Sample size reestimation by Bayesian prediction.

Area of Science:

• Clinical Trials
• Biostatistics
• Statistical Inference

Background:

• Interim data monitoring is crucial for adaptive clinical trial designs.
• Sample size reestimation is a key adaptive strategy to maintain trial power.
• Existing methods like predictive power and conditional power have limitations.

Purpose of the Study:

• To review and propose a Bayesian predictive approach for interim data monitoring.
• To apply this approach for interim sample size reestimation in clinical trials.
• To compare the proposed method with existing approaches.

Main Methods:

• Utilizing a Bayesian predictive framework to forecast trial outcomes.
• Predicting required sample size adjustments based on interim data to achieve a target success probability.
• Comparative analysis with predictive power and conditional power methods using real clinical trial data.

Main Results:

• The Bayesian predictive approach offers a robust method for sample size reestimation.
• Demonstrated ability to maintain desired success probabilities with adjusted sample sizes.
• Identified advantages over traditional predictive power and conditional power methods.

Conclusions:

• The Bayesian predictive approach is a valuable tool for adaptive clinical trial design.
• This method provides a principled way to adjust sample sizes based on accumulating evidence.
• It enhances the efficiency and success likelihood of clinical trials.