Treatment by stratum interaction test for equivalence trials

• Activity Number: 23
• Type: Contributed
• Date/Time: Sunday, August 11, 2002 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section*
• Author(s): Hui Quan*+
• Affiliation(s): Merck Research Laboratories
• Keywords: Null and alternative hypotheses ; Equivalence boundary ; Confidence interval ; Noninferiority trials ; P-value ; Power

Abstract

Stratification is often used in clinical trials to enhance feasibility, efficiency, and generalizability of the study. Trials designed to assess the equivalence of two treatments are no exceptions. The issue of treatment by stratum interaction, especially the so-called quantitative and qualitative interaction, has been discussed in the literature. However, previous discussions mainly focused on trials that aim to show difference between treatment groups on a study endpoint. In this paper, our focus is on trials that aim to show equivalence of two treatments. We highlight some key differences between these two types of trials regarding the definition of the treatment by stratum interaction, and propose two approaches for testing treatment by stratum interaction in equivalence trials. We also present computation and simulation results to show the relationships among sample sizes in the individual strata, individual ignorable bounds for defining the hypotheses for the interaction test, number of strata, and powers of the two approaches. An example is given to illustrate the procedures. These procedures are also extended for noninferiority trials.