CASE-CONTROL SURVIVAL ANALYSIS WITH A GENERAL SEMIPARAMETRIC SHARED FRAILTY MODEL - A PSEUDO FULL LIKELIHOOD APPROACH

Area of Science:

• Biostatistics
• Epidemiology
• Survival Analysis

Background:

• Correlated failure time data from population-based case-control studies present unique analytical challenges.
• Existing methods lack established large sample theory due to retrospective sampling schemes.
• Understanding risk factor effects and within-family dependencies is crucial for these data.

Purpose of the Study:

• To develop a novel statistical technique for estimating parameters in semiparametric shared frailty models with correlated failure time data.
• To provide a non-iterative nonparametric estimator for the cumulative baseline hazard function.
• To establish rigorous large sample theory for the proposed estimation methods.

Main Methods:

• Development of a novel estimation technique for a general semiparametric shared frailty model.
• Introduction of a simple, non-iterative nonparametric estimator for the cumulative baseline hazard function.
• Rigorous derivation of large sample theory for the proposed methods.

Main Results:

• A new method for estimating risk factor effects on hazard functions and within-family dependencies was developed.
• A computationally simple and non-iterative estimator for the cumulative baseline hazard was presented.
• The proposed method is supported by rigorous large sample theory.

Conclusions:

• The developed technique offers a robust approach for analyzing correlated failure time data in case-control studies.
• The method provides reliable parameter estimation and hazard function estimation with established theoretical guarantees.
• Simulation results and a real data example demonstrate the practical utility of the proposed statistical approach.