How to Strengthen Analyses by Dropping Risky Assumptions: Three Examples

• Activity Number: 300
• Type: Topic Contributed
• Date/Time: Tuesday, August 11, 2015 : 8:30 AM to 10:20 AM
• Sponsor: Biopharmaceutical Section
• Author(s): Devan Mehrotra*
• Companies: Merck
• Keywords: covariance structure ; Cox model ; crossover trial ; pharmacogenomics ; stratification ; unequal variance

Abstract

Randomized clinical trials are the backbone of all clinical development programs. Statistical analyses and conclusions from such trials can be unwittingly distorted by unnecessary and/or untenable assumptions. In this presentation, I will draw attention to risky assumptions associated with traditional analyses in a variety of clinical trial settings. Simple ways to strengthen the analyses by eliminating some assumptions will be demonstrated using theory, simulations and case studies. Specific topics covered will include crossover trials with baseline responses, stratified time-to-event trials, and genomewide association studies with continuous phenotypes.