A covariate-adjustment regression model approach to noninferiority margin definition

Area of Science:

• Biostatistics
• Clinical Trials
• Pharmaceutical Research

Background:

• Noninferiority trials rely on the constancy assumption for interpretable results.
• Discrepancies between historical and active control trials often violate this assumption.
• Current methods lack quantitative tools to assess the impact of these violations.

Purpose of the Study:

• To quantify the impact of population differences on constancy assumption violation in noninferiority trials.
• To redefine active control treatment effects when violations are unacceptable.
• To ensure accurate interpretation of experimental treatment effects.

Main Methods:

• Developed a covariate-adjustment generalized linear regression model.
• Quantified population differences and their effect on the constancy assumption.
• Redefined the noninferiority margin based on quantified violations.

Main Results:

• The model successfully quantifies the impact of population differences on constancy assumption violation.
• Identified conditions under which population differences lead to unacceptable violations.
• Demonstrated the ability to redefine the active control treatment effect and noninferiority margin.

Conclusions:

• The covariate-adjustment approach provides a quantitative method to address constancy assumption violations.
• This method enables accurate determination of experimental treatment effects in noninferiority trials.
• Facilitates robust interpretation of trial results despite population discrepancies.