Multiple Testing with Complex Hierarchical Structures in Clinical Trials

• Activity Number: 316
• Type: Invited
• Date/Time: Tuesday, August 2, 2011 : 10:30 AM to 12:20 PM
• Sponsor: General Methodology
• Author(s): Alex Dmitrienko*+ and Ajit C. Tamhane
• Companies: Eli Lilly and Company and Northwestern University
• Keywords: Multiple comparisons ; Familywise error rate ; Gatekeeping procedures ; Clinical trials

Abstract

This talk discusses approaches to addressing multiplicity issues in clinical trials with multiple objectives. The objectives define multiple testing problems with complex hierarchical structures, eg, problems with null hypotheses grouped into multiple families with general logical relationships. A new method has been developed for defining multiple testing procedures for problems with multiple families of null hypotheses (termed gatekeeping procedures). The method is based on constructing mixtures of procedures used in individual families, including p-value-based and parametric procedures, and is termed the mixture-based method. Resulting gatekeeping procedures account for arbitrary logical relationships and control the familywise error rate in the strong sense. This general method is illustrated using clinical trials with multiple objectives, including multiple dose-placebo comparisons/patient populations and multiple dose-placebo comparisons/noninferiority-superiority tests.