic, Inc. and University of Minnesota and The University of Texas M.D. Anderson Cancer Center and University of Minnesota
• Keywords: Bayesian hierarchical modeling ; Commensurate prior ; Flexible proportional hazards model ; Evidence synthesis ; Medical devices

Abstract

In this presentation we will consider both composite Kaplan-Meier and model based Bayesian methods using a piecewise exponential likelihood for synthesizing evidence from two sources of non-exchangeable time-to-event data. After a brief introduction of possible alternative approaches and an overview of the proposed methods, we will move on to applying the methods in real data. The setting will be a pair of post-market surveillance datasets capturing adverse events from persons on dialysis that underwent cardiac revascularization with a bare metal stent. We will focus on the manner in which the methods allow us to update fairly our current understanding of safety for the medical device as we progressively accumulate new information.