Adjustment for baseline covariates to increase efficiency in RCTs with binary endpoint: A comparison of bayesian and frequentist approaches