A Unified Approach to Covariate Adjustment for Survival Endpoints in Randomized Clinical Trials

Area of Science:

• Biostatistics
• Clinical Trials Methodology
• Survival Analysis

Background:

• Covariate adjustment enhances statistical efficiency in randomized trials by utilizing baseline covariate information.
• Established methods exist for continuous and binary endpoints, but covariate adjustment for survival endpoints remains complex and underutilized.
• Existing methods for survival data often require advanced statistical knowledge and computational skills.

Purpose of the Study:

• To present a novel, accessible augmentation approach for covariate adjustment in survival analysis.
• To improve the statistical efficiency and precision of treatment effect estimation for survival endpoints.
• To provide a widely applicable method that preserves key statistical properties under minimal assumptions.

Main Methods:

• Developed an augmentation approach that modifies existing treatment effect estimators for survival data.
• The augmentation preserves consistency and asymptotic normality, relying solely on randomization.
• Optimal augmentation terms are estimated using statistical and machine learning techniques, minimizing asymptotic variance.

Main Results:

• Simulation studies confirm substantial gains in statistical efficiency using the augmentation approach.
• The method demonstrates broad applicability across different effect measures for survival endpoints.
• The approach does not require restrictive assumptions like independent censoring or proportional hazards.

Conclusions:

• The augmentation approach offers a practical and effective solution for covariate adjustment in survival analysis.
• This method enhances the statistical power of randomized trials with survival endpoints.
• An R package, 'sleete', has been developed to implement this approach, facilitating its use in research.