|
Abstract:
|
There is an increasing interst for utilization of historical control data, including relevant early phase clinical trials and real-world data, to augment the control group in a randomized clinical trial (RCT). To control the confounding between the historical control and concurrent control, various methods, e.g., For example, propensity-score matching, proprensity-score-based prior, covariate-adjustment, have been proposed. Power priors, meta-analytic prior have been used to discount the historical information. The applicability of different methods depends on various practical issues, e.g., missing covariates, availability of historical data, outliers. In this study, we examine the impact of practical issues on the operating characteristics of the studies using augmented control. Simulation studies are used to compare the bias of treatment effect, sample size and type I errors of different methods. The objective of the study is to provide some empirical guidance of using the appropriate methods for control augmentation in RCT.
|
