Performance of Some Proof of Concept Decision Rules for Multiple Response Variables in a Multi-Dose Clinical Trial

• Activity Number: 473
• Type: Contributed
• Date/Time: Wednesday, August 3, 2011 : 8:30 AM to 10:20 AM
• Sponsor: Biopharmaceutical Section
• Author(s): Yiyun Tang*+
• Companies: Pfizer Inc.
• Keywords: Multiple Response Variables ; clinical trial design ; type I error ; Multiple Dose ; Proof of concept

Abstract

The proof of concept clinical trial often requires evaluation of multiple response variables for two or more doses of a test medicine. An example is a proof of concept study by a pharmaceutical company for the treatment of a psychiatric disorder. This paper describes statistical methods for a study design with multiple doses and balanced sample size, and it proposes reasonable statistical properties decision rules. Simulation studies are conducted to evaluate alternative decision rules with different degrees of flexibility. Discussion addresses the influence of correlations among the multiple response variables on the type I error, and the influence of the pattern of treatment differences on statistical power. A rule which requires a relatively strong result for at least one response variable and at least moderately favorable trends for all but one of the others is found to provide good statistical power with adequate control of type I error.