A0842
Title: A statistical tale of subgroup analysis for managerial decision making
Authors: Xuming He - University of Michigan (United States) [presenting]
Abstract: Clinical trials are often used for the assessment of therapeutic benefits. Subgroup analysis is routinely used in the clinical studies to understand heterogeneity of treatment effects in subgroups of patients. When used appropriately, subgroup analysis helps personalized treatments of patients and better designs of new studies. When used indiscriminately, subgroup analysis leads to costly false discoveries. We will start with a partial review of subgroup identification and confirmation methods, and examine the benefits and pitfalls of subgroup analysis associated with decision making in the pharmaceutical industry. We will discuss some important questions of how the risk for subgroup pursuit needs to be quantified to help managerial decision making. In this process, we identify several interesting statistical questions.
