Road to a New Study from Post-hoc Findings: a Bayesian Approach
*Chul Ahn, U.S. Food and Drug Administration  *Yunling Xu, CBER/FDA 

Keywords: Post-hoc findings, New study, Bayesian approach

In this presentation, unplanned subgroup analysis will be discussed within the context of using the result to initiate a new confirmatory study. A Bayesian approach is then used to integrate the subgroup data with the new confirmatory study data to increase the overall power of the study. Some statistical issues related to such study design and analysis of the study data, including simulation of operating characteristics and assessment of exchangeability between the subgroup and the new study, will be discussed from a regulatory perspective.