Double-robust semiparametric estimator for differences in restricted mean lifetimes in observational studies

Area of Science:

• Biostatistics
• Epidemiology
• Survival Analysis

Background:

• Restricted mean lifetime is crucial for comparing groups in epidemiologic studies with censored data.
• Clinical decisions, such as kidney transplant outcomes, require comparing survival across different groups.
• Confounding factors necessitate covariate adjustment for accurate comparisons in non-randomized studies.

Purpose of the Study:

• To develop a robust statistical estimator for differences in restricted mean lifetime.
• To account for confounding factors in observational survival studies.
• To provide a reliable method for comparing outcomes in clinical and epidemiological research.

Main Methods:

• Utilized semiparametric theory to create a novel estimator.
• Developed working models for both the time-to-event and the coarsening mechanism (group assignment and censoring).
• Demonstrated the double robust property of the proposed estimator.

Main Results:

• The estimator is consistent and asymptotically normal when either the time-to-event or coarsening model is correctly specified.
• Simulation studies confirmed the estimator's good finite-sample performance.
• The method was successfully applied to national kidney transplant data.

Conclusions:

• The proposed double robust estimator effectively compares restricted mean lifetimes in the presence of confounding.
• This method offers a reliable tool for epidemiologic research and clinical decision-making, particularly in transplantation.
• The approach enhances the validity of survival comparisons using observational data.