CoChat
Sign Up
Cited by 1,215Advanced Causal Inference TechniquesGenetic Associations and EpidemiologyStatistical Methods in Epidemiology

The DOI matches its Crossref record; 70 of the 70 references it cites were found in a registry or an index.

Mendelian randomisation for mediation analysis: current methods and challenges for implementation

Open access
Carter, Alice R., Sanderson, Eleanor, Hammerton, Gemma, Richmond, Rebecca C., Davey Smith, George, Heron, Jon, Taylor, Amy E., Davies, Neil M., Howe, Laura D.·2021·European Journal of Epidemiology
DOI 10.1007/s10654-021-00757-1PMC PMC8159796

Published version of a preprint

Open this paper in CoChat — free accountOpen PDF

Abstract

Mediation analysis seeks to explain the pathway(s) through which an exposure affects an outcome. Traditional, non-instrumental variable methods for mediation analysis experience a number of methodological difficulties, including bias due to confounding between an exposure, mediator and outcome and measurement error. Mendelian randomisation (MR) can be used to improve causal inference for mediation analysis. We describe two approaches that can be used for estimating mediation analysis with MR: multivariable MR (MVMR) and two-step MR. We outline the approaches and provide code to demonstrate how they can be used in mediation analysis. We review issues that can affect analyses, including confounding, measurement error, weak instrument bias, interactions between exposures and mediators and analysis of multiple mediators. Description of the methods is supplemented by simulated and real data examples. Although MR relies on large sample sizes and strong assumptions, such as having strong instruments and no horizontally pleiotropic pathways, our simulations demonstrate that these methods are unaffected by confounders of the exposure or mediator and the outcome and non-differential measurement error of the exposure or mediator. Both MVMR and two-step MR can be implemented in both individual-level MR and summary data MR. MR mediation methods require different assumptions to be made, compared with non-instrumental variable mediation methods. Where these assumptions are more plausible, MR can be used to improve causal inference in mediation analysis.

Overview

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1007/s10654-021-00757-1

Cites

70 of 70 references verified0 citation errors · 0 not checkable

Statistics

Figures

Fig. 1 The decomposed effects in a non-IV regression-based mediation analysis where c represents the total effect, c' represents the direct effect and the indirect effect can be calculated by subtracting c’ from c (difference method) or multiplying A times B (product of coefficients method) b multivariable Mendelian randomisation, using a combined genetic instrument for both the exposure and mediator of interest, to estimate the direct effect c' of the exposure and c two-step Mendelian randomisation, where the effect of the exposure on the mediator (A) and mediator on the outcome b are estimated separately, using separate genetic instrumental variables for both the exposure and mediator. These estimates are then multiplied together to estimate the indirect effect of the mediator (A*B)
Fig. 1 The decomposed effects in a non-IV regression-based mediation analysis where c represents the total effect, c' represents the direct effect and the indirect effect can be calculated by subtracting c’ from c (difference method) or multiplying A times B (product of coefficients method) b multivariable Mendelian randomisation, using a combined genetic instrument for both the exposure and mediator of interest, to estimate the direct effect c' of the exposure and c two-step Mendelian randomisation, where the effect of the exposure on the mediator (A) and mediator on the outcome b are estimated separately, using separate genetic instrumental variables for both the exposure and mediator. These estimates are then multiplied together to estimate the indirect effect of the mediator (A*B)CC BY · PubMed Central
Fig. 2 Schematic diagram illustrating the causal assumptions (dashed lines) in a non-IV regression-based mediation methods and b Mendelian randomisation mediation analysis with the measured associations in solid black lines. Additional assumptions: in non-IV mediation there is no measurement error in the exposure or mediator; in Mendelian randomisation mediation there is no exposure-mediator interaction. In Mendelian randomisation, the exclusion restriction criteria mean there are no alternative pathways from the instrument to the outcome other than via the exposure (or mediator) of interest
Fig. 2 Schematic diagram illustrating the causal assumptions (dashed lines) in a non-IV regression-based mediation methods and b Mendelian randomisation mediation analysis with the measured associations in solid black lines. Additional assumptions: in non-IV mediation there is no measurement error in the exposure or mediator; in Mendelian randomisation mediation there is no exposure-mediator interaction. In Mendelian randomisation, the exclusion restriction criteria mean there are no alternative pathways from the instrument to the outcome other than via the exposure (or mediator) of interestCC BY · PubMed Central
Fig. 3 Size of absolute bias for the indirect effect of an exposure on a continuous outcome, rare binary outcome and common binary outcome through a continuous mediator, for a range of fixed true total effect sizes (0.2, 0.5 and 1.0) and range of true indirect effect sizes using non-IV regression based mediation methods or Mendelian randomisation, on the relative scale (simulated N = 5000). In all scenarios, unmeasured confounding is simulated
Fig. 3 Size of absolute bias for the indirect effect of an exposure on a continuous outcome, rare binary outcome and common binary outcome through a continuous mediator, for a range of fixed true total effect sizes (0.2, 0.5 and 1.0) and range of true indirect effect sizes using non-IV regression based mediation methods or Mendelian randomisation, on the relative scale (simulated N = 5000). In all scenarios, unmeasured confounding is simulatedCC BY · PubMed Central
Fig. 4 Flow chart for analytical processes when carrying out mediation analyses using individual level Mendelian randomisation
Fig. 4 Flow chart for analytical processes when carrying out mediation analyses using individual level Mendelian randomisationCC BY · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Show all 67 references

