On Model Selections for the Power Analysis of Repeated Measurement Data in Clinical Studies

• Activity Number: 336
• Type: Contributed
• Date/Time: Tuesday, August 5, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Baiming Zou*+
• Companies: University of Florida
• Keywords: Repeated measurement ; Model selection ; Bootstrap ; Missing data ; Mixed effects model ; t-test

Abstract

Repeated measurement designs have been widely used in randomized trials for evaluating intervention efficacies. For some clinical trials, the last time point effect is of primary interest and tested by simple procedures, such as t-test. Though simple and convenient, this type of analysis throws away a large amount of information.

Alternatively, the mixed effects model is often used for the repeated measurement data. It models all available data jointly and allows explicit assessment of the overall treatment effects across time. In this paper, we propose an analytic strategy for the longitudinal clinical trial data where the mixed effects model is coupled with a model selection scheme. Our numerical results demonstrate that the proposed analytic procedure is more powerful than the simple t-test for the last time point effect testing. Simulations also reveal that the proposed method outperforms the mixed effects model for testing the overall treatment effects. In addition, the proposed framework is more robust, convenient and flexible in dealing with missing data. The proposed method is demonstrated by a clinical trial data on the cognitive effect of testosterone.