Ascertainment-adjusted maximum likelihood estimation for the additive genetic gamma frailty model

Area of Science:

• Biostatistics
• Genetic Epidemiology
• Statistical Genetics

Background:

• Complex diseases require sophisticated genetic linkage analysis methods.
• The additive genetic gamma frailty model accounts for variable age of onset and covariates.
• Ascertainment biases can affect parameter estimates; retrospective likelihood methods may lose efficiency.

Purpose of the Study:

• To evaluate two parameter estimation approaches for the additive gamma frailty model.
• To address sibship ascertainment based on having at least two affected individuals before a specified age.
• To compare a conditional likelihood approach with a full ascertainment-adjusted likelihood approach.

Main Methods:

• Derivation of explicit forms for likelihood functions under two estimation strategies.
• Simulation studies to assess parameter estimation accuracy and bias.
• Evaluation under conditions of pre-specified versus simultaneously estimated baseline hazard functions.

Main Results:

• Both conditional and ascertainment-adjusted likelihood methods yield accurate estimates when the baseline hazard is pre-specified.
• Only the ascertainment-adjusted likelihood method provides unbiased parameter estimates when the baseline hazard must be estimated.
• Simulation results highlight the superiority of the ascertainment-adjusted approach under complex scenarios.

Conclusions:

• The ascertainment-adjusted likelihood method is crucial for unbiased parameter estimation in additive gamma frailty models.
• This method effectively handles ascertainment biases in genetic linkage analysis for complex diseases.
• The ascertainment-adjusted likelihood ratio test is a viable tool for genetic linkage analysis using this model.