Robust Mendelian randomization in the presence of residual population stratification, batch effects and horizontal pleiotropy

Abstract

Mendelian Randomization (MR) studies are threatened by population stratification, batch effects, and horizontal pleiotropy. Although a variety of methods have been proposed to mitigate those problems, residual biases may still remain, leading to highly statistically significant false positives in large databases. Here we describe a suite of sensitivity analysis tools that enables investigators to quantify the robustness of their findings against such validity threats. Specifically, we propose the routine reporting of sensitivity statistics that reveal the minimal strength of violations necessary to explain away the MR results. We further provide intuitive displays of the robustness of the MR estimate to any degree of violation, and formal bounds on the worst-case bias caused by violations multiple times stronger than observed variables. We demonstrate how these tools can aid researchers in distinguishing robust from fragile findings by examining the effect of body mass index on diastolic blood pressure and Townsend deprivation index.

Strategies to investigate and mitigate collider bias in genetic and Mendelian randomisation studies of disease progression

“…Finally, we should mention that quantifying and adjusting for index event bias in case-only studies is an active area of research, and we expect new methods to emerge in the coming years. For example, Cinelli et al [ 57 ] recently developed a sensitivity analysis tool for pleiotropic bias and population stratification in MR. Their approach can be used to model collider bias in MR studies where conditioning on a collider opens a pleiotropic path between the instrument and the outcome, meaning that it may be applicable to studies of disease progression, although this was only briefly explored by Cinelli et al and more research in that direction may be needed. As another example, Heckman-type sample selection models [ 58 ] are often used in econometrics to adjust for bias due to missing data, and a recent paper explored whether these models can also be applied to genetic studies [ 59 ].…”

What can Mendelian randomization contribute to biological anthropology?

“…Another potential source of bias is population stratification, which occurs when groups of individuals differ systematically in genetic ancestry and the phenotype of interest, producing spurious associations, something that can be especially problematic in admixed populations, such as those from Latin America. Several MR extensions that account for the violation of its assumptions have been—and continue to be—developed, allowing the researcher to run multiple sensitivity analyses that will detect bias and deliver corrected estimates (Burgess et al, 2017; Cinelli et al, 2022). The best known of these are the MR Egger, weighted median, weighted mode and MR‐PRESSO methods (Hartwig et al, 2017; Haycock et al, 2016; Verbanck et al, 2018).…”

Body Mass Index and the Risk of Rheumatic Disease: Linear and Nonlinear Mendelian Randomization Analyses

“…The presence of weak instruments is a violation of the first assumption, while the second and third assumptions are violated by the presence of horizontal pleiotropy. Any violation of these 3 assumptions may result in biased estimates (10). MR investigations are commonly designed as 1‐ or 2‐sample studies.…”