A Bayesian Adaptive Randomization Design Using Placebo and Active Control

• Activity Number: 298
• Type: Contributed
• Date/Time: Tuesday, July 31, 2012 : 8:30 AM to 10:30 AM
• Sponsor: Biopharmaceutical Section
• Author(s): Yili Pritchett and Qi Tang*+ and Kristine Broglio and Weining Robieson and Scott Berry
• Companies: Abbott Laboratories and Abbott Laboratories and Berry Consultants and Abbott Laboratories and Berry Consultants
• Keywords: Bayesian ; adaptive randomization ; active control ; clinical trial ; placebo control ; stopping criteria

Abstract

Bayesian adaptive randomization designs are gaining more and more popularity in dose ranging clinical trials due to their efficiency. In our design, the efficacy of the experimental drug needs to be assessed by comparing with placebo as well as with an active control, which does not have consistent historical responses. We create a Bayesian adaptive randomization design comparing the experimental agent to an active agent and placebo. The goal of the trial is to characterize the dose-response relationship, identify the optimal dose, and stop the trial for success or futility if warranted. The performance of the active control has been taken into consideration when creating the stopping criteria. In this talk, we illustrate the adaptive design algorithm and the statistical properties of the design by comprehensive simulations using the software FACTS (Fixed and Adaptive Clinical Trial Simulator). Operating characteristics with various scenarios will be presented. Design properties are also examined under varying values of key sensitive factors including enrollment speed, standard deviation of the primary efficacy endpoint, longitudinal profile and effect of the active control.