Robust Use of External Control Information in Clinical Trials

• Activity Number: 45 - Statistical Methods Utilizing Historical and Real-World Data for Clinical Studies
• Type: Topic-Contributed
• Date/Time: Sunday, August 8, 2021 : 3:30 PM to 5:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Heinz Schmidli*
• Companies: Novartis
• Keywords: Bayesian; clinical trial; external control; historical control; rare disease; Mixture prior

Abstract

Use of external control information in randomized controlled clinical trials allows us to reduce the number of patients randomized to control. This is attractive from an ethical, feasibility and cost perspective. Robust Bayesian meta-analytic-predictive methods (Schmidli et al., 2014, Biometrics; Schmidli et al., 2020, Clinical Pharmacology and Therapeutics) synthesize the evidence of relevant external control data, and predict the control effect in the planned study. The inference is made robust to conflicts between prediction and observed data by use of mixture priors. The approach is illustrated by a planned randomized adaptive trial in children with multiple sclerosis. The proposed trial design has been evaluated under US FDA’s Complex Innovative Designs pilot program.