Bayesian Analysis for Noninferiority Trial

• Activity Number: 547
• Type: Contributed
• Date/Time: Wednesday, August 7, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Shu-Chih Su*+ and G. Frank Liu
• Companies: Merck Research Laboratories and Merck Res Labs
• Keywords: Bayesian ; non-inferiority

Abstract

In non-inferiority trials, the experimental treatment is compared to an active control instead of placebo. The design of those trials is often to demonstrate the experimental treatment is non-inferior to the control by certain pre-specified margin. For the studies with existing historical data, Bayesian approaches with the incorporation of informative priors might outperform the frequentist approaches as has been shown in other published work. In this work, we will explore the posterior distribution of the primary non-inferiority endpoint by varying prior distributions to accommodate prior information obtained from previous studies. We aim to demonstrate the potential efficiency gains of this Bayesian approach versus traditional frequentist approaches in the context of real vaccine trial data to demonstrate the usage of Bayesian in noninferiority trials.