Bayesian Two-Stage Sequential Design for Phase II Clinical Trials

• Activity Number: 405
• Type: Contributed
• Date/Time: Tuesday, July 31, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Sumihiro Suzuki*+ and Jun Zhang and Shande Chen
• Companies: University of North Texas Health Science Center and University of North Texas Health Science Center and University of North Texas Health Science Center
• Keywords: two-stage design ; sequential analysis ; clinical trial ; Bayesian ; optimal design

Abstract

Phase II clinical trials are generally small scales studies involving 50 to 300 patients in order to evaluate both efficacy and safety aspects of a new drug. In such trials, it is advantageous to have interim analysis so that the trial may be terminated early when necessary. Our goal is to focus on the efficacy aspect to develop a Bayesian two-stage sequential sampling scheme for a hypothesis test about a binomial probability that minimizes the risk plus cost given a prior distribution, decision loss function, and cost function. Based on what we observe in the first stage, we terminate and reject, terminate and accept, or continue to the second stage. The approach involves first determining an optimal sample size, N, for the non-sequential (one-stage) case. We set this N as the allowable total sample size for the sequential (two-stage) case. We then attempt to minimize the risk plus cost in terms of N1 (sample size for stage one), N2 (sample size for stage two), and the continuation criteria for stage one, i.e. when do we proceed to stage two.