Mind the gap: Bayesian equipoise calibration of clinical trial designs

A key objective of randomized clinical trial design is to ensure strong control of the conditional error rates associated with the primary analysis outcome, but no link between trial design and the probabilities of the design hypotheses is currently established. This paper contributes to bridging this gap by calibrating the operational characteristics of the trial design with respect to the pre-specified level of equipoise imbalance , which is a reduction in pre-study uncertainty about whether the null or the alternative hypothesis are more likely true, provided by the primary trial outcome. First, equipoise imbalance is quantified as the percentile of the post-study odds of the design hypotheses on a prior distribution reflecting pre-study equipoise within the population of medical experts. Under this prior, common late phase design outcomes are shown to provide at least 90% evidence of equipoise imbalance. Designs carrying 95% power at 5% false positive rate are also shown to provide strong equipoise imbalance against the alternative hypothesis when the trial outcome fails to reject the null, providing a robust statistical basis to informing further clinical development decisions. Finally, equipoise calibration is applied to design of sequential clinical development plans in oncology. Commonly used power and false positive error rates are shown to provide strong equipoise imbalance when positive outcomes are observed in both phase 2 and phase 3 trials. Establishing strong equipoise imbalance based on inconsistent outcomes of phase 2 and phase 3 is shown to require large sample sizes unlikely to improve current evidence standards.