Bayesian Design for Pediatric Clinical Trials with Binary Endpoints When Borrowing Historical Information of...

Area of Science:

• Biostatistics
• Clinical Pharmacology
• Pediatric Drug Development

Background:

• Pediatric efficacy evaluation is crucial for drug development but faces enrollment challenges, especially for rare diseases.
• Bayesian frameworks offer methods to leverage historical data, potentially increasing pediatric trial efficiency and reducing sample sizes.

Purpose of the Study:

• To extend existing Bayesian hierarchical models for pediatric efficacy extrapolation.
• To develop methods for efficiently borrowing strength from multiple historical trials.
• To adapt models for both continuous and binary endpoints in pediatric drug development.

Main Methods:

• Extension of Schoenfeld et al.'s Bayesian hierarchical model to incorporate multiple historical trials.
• Development of a quantitative method for efficient historical information borrowing.
• Adaptation of the model for binary endpoints using a hierarchical binomial model.

Main Results:

• The proposed methods allow for more efficient utilization of historical data from multiple sources.
• The extended model successfully extrapolates efficacy for both continuous and binary endpoints.
• Simulations and a case study demonstrate the practical application and robustness of the methods.

Conclusions:

• The enhanced Bayesian approach improves the efficiency of pediatric drug trials by effectively utilizing historical data.
• The methods provide a flexible framework for efficacy extrapolation across different endpoint types.
• Careful consideration of prior distributions and sensitivity analyses are important for reliable application.