# Utilizing offspring genotype-by-proxy Mendelian randomization to investigate the causal effect of offspring perinatal traits on maternal health

## TL;DR

This paper introduces a new method to study how a baby's traits during birth can affect the mother's health using genetic data from fathers.

## Contribution

The novel 'offspring genotype-by-proxy' Mendelian randomization framework allows causal inference without direct offspring genetic data.

## Key findings

- The framework uses paternal genotypes to proxy offspring genotypes for causal effect estimation.
- Model misspecification and spousal misclassification can impact statistical power and bias.
- The approach can be implemented in large biobanks with spousal genotype data but not mother-offspring pairs.

## Abstract

During the perinatal period, the fetus can exert profound effects on processes that alter pre- and postnatal maternal physiology. It is possible to investigate the causal effect of offspring perinatal exposures on their mother’s health using Mendelian randomization (MR). However, analyses need to be adjusted for maternal genotype to avoid confounding. Such analyses are difficult to perform at scale because of the paucity of cohorts across the world with large numbers of genotyped maternal–offspring dyads and parent–offspring trios.

We introduce the “offspring genotype-by-proxy” MR framework which can be employed in the absence of offspring genetic information to complement existing approaches in the triangulation of causal inference. The basic idea is to use paternal genotypes to proxy the direct effect of their offspring’s genotype on their offspring’s own exposures.

We compare our framework to other MR designs and investigate the consequences of model misspecification and spousal misclassification on statistical power, consistency, and bias. In addition, we discuss the key MR assumptions that prevent these approaches from being appropriate for investigating the effect of many offspring postnatal and later life exposures on maternal health.

Given the increasing availability of datasets such as the UK Biobank that (incidentally) include tens of thousands of genome-wide genotyped spousal pairs and large population biobanks with linked health record data for first-degree relatives, the offspring genotype-by-proxy MR approach could augment causal analyses of offspring perinatal exposures on their mother’s outcomes as implementation is not restricted to datasets with mother–offspring genotype information.