Research works printed but not linked: 1

  1. Richardson TG, Sanderson E, Elsworth B, Tilling K, Davey SG. Use of genetic variation to separate the effects of early and later life adiposity on disease risk: mendelian randomisation study. BMJ. 2020;369:m1203. https://doi.org/10.1136/bmj.m1203. · DOI · Paper page

Cited by

Sorted by citations

Items 501–600 of 1,001

Related

See all

Metadata

Authors
Carter, A. R., Sanderson, E., Hammerton, G., Richmond, R. C., Davey Smith, G., Heron, J., Taylor, A. E., Davies, N. M., & Howe, L. D.
Affiliations
  • University of Bristol (GB)
  • University Hospitals Bristol NHS Foundation Trust (GB)
  • Norwegian University of Science and Technology (NO)
Corresponding author
Carter, Alice R.
ORCID
Venue
European Journal of Epidemiology
Publisher
Springer Science and Business Media LLC
Published
2021-05
Volume
36
Issue
5
Pages
465-478
ISSN
  • 0393-2990 (ISSN-L)
  • 1573-7284
DOI
10.1007/s10654-021-00757-1
PMC
PMC8159796
OpenAlex
W3159102563
Cited by
1,215
References
67
Citations by year
2026: 196 · 2025: 407 · 2024: 386 · 2023: 143 · 2022: 54 · 2021: 18
FWCI
117.42
Type
journal article
Language
en
Open access
Licence
Locations
  • European Journal of Epidemiology (publishedVersion, cc-by)
  • bioRxiv (Cold Spring Harbor Laboratory) (acceptedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • Bristol Research (University of Bristol) (submittedVersion, cc-by)
Funders
  • UK Medical Research Council (MC_UU_00011/1, MR/M020894)
  • UK Economic and Social Research Council (ES/N000757/1)
  • Wellcome Trust (209138/Z/17/Z)
  • Norwegian Research Council (295989)
Topics
  • Physical Sciences › Mathematics › Statistics and Probability › Advanced Causal Inference Techniques
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Genetics › Genetic Associations and Epidemiology
  • Physical Sciences › Mathematics › Statistics and Probability › Statistical Methods in Epidemiology
Keywords
Mendelian randomization, mediation analysis, causal inference, confounding, measurement error, weak instrument bias, pleiotropy, multiple mediators
MeSH terms
Mediation Analysis, Humans, Genetic Variation, Bias, Causality, Genome-Wide Association Study, Mendelian Randomization Analysis, Genetic Pleiotropy

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review