Bayesian Applications in Medical Device Clinical Trials

• Activity Number: 6
• Type: Invited
• Date/Time: Sunday, August 3, 2003 : 2:00 PM to 3:50 PM
• Sponsor: Section on Bayesian Stat. Sciences
• Author(s): Andrew S. Mugglin*+ and H. Amy Xia*+
• Companies: Amgen, Inc. and Medtronic, Inc.
• Keywords: medical device clinical trials ; hierarchical modeling ; sequential monitoring ; multiple outcomes

Abstract

Medical device clinical trials are unique in the sense that a great deal of historical information on similar devices often exists, or sometimes a staged clinical strategy is taken in the new product development process. A Bayesian hierarchical approach is very suitable for such settings because by using good prior information, the size of a trial can be reasonably reduced. Other advantages of using Bayesian methods include sequential monitoring, prediction, and random effects modeling. Several case studies will be presented. One involves hierarchical modeling with borrowing of information from previous studies and faces some modeling challenges when the parameter of interest is on the edge of the parameter space. Another study features sequential monitoring of multiple safety and efficacy outcomes, while incorporating risk-benefit assessment into the stopping rules. In addition, practical challenges when planning a Bayesian trial in a regulatory setting will be addressed.