Frequentist-calibrated Bayesian group sequential design with dynamic borrowing

2026-07-31

AI Summary

This work proposes a Bayesian group sequential design with dynamic information borrowing that reconciles the efficiency gains of incorporating historical data with the regulatory requirement of strict Type I error control. By establishing an explicit correspondence between posterior probability decision rules and uniformly most powerful frequentist tests, the method employs dual thresholds at each interim analysis: one to guarantee exact Type I error control and another to adaptively borrow strength from historical data when appropriate. This approach uniquely unifies frequentist error calibration with Bayesian information borrowing without sacrificing power. Numerical experiments demonstrate that the design maintains the nominal Type I error rate while substantially improving statistical power, and it has been successfully implemented in the design of a Phase III tuberculosis prevention trial integrating historical data from both adult and pediatric populations.

Abstract

Bayesian analysis is increasingly used in clinical trials. However, assessment of the design with respect to the frequentist operating characteristics, such as type I error and power, remains a regulatory requirement in many cases. It is well established that, when information is borrowed from external sources to the trial, imposing strict frequentist type I error rate control is equivalent to offsetting the borrowing, which results in no power gains. We propose a Bayesian group sequential design with dynamic borrowing that exploits an explicit correspondence between Bayesian decision criteria based on posterior odds and frequentist uniformly most powerful (UMP) tests. At each interim analysis, two evidential thresholds are made available: the one that exactly retrieves the frequentist UMP decision; the other, that allows the investigator to incorporate historical information when appropriate. We assess the performance of the proposed approach in numerical studies, and apply the framework to the design of a phase III tuberculosis prevention trial, incorporating historical adult and pediatric trial data.

Problem

• Bayesian analysis
• frequentist operating characteristics
• type I error
• dynamic borrowing
• clinical trials

Innovation

• Bayesian group sequential design
• dynamic borrowing
• frequentist calibration
• posterior odds
• uniformly most powerful test