Adaptive Bayesian Clinical Trial Design for Improving Robustness to Sample Size Calculation Assumptions: A Simulation Study

• Activity Number: 341 - Topics in Adaptive Designs: Sample Size, Randomization and Related Topics
• Type: Contributed
• Date/Time: Wednesday, August 5, 2020 : 10:00 AM to 2:00 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Shirley Liao* and Fabio Pellegrini and Bernd Kieseier and Christophe Hotermans and Jason Mendoza and Christine Lebrun and Daniel Pelletier and Aksel Siva and Maria Pia Sormani and Orhun Kantarci and Darin Okuda and Carl de Moor
• Companies: Biogen and Biogen International GmbH and Biogen and Biogen and Biogen and Centre Hospitalier Universitaire de Nice and University of Southern California and e Department of Neurology of Istanbul University, Cerrahpa?a School of Medicine and Università degli Studi di Genova and Mayo Clinic and UT Southwestern Medical Center and Biogen Inc
• Keywords: Bayesian; adaptive clinical trial design; interim analysis

Abstract

Bayesian clinical trial design has gained traction in rare disease areas where the utilization of Bayesian priors in sample size estimation may allow a trial to maintain minimum required power with a smaller sample size. However, the purported power of the design may be much lower than expected if parametric assumptions typical for a Bayesian sample size calculation are violated, which is especially likely in the study of rare diseases where little if any preliminary data is available to inform event rates, treatment effect or strong Bayesian priors. We present a realistic scenario for Bayesian clinical trial design for survival data and demonstrate through simulations how strong parametric assumptions may lead to overconfidence in expected power as well as an inflated Type-1 error rate when assumptions are violated. We then compare multiple options for optimizing the completion of this trial, including a Bayesian adaptive design which incorporates an interim analysis for sample size recalculation.