Covariate adjustment in subgroup analyses of randomized clinical trials: A propensity score approach

Background: Subgroup analyses are frequently conducted in randomized clinical trials to assess evidence of heterogeneous treatment effect across patient subpopulations. Although randomization balances covariates within subgroups in expectation, chance imbalance may be amplified in small subgroups and harm the precision of subgroup analyses. Covariate adjustment in overall analysis of RCT is often conducted, via either analysis of covariance or propensity score weighting, but covariate adjustment for subgroup analysis has been rarely discussed. In this article, we develop propensity score weighting for covariate adjustment to improve the precision and power of subgroup analyses in RCTs. Methods: We extend the propensity score weighting methodology to subgroup analyses by fitting a logistic regression propensity model with prespecified covariate-subgroup interactions. We show that overlap weighting exactly balances the covariates with interaction terms in subgroups. Extensive simulations were performed to compare the operating characteristics of unadjusted, propensity score weighting and ANCOVA estimator. We apply these methods to the HF-ACTION trial to evaluate the effect of exercise training on 6-minute walk test in prespecified subgroups. Results: Efficiency of the adjusted estimators is higher than that of the unadjusted estimator. The propensity score weighting estimator is as efficient as ANCOVA, and may be more efficient when subgroup sample size is small (N<125), or when ANCOVA model is misspecified. The weighting estimators with full-interaction propensity model consistently outperform traditional main effect propensity model. Conclusion: Propensity score weighting serves as an objective alternative to adjust covariate chance imbalance in subgroup analyses of RCTs. It is important to include the full set of covariate-subgroup interactions in the propensity score model.