Multiple Comparisons for Partial Covariance Matrices of Two Treatment Groups in Clinical Trial

• Activity Number: 132
• Type: Contributed
• Date/Time: Monday, August 1, 2011 : 8:30 AM to 10:20 AM
• Sponsor: Biopharmaceutical Section
• Author(s): Yoshiomi Nakazuru*+ and Takashi Seo
• Companies: Tokyo University of Science and Tokyo University of Science
• Keywords: Multiple comparisons ; Covariance matrix ; Clinical trial

Abstract

Generally, the main focus of clinical trial is to demonstrate the efficacy of new treatment in terms of mean(s) of the primary endpoint(s). In some cases, however, variance(s) of the endpoints also may be the focus of interest. For example, a treatment with smaller variability, and thus more predictable efficacy, may be preferable, given two treatments with equal efficacy in terms of the means. In this presentation we consider multiple comparison procedure for partial covariance matrices of endpoints in clinical trials to demonstrate the superiority of a new treatment in terms of the variability. First, we review and discuss the tests for the equality of two covariance matrices based on the Union-Intersection test procedure. Second, we propose a multiple comparison procedure for partial covariance matrices that limiting the number of comparisons and compare its performance with the method discussed in first section, using Monte Carlo simulation with a normality assumption. The simulation results suggest that powers of the proposed procedure are generally higher than those of the previous method, while keeping type I error rates nearly within the nominal level.