Model-Free Tests for Hypotheses of No Quantitative and No Qualitative Interactions in Survival Analysis

• Activity Number: 573
• Type: Contributed
• Date/Time: Wednesday, August 1, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Kallappa Koti*+
• Companies: FDA
• Keywords: Stratified randomization ; Kaplan-Meier method ; Simple effects ; Gail-Simon test ; Likelihood ratio test ; Predictive biomarker

Abstract

Often it is of interest to know whether treatment effects are heterogeneous over various subgroups of patients defined by the prognostic factors. We address this problem in a survival analysis setting. We extend the standard definition of interaction used in a standard 2×2 factorial experiment to survival analysis. We define the interaction in terms of median time. We use the Efron's bootstrap for estimating the standard error of median time. We propose model-free tests and discuss sample size determination. We discuss some possible applications of these tests in biomarker validation investigations. No official support or endorsement of this abstract by the Food and Drug Administration is intended or should be inferred. This presentation reflects the views of the author and should not be construed to represent FDA's views or policies.