Bias in parameter estimates due to omitting gene-environment interaction terms in case-control studies

Area of Science:

• Genetics
• Biostatistics
• Epidemiology

Background:

• Genetic studies generate vast data for examining genetic effects.
• Gene-environment interaction (G×E) occurs when genetic variant effects differ based on nongenetic factors.
• Neglecting G×E can lead to substantial bias in estimates of main genetic effects.

Purpose of the Study:

• To derive a general and convenient approximation for the magnitude of bias caused by omitting the G×E term in genetic analyses.
• To assess the accuracy of this approximation in finite samples.
• To apply the approximation to a real-world genetic study, specifically in Alzheimer's disease.

Main Methods:

• Derivation of a mathematical approximation for bias due to omitted G×E.
• Validation of the approximation's accuracy using simulations or finite sample analysis.
• Application of the derived approximation in a case study involving Alzheimer's disease genetics.

Main Results:

• A general approximation for bias magnitude due to omitted G×E was successfully derived.
• The approximation demonstrated reasonable accuracy in finite samples.
• The method was applied to analyze genetic data in Alzheimer's disease, providing insights into potential biases.

Conclusions:

• Omitting G×E in genetic studies can significantly bias main effect estimates.
• The derived approximation offers a practical tool to quantify and potentially correct for this bias.
• This approach enhances the reliability of genetic association studies, including those for complex diseases like Alzheimer's.