Statistical analysis of two arm randomized pre-post design with one post-treatment measurement

Abstract: Randomized pre-post designs, with outcomes measured at baseline and follow-ups, have been commonly used to compare the clinical effectiveness of two competing treatments. There are vast, but often conflicting, amount of information in current literature about the best analytic methods for pre-post design. It is challenging for applied researchers to make an informed choice. We discuss six methods commonly used in literature: one way analysis of variance (ANOVA), analysis of covariance main effect and interaction models on post-treatment measurement (ANCOVA I and II), ANOVA on change score between baseline and post-treatment measurements, repeated measures and constrained repeated measures models (cRM) on baseline and post-treatment measurements as joint outcomes. We review a number of study endpoints in pre-post designs and identify the difference in post-treatment measurement as the common treatment effect that all six methods target. We delineate the underlying differences and links between these competing methods in homogeneous and heterogeneous study population. We demonstrate that ANCOVA and cRM outperform other alternatives because their treatment effect estimators have the smallest variances. cRM has comparable performance to ANCOVA I main effect model in homogeneous scenario and to ANCOVA II interaction model in heterogeneous scenario. In spite of that, ANCOVA has several advantages over cRM, including treating baseline measurement as covariate because it is not an outcome by definition, the convenience of incorporating other baseline variables and handling complex heteroscedasticity patterns in a linear regression framework.