Marginal hazard regression for correlated failure time data with auxiliary covariates

Area of Science:

• Biostatistics
• Epidemiology
• Survival Analysis

Background:

• Biomedical studies often face budget constraints limiting primary covariate collection to a subset of the cohort.
• Inexpensive auxiliary covariates are frequently available for the entire cohort, presenting an opportunity for enhanced data analysis.

Purpose of the Study:

• To develop a statistical method that leverages auxiliary covariate information to improve the efficiency of correlated failure time data analysis.
• To address the challenge of incomplete primary covariate data in cohort studies.

Main Methods:

• Development of an estimated partial likelihood approach for correlated failure time data incorporating auxiliary information.
• Assumption of a marginal hazard model with a common baseline hazard function.
• Utilizing modern empirical process theory for asymptotic property proofs, as classical martingale theory is insufficient.

Main Results:

• The proposed statistical method demonstrated increased efficiency compared to existing methods in simulation studies.
• Asymptotic properties of the developed estimators were rigorously derived using advanced statistical theory.

Conclusions:

• The novel estimated partial likelihood approach effectively utilizes auxiliary information to enhance the efficiency of correlated failure time data analysis in biomedical research.
• The method provides a statistically valid and more powerful alternative for studies with budget-limited primary data collection.