Multivariate logistic regression for familial aggregation in age at disease onset

Area of Science:

• Biostatistics
• Genetic Epidemiology
• Public Health

Background:

• Familial aggregation studies identify diseases clustering within families to find hereditary risk factors.
• Accurate assessment requires accounting for age at disease onset, particularly for late-onset diseases.
• Existing models may not fully capture age-dependent familial associations.

Purpose of the Study:

• To propose a novel discrete time statistical model for analyzing familial disease aggregation.
• To incorporate age at disease onset and allow familial associations to vary with age.
• To provide interpretable parameters for cancer risk assessment.

Main Methods:

• Developed a discrete time model to analyze familial disease aggregation.
• The model accounts for age at disease onset and allows familial association to vary with age.
• Parameters are interpreted as conditional log-odds ratios, akin to discrete time conditional cross hazard ratios.

Main Results:

• The proposed model effectively accounts for age at disease onset in familial aggregation studies.
• Familial associations can be modeled to vary with age and be modified by covariates.
• The model's parameters offer interpretable insights into disease risk, particularly for cancer.

Conclusions:

• The discrete time model provides a robust framework for studying hereditary disease risk.
• It enhances the analysis of familial aggregation by incorporating age-specific associations.
• The model is applicable to large family studies, such as the Cancer Genetics Network (CGN).