Futility Designs

Limited resources require that interventions be evaluated for an efficacy signal in Phase II prior to initiation of large and costly confirmatory Phase III clinical trials. The standard concurrently controlled superiority design is not well-suited for this evaluation. Because the Phase II superiority design is often underpowered to detect clinically meaningful improvements, investigators are left to make subjective decisions in the face of a nonsignificant test result. The futility design reframes the statistical hypothesis in order to discard interventions which do not demonstrate sufficient promise. The alternative hypothesis is that the effect is less than some minimally worthwhile threshold. In this way, the trial can be appropriately powered to evaluate whether the intervention is worth pursuing in Phase III and thus provides a clear “no go” signal. We briefly describe the superiority design in order to compare and contrast with the futility design. We then describe both the single-arm and concurrently controlled futility designs and present case studies of each. Lastly, we discuss some key considerations related to sample size calculation and interim analysis.