Interpreting findings from Mendelian randomization using the MR-Egger method - PubMed Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2017 May;32(5):377-389.
doi: 10.1007/s10654-017-0255-x. Epub 2017 May 19.

Interpreting findings from Mendelian randomization using the MR-Egger method

Stephen Burgess  1   2 , Simon G Thompson  3
Affiliations

Interpreting findings from Mendelian randomization using the MR-Egger method

Stephen Burgess et al. Eur J Epidemiol. 2017 May.

Erratum in

Abstract

Mendelian randomization-Egger (MR-Egger) is an analysis method for Mendelian randomization using summarized genetic data. MR-Egger consists of three parts: (1) a test for directional pleiotropy, (2) a test for a causal effect, and (3) an estimate of the causal effect. While conventional analysis methods for Mendelian randomization assume that all genetic variants satisfy the instrumental variable assumptions, the MR-Egger method is able to assess whether genetic variants have pleiotropic effects on the outcome that differ on average from zero (directional pleiotropy), as well as to provide a consistent estimate of the causal effect, under a weaker assumption-the InSIDE (INstrument Strength Independent of Direct Effect) assumption. In this paper, we provide a critical assessment of the MR-Egger method with regard to its implementation and interpretation. While the MR-Egger method is a worthwhile sensitivity analysis for detecting violations of the instrumental variable assumptions, there are several reasons why causal estimates from the MR-Egger method may be biased and have inflated Type 1 error rates in practice, including violations of the InSIDE assumption and the influence of outlying variants. The issues raised in this paper have potentially serious consequences for causal inferences from the MR-Egger approach. We give examples of scenarios in which the estimates from conventional Mendelian randomization methods and MR-Egger differ, and discuss how to interpret findings in such cases.

Keywords: Instrumental variable; MR-Egger; Mendelian randomization; Robust methods; Summarized data.

PubMed Disclaimer

Figures

Fig. 1
Fig. 1
Decomposing association for genetic variant Gj with the outcome into a indirect (causal) effect via the risk factor and an direct (pleiotropic) effect (see Eq. 1)
Fig. 2
Fig. 2
Graph showing simulated (left panel) and real-data (right panel) examples in which inverse-variance weighted estimate (solid line) and MR-Egger estimate (dashed line) differ substantially. Each point represents the per allele associations of a single genetic variant (lines from each point are 95% confidence intervals for the associations). In both cases, the inverse-variance weighted estimate is positive, whereas the MR-Egger causal estimate is null with intercept term differing from zero
Fig. 3
Fig. 3
Graph showing same simulated example as in Fig. 2 (left panel), except that three variants are positively orientated and two negatively. The inverse-variance weighted estimate (solid line) is unaffected by the orientation of variants, whereas the MR-Egger estimate (dashed line) is affected by the choice of orientation, with the intercept term attenuating and the MR-Egger estimate approaching the inverse-variance weighted estimate
Fig. 4
Fig. 4
Graph showing hypothetical example in which genetic associations with the risk factor and with the outcome are similar for all variants. Left panel inverse-variance weighted estimate (solid line) and 95% confidence interval (grey area) suggest strong evidence for a positive causal effect. Right panel MR-Egger estimate (dashed line) and 95% confidence interval (grey area) suggest no evidence against the instrumental variable assumptions (intercept test), but also no evidence for a causal effect (causal test)
Fig. 5
Fig. 5
Graph showing same hypothetical example as Fig. 4 (left panel) except for the addition of a single extra genetic variant (right panel). Left panel inverse-variance weighted estimate (solid line) and MR-Egger estimate (dashed line) are similar. Right panel inverse-variance weighted estimate (solid line) and MR-Egger estimate (dashed line) are markedly different, as the influential genetic variant changes the sign of the MR-Egger estimate
Fig. 6
Fig. 6
Potential violations of the InSIDE assumption. Top panel pleiotropic effects act directly on the outcome (InSIDE satisfied); middle panel pleiotropic effects act on the outcome via single confounder (InSIDE violated); bottom panel pleiotropic effects act on the outcome via different confounders (InSIDE still violated). Arrows from the genetic variants to the risk factor may not be present for all variants; some variants may affect the confounder directly and not the risk factor. Notation: G1, G2, …, GJ, genetic variants; X, risk factor; Y, outcome; U, confounder. Pleiotropic effects are signified by curved arrows
Fig. 7
Fig. 7
Graph showing further real example in which inverse-variance weighted estimate (solid line) and MR-Egger estimate (dashed line) differ substantially. Each point represents the per allele associations of a single genetic variant (lines from each point are 95% confidence intervals for the associations). Associations with HDL-cholesterol are in standard deviation units and associations with CHD risk are log odds ratios
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

See all "Cited by" articles

References

    1. Davey Smith G, Ebrahim S. ‘Mendelian randomization’: can genetic epidemiology contribute to understanding environmental determinants of disease? Int J Epidemiol. 2003;32(1):1–22. doi: 10.1093/ije/dyg070. - DOI - PubMed
    1. Burgess S, Thompson SG. Mendelian randomization: methods for using genetic variants in causal estimation. London: Chapman & Hall; 2015.
    1. Davey Smith G, Ebrahim S. Mendelian randomization: prospects, potentials, and limitations. Int J Epidemiol. 2004;33(1):30–42. doi: 10.1093/ije/dyh132. - DOI - PubMed
    1. Didelez V, Sheehan N. Mendelian randomization as an instrumental variable approach to causal inference. Stat Methods Med Res. 2007;16(4):309–330. doi: 10.1177/0962280206077743. - DOI - PubMed
    1. Greenland S. An introduction to instrumental variables for epidemiologists. Int J Epidemiol. 2000;29(4):722–729. doi: 10.1093/ije/29.4.722. - DOI - PubMed

Publication types

LinkOut - more resources