The illness-death model for family studies

Area of Science:

• Biostatistics
• Epidemiology
• Genetic Epidemiology

Background:

• Family studies often use biased sampling, selecting affected individuals with right-truncated disease onset ages.
• Existing statistical methods are limited for complex life history processes in such family data.
• Analyzing within-family dependence and covariate effects on disease onset requires advanced modeling.

Purpose of the Study:

• To develop statistical methods for analyzing clustered illness-death processes in family studies with biased sampling.
• To investigate covariate effects on the marginal intensity of disease onset.
• To quantify within-family dependence in disease onset times.

Main Methods:

• A simple joint model for clustered illness-death processes was formulated.
• Likelihood and composite likelihood methods were constructed for biased family data.
• Auxiliary data augmentation of composite likelihood was used for insufficient data scenarios.

Main Results:

• The proposed joint model effectively analyzes covariate effects on disease onset intensity.
• Within-family dependence in disease onset times can be accurately studied.
• Composite likelihood with auxiliary data improves estimation for rare diseases with limited data.

Conclusions:

• The developed statistical framework accommodates biased sampling in family studies for complex disease processes.
• The methods provide robust estimation of covariate effects and familial aggregation.
• The approach was successfully applied to a family study of psoriatic arthritis.