Use of Prior Information for Improving the Power of Multiple Testing Procedures

• Activity Number: 277
• Type: Invited
• Date/Time: Tuesday, August 2, 2011 : 8:30 AM to 10:20 AM
• Sponsor: Biopharmaceutical Section
• Author(s): Mohammad Huque*+ and Mohamed A. Alosh
• Companies: U.S. Food and Drug Administration and U.S. Food and Drug Administration
• Keywords: Multiplicity ; adaptive allocation ; prior information

Abstract

A clinical trial might involve more than one clinically relevant endpoint each of which is sufficient on its own to characterize the treatment benefit, yet due to power consideration one of these endpoints might be given a lower priority in a sequential testing for establishing an efficacy claim. The same situation arises when testing for a pre-specified clinically 'important' component of a composite endpoint following testing for the composite endpoint, or testing for pre-specified subgroup after testing for the total population. Various multiple testing procedures have been proposed in the literature with different allocation of the type I error rates between the two endpoints and possibly allowing the significant level for testing the second endpoint to adapt to the findings of the first endpoint. However, it is well-recognized in the literature that no single method outperforms other methods. In this presentation we investigate the utility of using additional information, which might be available at the design stage, in comparing the power performance of different multiplicity procedures for rejecting at least one null hypothesis as the criteria for a positive trial.