A Bayesian Constancy-Enforced Non-Inferiority Design in Medical Device Trials with a Binary Endpoint

Abstract

Non-inferiority trials are commonly seen in medical device trials. However, the violation of the constancy assumption could be a serious concern. Some approaches have been proposed to address the impact of violation of the constancy assumption on the validity of a non-inferiority trial such as prevention through good trial conduct and adjustment through analysis. In this paper, we propose a new approach via adding a companion constancy test to the non-inferiority test that consequently protects the validity of a non-inferiority trial under Bayesian framework. In addition, as some clinical data from historical studies are available for the active control prior to the start of the investigational study, we consider two different methods to borrow the available information. We use a simulation study to assess the performance of the proposed approach.