Adaptive design and estimation in randomized clinical trials with correlated observations

Area of Science:

• Biostatistics
• Clinical Trial Design
• Biomedical Research

Background:

• Clinical trials with correlated data are common in biomedical research.
• Intracluster correlation significantly impacts sample size and power calculations.
• Fixed-sample designs may not be optimal for trials with correlated data.

Purpose of the Study:

• To propose a flexible, adaptive clinical trial design for correlated data.
• To incorporate adaptive monitoring and inference procedures.
• To ensure valid sample size and power calculations in the presence of intracluster correlation.

Main Methods:

• Adaptive sample size re-estimation using observed data.
• Weighted average of block-wise test statistics via generalized estimating equations.
• A stopping rule for early termination and acceptance of the null hypothesis.
• Updating effect size and within-cluster correlation estimates.

Main Results:

• The proposed design adaptively re-estimates sample size, unlike fixed-sample designs.
• Final inference uses a weighted average of test statistics, with weights based on accumulated data.
• The method allows for early trial termination if no significant treatment effect is found.
• Simulation studies demonstrate the operating characteristics of the adaptive design.

Conclusions:

• The proposed adaptive design offers flexibility and efficiency for clinical trials with correlated data.
• It provides a robust framework for sample size and power calculations, accounting for intracluster correlation.
• The method facilitates valid inference and potential early stopping, improving trial management.