All Times EDT
Keywords: randomized trial, survival analysis
Auxiliary covariates are routinely collected in randomized clinical trials. Although it has been long recognized in statistical literature that incorporating covariates can improve the efficiency of inference and reduce chance imbalance, covariates adjustment has not been used as often as it should be in primary analysis of clinical trials in practice. Part of the reason is that people are concerned about potential model misspecification that may result in invalid inference. We discuss a framework that allows one to adjust covariates robustly in the sense that it is guaranteed to improve efficiency and make valid inference regardless of whether the specified models for covariates are correct or not, hence eliminating the concern over model misspecification. This robustness is guaranteed by the fact that treatments are randomized in a clinical trial. In the event the specified models are correct or nearly correct, one can achieve the optimal efficiency gain. We illustrate the framework for continuous, binary and survival outcomes and discuss various ways to implement it in practice.