Notes on testing noninferiority in multivariate binary data under the matched-pair design

Area of Science:

• Biostatistics
• Clinical Trial Design
• Statistical Modeling

Background:

• Therapeutic efficacy is often assessed using multiple endpoints, necessitating methods to integrate this information.
• Univariate noninferiority tests may lack power when dealing with multiple response variables.

Purpose of the Study:

• To develop and evaluate procedures for testing noninferiority in multivariate binary data using a matched-pair design.
• To improve the statistical power of noninferiority tests by incorporating information from multiple response variables.
• To apply these methods to real-world clinical trial data.

Main Methods:

• Development of a mixed-effects logistic regression model for multivariate binary data.
• Application of Bonferroni's and Scheffe's methods for controlling Type I error rates.
• Utilizing Monte Carlo simulations to assess the performance of the proposed test procedures.
• Illustration using data from a crossover clinical trial on an antidepressive drug.

Main Results:

• The proposed mixed-effects logistic regression model effectively handles multivariate binary data in matched-pair designs.
• The developed noninferiority tests show improved power compared to univariate approaches.
• Bonferroni's and Scheffe's methods successfully controlled Type I error inflation.
• The methods were successfully applied to analyze adverse events in a clinical trial.

Conclusions:

• The proposed statistical procedures offer a powerful approach for noninferiority testing with multivariate binary outcomes in matched-pair studies.
• Integrating multiple endpoints enhances the ability to detect noninferiority.
• These methods are valuable for clinical trial analysis, particularly when assessing multiple adverse events or efficacy measures.