Bias in Mendelian randomization due to assortative mating

Area of Science:

• Epidemiology
• Statistical Genetics

Background:

• Mendelian randomization (MR) strengthens causal inference in observational studies.
• Existing MR methods address bias from horizontal pleiotropy.
• Assortative mating presents a novel source of bias in MR.

Purpose of the Study:

• To investigate the impact of assortative mating on Mendelian randomization.
• To determine if assortative mating introduces bias in MR analyses.
• To explore methods for detecting and correcting this bias.

Main Methods:

• Simulations were conducted to model assortative mating effects.
• Assessed bias from cross-trait and single-trait assortative mating.
• Evaluated MR methods' robustness to assortative mating bias.

Main Results:

• Assortative mating causes bias in MR, irrespective of the phenotypes under assortment.
• Bias accumulates across generations and affects pleiotropy-robust MR methods.
• Genetic data from mother-father-offspring trios can detect and correct this bias.

Conclusions:

• Assortative mating is a significant confounder in Mendelian randomization studies.
• Standard MR bias correction methods do not account for assortative mating.
• Trio data offers a viable solution for mitigating assortative mating bias in MR.