A simple frailty model for family studies with covariates

Area of Science:

• Biostatistics
• Epidemiology
• Statistical Modeling

Background:

• Frailty, an unobserved heterogeneity, can significantly impact survival analysis and risk factor estimation.
• Traditional models may not adequately account for this unobserved variability, leading to biased results.
• Accurate modeling of frailty is essential for reliable epidemiological and clinical research.

Purpose of the Study:

• To develop and analyze a simple, user-friendly frailty model using logistic regression with binary frailties.
• To present methods for testing frailty, estimating frailty parameters, and covariate effects.
• To investigate the consequences of ignoring or mis-specifying frailty in statistical analyses.

Main Methods:

• Development of a logistic regression model incorporating binary frailties.
• Presentation of statistical methods for frailty testing and parameter estimation.
• Simulation studies and analysis of a real-world dataset (palmar keratoses) to evaluate model performance.

Main Results:

• The proposed frailty model is simple and amenable to analysis.
• Ignoring or mis-specifying frailty leads to downward bias in covariate effect estimates, even with over-estimation.
• Accurate estimation of large frailties can be challenging due to likelihood flatness, but this minimally impacts covariate effect estimates.

Conclusions:

• The developed frailty model provides a practical approach for accounting for unobserved heterogeneity.
• The study highlights the importance of correctly specifying frailty models to avoid biased covariate effect estimates.
• The methodology is applicable to various scenarios, including those with multiple frailty sources like sibships and spouse pairs.