Gatekeeping procedures in dose-response clinical trials based on the Dunnett test

Area of Science:

• Biostatistics
• Clinical Trial Methodology
• Pharmaceutical Research

Background:

• Multiple testing procedures are crucial in clinical trials to control Type I error rates.
• Dose-response studies require specialized statistical methods to evaluate treatment effects across different dosages.
• Existing gatekeeping procedures may not fully leverage hierarchical endpoint structures.

Purpose of the Study:

• To propose a general gatekeeping framework for multiple testing in dose-response clinical trials.
• To extend existing gatekeeping methods, including Dunnett and Bonferroni-based approaches.
• To improve statistical power by accounting for the hierarchical relationship between primary and secondary endpoints.

Main Methods:

• Development of a novel gatekeeping framework incorporating hierarchical testing.
• Extension of the Dunnett test and Bonferroni-based gatekeeping procedures.
• Illustration with a simulated dose-response clinical trial in diabetes patients.
• Monte-Carlo simulations to compare power with existing methods.

Main Results:

• The proposed gatekeeping procedure demonstrates a power advantage over the Bonferroni gatekeeping procedure.
• Power gains are more pronounced with higher correlations among endpoints.
• The procedure effectively restricts testing of secondary endpoints based on primary endpoint significance.

Conclusions:

• The novel gatekeeping framework offers a more powerful approach for multiple testing in dose-response clinical trials.
• This method is particularly beneficial when endpoints are correlated and primary comparisons are significant.
• The proposed procedure enhances statistical efficiency in evaluating drug efficacy across dose levels.