A Bayesian Sequential Design for Clinical Trials with Time-to-Event Outcomes.

Area of Science:

• Biostatistics
• Clinical Trial Design
• Statistical Inference

Background:

• Increasing interest in Bayesian group sequential designs for clinical trials.
• Need for improved efficiency, shorter development times, and enhanced statistical precision.
• Maintaining clinical trial integrity and validity is paramount.

Purpose of the Study:

• To propose a Bayesian sequential design for clinical trials with time-to-event outcomes.
• To control the overall type I error rate using alpha spending functions.
• To adapt Bayes factor for decision-making during interim analyses.

Main Methods:

• Development of algorithms for decision rules and power calculation.
• Implementation of sensitivity analysis for prior parameter impact.
• Simulation studies comparing Bayesian and frequentist group sequential designs.

Main Results:

• The proposed Bayesian design can achieve greater power at a fixed number of events.
• The Bayesian design requires a smaller expected event size with appropriate priors.
• Feasibility demonstrated using a real-world clinical trial dataset.

Conclusions:

• Bayesian group sequential designs offer advantages in power and efficiency over frequentist approaches.
• Appropriate prior selection is crucial for optimizing Bayesian sequential trial performance.
• The proposed Bayesian design is a feasible and potentially superior alternative for clinical trials.