November 28 2022 Henry Ntanda, MS

PROMETRIKA team members make continuous learning a priority. As a statistician, it is important to keep up-to-date with the FDA’s current thinking on analytic approaches applied in clinical trials. In May 2021, the FDA published a draft guidance for industry, “Adjusting for Covariates in Randomized Clinical Trials for Drugs and Biological Products,” which provides recommendations for use of covariates in the analysis of randomized, parallel group clinical trials that are applicable to superiority trials and noninferiority trials.  I share some highlights and insights on this topic below.

According to the guidance, to improve the precision of treatment effect estimation and inference, sponsors should consider adjusting for prespecified prognostic baseline covariates (e.g., age, baseline severity, comorbidities) in the primary efficacy analysis and should propose methods of covariate adjustment (FDA Guidance on COVID-19 treatment and prevention trials). Also, as discussed in the ICH E9 Statistical Principles for Clinical Trials, “Pre-trial deliberations should identify those covariates and factors expected to have an important influence on the primary variable(s), and should consider how to account for these in the analysis to improve precision…”

Covariate adjustment in randomized controlled trials provides advantages. Random assignment of trial participants to treatments or control groups is a measure to generate comparable, well balanced groups between treatment arms in terms of both known and unknown covariates (Rosenberger WF et al 2002). However, randomization will not guarantee balance; in any given trial, there may be large imbalances in important prognostic covariates at baseline between treatment arms. Such imbalances can give an unfair advantage to one treatment arm over another if not accounted for in the analysis. Therefore, prespecifying important baseline covariates to be included in the analysis will help to ensure that random imbalances in these covariates will not impact treatment effect estimates (Egbewale BE et al 2014).

Using information measured on a subject before randomization to estimate and test treatment effects between randomized groups is referred to as covariate adjustment. Information measured on a subject can include disease characteristics, demographics factors, etc. In clinical trials, it is important that study populations reflect the variability of the target population, however, this may make it harder to detect a treatment effect. In situations like this, covariate adjustment allows for incorporation of prespecified prognostic factors in the statistical analysis and can result in narrower confidence intervals and greater statistical power to detect treatment effects. Further benefits of covariate adjustment include protection against random imbalances in important baseline covariates (Senn S et al 2007) and maintaining correct type I error rates when covariates are used in the randomization process (Paren M, et al 1998). Despite the benefit of covariate adjustment, unadjusted analyses still dominate in practice with covariate adjustment used in 24 to 34 % of clinical trials (Austin PC, et al, Saquib N et al 2013).

There are several approaches to covariate-adjusted analyses whose validity does not depend on a correctly specified model or that provide inference with greater robustness to misspecification than those used in common practice. For more information on these approaches, please visit the following links: and

In conclusion, during the clinical development process, the FDA strongly recommends that sponsors should consider use of covariate adjustment in a primary analysis and prespecify the details of the analysis, and prospectively specify the covariates and the mathematical form of the covariate adjusted estimator. These details should be included in the protocol or statistical analysis plan. If the number of covariates is large relative to the sample size or they are adaptively selected, sponsors should discuss the method of adjustment with the FDA during review of the protocol or statistical analysis plan as per the guidance provided at We at PROMETRIKA can guide our sponsors on how best to implement covariate adjustment and interact with the FDA should questions arise.

“In clinical trials, it is important that study populations reflect the variability of the target population, however, this may make it harder to detect a treatment effect.”

Share This Article