All Times EDT
Keywords: COVID-19, Predictive Power, Type-I Error, Uncertainty
We present a Bayesian sequential design with the aim of providing a framework to rapidly establish a proof-of-concept (PoC) for vaccine efficacy of Bacillus Calmette-Guérin (BCG) in providing protection against COVID-19 infection under a constantly evolving incidence rate and in absence of prior data. The trial design is based on taking several interim looks and calculating the Bayesian predictive power with the current cohort at each look. At each interim look, a decision to either stop the trial for futility or stopping enrolment for efficacy or to continue to the next planned interim is made. If the decision is to stop enrolment for efficacy then the final analysis is carried out once the current cohort completes the pre-specified fixed follow-up time of 3 months. We present the simulation based operating characteristics of the design and also discuss possible approaches to the computation of the current cohort predictive power.