Robust estimation for secondary trait association in case-control genetic studies.

Area of Science:

• Genetics
• Biostatistics
• Epidemiology

Background:

• Secondary trait genetic association studies offer insights into disease etiology.
• Estimating these associations requires careful consideration of potential biases, particularly from case-control sampling.
• Maximum likelihood (ML) and inverse probability weighted (IPW) estimators are common methods for such analyses.

Purpose of the Study:

• To compare the efficiency and robustness of various inverse probability weighted (IPW) and maximum likelihood (ML) estimators for secondary trait genetic association.
• To evaluate the impact of model misspecification on these estimators.
• To identify reliable methods for genetic association studies with complex sampling schemes.

Main Methods:

• Comparison of inverse probability weighted (IPW) estimators and maximum likelihood (ML) estimators.
• Assessment of estimator performance under correct and misspecified models for primary and secondary traits.
• Evaluation of IPW estimators with nonparametrically estimated selection probabilities and augmented IPW estimators.

Main Results:

• Maximum likelihood (ML) methods can severely inflate type I error rates when the primary trait model is misspecified.
• Inverse probability weighted (IPW) estimators are generally less efficient than ML estimators but demonstrate robustness against model misspecification.
• Augmented IPW and IPW with nonparametrically estimated selection probabilities show improved efficiency when secondary trait data is available for the entire cohort.

Conclusions:

• Inverse probability weighted (IPW) estimators provide a flexible and robust approach for secondary trait genetic association studies, especially those with complex sampling designs.
• IPW-based methods are a viable and reliable option for analyzing large genetic association studies.
• Caution is advised with ML methods due to their sensitivity to model misspecification in genetic association analyses.