Blinded sample size reestimation in non-inferiority trials with binary endpoints.

Area of Science:

• Clinical Trials
• Biostatistics
• Statistical Methodology

Background:

• Accurate sample size calculation is crucial for clinical trial success.
• Uncertainty in parameter estimates can compromise trial validity.
• Internal pilot studies allow for sample size reestimation.

Purpose of the Study:

• To compare fixed-size designs with sample size reestimation designs for non-inferiority trials.
• To evaluate the impact of parameter misspecification on Type I error rate and power.
• To demonstrate the benefits of sample size reestimation in addressing estimation uncertainty.

Main Methods:

• Simulation studies comparing fixed-size and reestimation designs.
• Analysis of Type I error rate and statistical power.
• Focus on non-inferiority trials with binary outcomes.

Main Results:

• Sample size reestimation designs effectively correct for initial parameter misspecification.
• Designs with reestimation maintain desired power and Type I error rates.
• Fixed-size designs are sensitive to errors in nuisance parameter estimation.

Conclusions:

• Sample size reestimation is a valuable strategy for non-inferiority trials with binary outcomes.
• This approach enhances the robustness of clinical trial designs against parameter uncertainty.
• Implementing reestimation improves the reliability of trial results and resource allocation.