Choosing covariates in the analysis of clinical trials
Generate an AI Snapshot to get a quick, structured summary of this paper.
A concise AI-generated summary of the paper will appear here once you click Generate AI Snapshot.
TL;DR
With the aid of Canner's equation, using both formal analysis and simulation, the impact of covariate adjustment is further explored and it is shown that unless tight control over the analysis plans is established in advance, covariates can lead to seriously misleading inferences.
Abstract
Much of the literature on clinical trials emphasizes the importance of adjusting the results for any covariates (baseline variables) for which randomization fails to produce nearly exact balance, but the literature is very nearly devoid of recipes for assessing the consequences of such adjustments. Several years ago, Paul Canner presented an approximate expression for the effect of a covariate adjustment, and he considered its use in the selection of covariates. With the aid of Canner's equation, using both formal analysis and simulation, the impact of covariate adjustment is further explored. Unless tight control over the analysis plans is established in advance, covariate adjustment can lead to seriously misleading inferences. Illustrations from the clinical trials literature are provided.