Trial Covariate Adjustment Guidance

FDA has published a guidance, Adjusting for Covariates in Randomized Clinical Trials for Drugs and Biological Products, with current agency recommendations on adjusting for covariates in the statistical analysis of randomized clinical trials in drug development programs. The document provides recommendations for using covariates in the analysis of randomized parallel group clinical trials that apply to both superiority and noninferiority trials.

The guidance’s main focus, FDA says, is on using prognostic baseline covariates to improve statistical efficiency for estimating and testing treatment effects. It does not address using covariates to control for confounding variables in non-randomized trials, using covariates in models to account for missing outcome data, using covariate adjustment for analyzing longitudinal repeated measures data, using Bayesian methods for covariate adjustment, or using machine learning methods for covariate adjustment.