Inference on Subgroups and All-Comers Cognizant of Logical Relationships Among Efficacy Parameters

• Activity Number: 86
• Type: Contributed
• Date/Time: Sunday, August 9, 2015 : 4:00 PM to 5:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Szu-Yu Tang* and Jason Hsu
• Companies: Ventana Medical Systems, Inc. and The Ohio State University
• Keywords: Partition Testing ; Subgroup Analysis ; Logic Relationship

Abstract

In one aspect of personalized medicine development, the patient population is thought of as a mixture of two subgroups that might derive differential efficacy, given treatment versus control. An important decision to make is whether to target the entire patient population (so-called all-comers), or just a subgroup of the patients. There are logical relationships among efficacy parameters in the subgroups and all-comers. This presentation shows the Partition Principle in multiple testing can formulate null hypotheses that respect such logic, and discusses to what extent statistical inference should respect these logical relationships.