Dose-Weighted Adjusted Mantel-Haenszel Tests for Numeric Scaled Strata in a Randomized Trial

Area of Science:

• Clinical Trials
• Biostatistics
• Statistical Methodology

Background:

• Randomized clinical trials may experience protocol deviations, such as reduced active treatment doses.
• Standard statistical methods may not adequately account for dose variations caused by such deviations.
• The extended Mantel-Haenszel (EMH) trend test was the original analysis plan for a three-arm trial.

Purpose of the Study:

• To develop and evaluate a novel statistical method to address protocol deviations involving fewer active treatment doses than planned.
• To compare the performance of the new method against existing statistical tests in simulated scenarios.

Main Methods:

• A dose-weighted adjusted extended Mantel-Haenszel (DWAEMH) test was developed, incorporating weights for actual available doses.
• Monte Carlo simulations using a logistic model were performed to estimate empirical power and size.
• The DWAEMH test was compared to the standard EMH trend test and logistic regression Wald tests.

Main Results:

• The DWAEMH test maintained statistical size (Type I error rate) under simulated conditions.
• The DWAEMH test demonstrated improved statistical power compared to EMH and logistic regression Wald tests when dose-response effects and interactions were present.
• The proposed method is implementable using standard statistical software.

Conclusions:

• The dose-weighted adjusted extended Mantel-Haenszel (DWAEMH) test is a robust statistical approach for analyzing clinical trial data with dose deviations.
• This method offers enhanced power and maintains accuracy, making it a valuable tool for biostatisticians dealing with real-world trial complexities.
• The DWAEMH test provides a practical solution for improving the analysis of randomized prevention trials with protocol deviations.