Improving Precision and Power in Randomized Trials Using Covariate Adjustment

• Activity Number: 171 - Biopharmaceutical Section P.M. Roundtable Discussion (Added Fee)
• Type: Roundtables
• Date/Time: Tuesday, August 10, 2021 : 12:00 PM to 1:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Michael Rosenblum* and Bingkai Wang and Alex Luedtke and David Benkeser and Ivan Diaz and Daniel Scharfstein and Jodi Seagal
• Companies: Johns Hopkins University and Johns Hopkins Bloomberg School of Public Health and University of Washington and Emory University and Weill Cornell Medical College and University of Utah and Johns Hopkins University
• Keywords: clinical trial; model robust; regulatory science

Abstract

In randomized clinical trials with baseline variables that are correlated with the outcome, there is potential to substantially improve precision and reduce the required sample size by appropriately adjusting for these variables in the statistical analysis (called covariate adjustment). Despite regulators such as the U.S. Food and Drug Administration and the European Medicines Agency recommending covariate adjustment, it remains highly underutilized leading to inefficient trials in many disease areas. This is especially true for trials with binary, ordinal, and time-to-event outcomes, which are quite common. We discuss how to implement covariate adjustment in the context of several disease areas including: COVID-19 treatment trials, Alzheimer's disease trials, and stroke treatment trials. We also discuss new estimators that leverage precision gains from both stratified randomization and covariate adjustment.