Bayesian predictive monitoring with bivariate binary outcomes in phase II clinical trials

Abstract Traditionally, phase II single-arm trials are based on a binary response variable that represents the efficacy of the experimental treatment. However, the introduction of an additional binary endpoint to assess whether the new therapy is also sufficiently safe for a further evaluation in larger phase III studies is often suggested. A Bayesian predictive strategy for interim monitoring in phase II trials focused on bivariate binary outcomes is proposed. At any interim analysis, the stopping rules are based on the evaluation of the predictive probability that the trial will show a conclusive result at the planned end of the study, given the observed data. The proposed procedure is applied using hypothetical scenarios that represent different situations which may occur at the interim stage. A real data application is also illustrated with the use of both non-informative and informative prior distributions. Finally, simulation studies to evaluate the operating characteristics of the design have been performed.