Contextualizing selection bias in Mendelian randomization: how bad is it likely to be?

TL;DR

This study assesses how selection bias influences Mendelian randomization results, finding it can cause significant bias under large effects but is often moderate, with mitigation possible through inverse probability weighting if models are correct.

Contribution

The paper provides a comprehensive simulation analysis of selection bias in Mendelian randomization, highlighting conditions under which bias is severe and evaluating correction methods.

Findings

Selection bias can severely impact Mendelian randomization estimates with large effects.

Moderate effects of selection bias generally lead to small bias and minimal Type 1 error inflation.

Inverse probability weighting can reduce bias if the selection model is correctly specified.

Abstract

Selection bias affects Mendelian randomization investigations when selection into the study sample depends on a collider between the genetic variant and confounders of the risk factor-outcome association. However, the relative importance of selection bias for Mendelian randomization compared to other potential biases is unclear. We performed an extensive simulation study to assess the impact of selection bias on a typical Mendelian randomization investigation. Selection bias had a severe impact on bias and Type 1 error rates in our simulation study, but only when selection effects were large. For moderate effects of the risk factor on selection, bias was generally small and Type 1 error rate inflation was not considerable. The magnitude of bias was also affected by the strength of confounder-risk factor and confounder-outcome associations, the structure of the causal diagram and selection frequency. The use of inverse probability weighting ameliorated bias when the selection model was correctly specified, but increased bias when selection bias was moderate and the model was misspecified. Finally, we investigated whether selection bias may explain a recently reported finding that lipoprotein(a) is not a causal risk factor for cardiovascular mortality in individuals with previous coronary heart disease.