Covariate adjustment in randomized controlled trials with dichotomous outcomes increases statistical power and reduces sample size requirements

摘要: Abstract Objective Randomized controlled trials (RCTs) with dichotomous outcomes may be analyzed or without adjustment for baseline characteristics (covariates). We studied type I error, power, and potential reduction in sample size several covariate strategies. Study Design Setting Logistic regression analysis was applied to simulated data sets ( n  = 360) different treatment effects, outcome incidences, prevalences. Treatment effects were estimated a single covariate. Strategies included always adjusting the (“prespecified”), only when predictive imbalanced. Results found that error generally at nominal level. The power highest prespecified adjustment. higher stronger (from 3 46%, 50% incidence prevalence) independent of effect. At lower incidences and/or prevalences, lower. Conclusion conclude characteristic lead potentially important increase analyses Adjusted should, hence, considered more often RCTs outcomes.