Name: Peiling YANG

Firstname: Peiling

Title: Mathematical Statistician

Institution: Division of Biometrics I, CDER, FDA

Street: 1451 Rockville Pike, HFD-710, Room 2022

City: Rockville

Zip-Code: 20852

Country: USA

Phone: (301)827-1521

Fax: (301)594-6593

Email: yangp@cder.fda.gov

Authors: Peiling Yang, Gang Chen, and George H. Chi

Title: Controlling the False Positive Rate Due to Multiple Analyses

Abstract: In the design of a clinical trial, randomization is often stratified by a few important prognostic factors to avoid imbalance in distributions of the prognostic factors between treatment groups. In such a trial, either stratified or unstratified analysis is often acceptable as the primary analysis for the primary endpoint. However, if both analyses are considered as the primary analyses, then there will be a potential inflation of the false positive rate, depending upon how the two analyses will be used for the drug approval, as to winning on both or on either analysis.

In this presentation, simulation results for the false positive rate when using two correlated analyses for one primary endpoint will be discussed. In addition, a methodology for controlling the false positive rate with two primary analyses will be proposed and illustrated.