Incorporating treatment-covariate interaction in designing a covariate adjusted response adaptive allocation for binary response trials

Abstract

Assuming binary treatment outcomes, a covariate adjusted response adaptive (CARA) design is developed for multi-arm clinical trials assuming the presence of treatment-covariate interaction. The proposed allocation skews the allocation dynamically according to the performance of the treatments and the covariate profile of any incoming subject. Relevant de- sign and precision based measures of the performance are further investigated empirically, which established the usefulness of the proposed allocation design over its competitors.