Abstract:
|
Binary endpoints such as success rates or response rates have been widely used in many clinical trials. The difference in the success rates between two treatments, sometimes called the risk difference or treatment difference, is one frequently used measure of relative efficacy. The statistical methodology used to estimate the relative efficacy is becoming increasingly important, and may be complicated by the presence of categorical or continuous baseline covariates which are also associated with response. While different analysis approaches have been proposed in the literature, the selection among those approaches remains unclear to researchers in practice. In this study, through a comparison of biasness, efficiency and robustness of the estimates, a statistical simulation is conducted to investigate the performance of various methods on the estimation of risk difference in the presence of continuous baseline covariates.
|