Blinded sample size recalculation in multiple composite population designs with normal data and baseline adjustments.

Area of Science:

• Clinical Trials
• Biostatistics
• Personalized Medicine

Background:

• Growing interest in subpopulation analysis for personalized medicine and targeted therapies.
• Need for robust trial designs and analysis methods to evaluate treatment effects in specific patient groups.

Purpose of the Study:

• To propose a novel trial design for analyzing composite populations, defined as collections of disjoint subsets.
• To develop methods for testing treatment effects and calculating sample sizes for these composite populations.

Main Methods:

• Defining composite populations and applying the trial design to normally distributed endpoints with random baseline covariates.
• Combining p-values from subset levels using the inverse normal combination function for composite population tests.
• Utilizing closed testing procedures and multivariate normal distributions for hypothesis testing and sample size calculations.

Main Results:

• Simulations show no significant inflation of the type I error rate.
• The proposed methods effectively achieve or closely meet target power after sample size recalculation.

Conclusions:

• The developed trial design and analysis methods are effective for composite populations in personalized medicine.
• The approach provides a statistically sound framework for evaluating targeted therapies in subpopulations.