CoChat
Sign Up
Cited by 1,648Meta-analysis and systematic reviewsStatistical Methods in EpidemiologyAdvanced Causal Inference Techniques

The DOI matches its Crossref record; 138 of the 138 references it cites were found in a registry or an index; none of the 101 sources its claims cite is checked for support yet.

Strengthening the Reporting of Observational Studies in Epidemiology using Mendelian Randomisation (STROBE-MR): explanation and elaboration.

Open access
Skrivankova, Veronika W, Richmond, Rebecca C, Woolf, Benjamin A R, Davies, Neil M, Swanson, Sonja A, VanderWeele, Tyler J, Timpson, Nicholas J, Higgins, Julian P T, Dimou, Niki, Langenberg, Claudia, Loder, Elizabeth W, Golub, Robert M, Egger, Matthias, Davey Smith, George, Richards, J Brent·2021·BMJ
DOI 10.1136/bmj.n2233PMC PMC8546498
Open this paper in CoChat — free accountOpen PDF

Abstract

Mendelian randomisation (MR) studies allow a better understanding of the causal effects of modifiable exposures on health outcomes, but the published evidence is often hampered by inadequate reporting. Reporting guidelines help authors effectively communicate all critical information about what was done and what was found. STROBE-MR (strengthening the reporting of observational studies in epidemiology using mendelian randomisation) assists authors in reporting their MR research clearly and transparently. Adopting STROBE-MR should help readers, reviewers, and journal editors evaluate the quality of published MR studies. This article explains the 20 items of the STROBE-MR checklist, along with their meaning and rationale, using terms defined in a glossary. Examples of transparent reporting are used for each item to illustrate best practices.

Overview

AI snapshot

Written from the paper’s full text.

TL;DR

STROBE-MR is a 20-item reporting checklist for Mendelian randomisation (MR) studies, and this article explains each item and provides examples of transparent reporting. It aims to help authors, reviewers, and journal editors communicate and appraise MR studies more clearly.

Key findings

  • STROBE-MR comprises 20 reporting items covering title/abstract, introduction, methods, results, discussion, and other information; it was developed by an expert group and revised after a July 2019 preprint and debate at the 4th International Mendelian Randomisation Conference.
  • The paper states that empirical evidence indicates many MR reports do not clearly state or examine MR assumptions and report insufficient details on data sources, making quality and reliability difficult to evaluate.
  • This explanation and elaboration document explains the rationale of each checklist item and uses published MR studies as examples of transparent reporting; the examples are not necessarily ideal but highlight the intended issue for each item.
  • The authors encourage MR study authors, reviewers, and journal editors to use STROBE-MR to improve reporting.

Methods

  • Guideline development: the initiative was established in 2018 following guidance for developing medical research reporting guidelines; an expert group met face-to-face in Bristol, UK, over two days in May 2019 to discuss empirical evidence and draft checklist items.
  • The draft checklist was published as a preprint in July 2019, debated on the preprint platform, social media, and at the 4th International Mendelian Randomisation Conference, then revised.
  • This article is an explanation and elaboration document complementing the STROBE-MR statement; it explains the 20 checklist items using a glossary and examples identified from published MR studies, with some examples edited by removing citations and unrelated sections.

Limitations

  • The paper states that STROBE-MR is not intended as a formal tool for assessing methodological or reporting quality and should not be transformed into a quality scale; it is also not a formal guideline to design and conduct MR studies.
  • The paper states that some items will not apply to MR studies that do not use an instrument for the exposure or do not report instrumental variable estimates, though the checklist still provides useful guidance.
  • The paper states that the examples do not necessarily represent the ideal statement for each checklist item; evident from the text, this is a consensus/guideline explanation and elaboration without reported empirical validation of whether STROBE-MR improves reporting quality.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1136/bmj.n2233

Cites

138 of 138 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Verification

Claims the paper makes

0 of 101 cited sources checked · 0 supported · 0 flagged

Show all 48 claims

  • It may state two separate claims as one. 38
  • Probably not a substantial claim. 103
  • The quote only partly states this claim. 1
120 of 120

  • The paper relies on this without support.

    Reporting guidelines help authors effectively communicate all critical information about what was done and what was found.

    “Reporting guidelines help authors effectively communicate all critical information about what was done and what was found.”

  • The paper relies on this without support.

    Genetic variation is essentially randomly inherited from parents to offspring at conception.

    “Genetic variation is essentially randomly inherited from parents to offspring at conception”

  • The paper relies on this without support.

    Mendelian randomisation studies have increased in popularity and have begun to inform understanding of disease causation.

    “Given these advantages, MR studies have increased in popularity and have begun to inform understanding of disease causation.”

  • The authors state this as an assumption.

    Relevance: genetic variant G is associated with the exposure of interest X.

    “I. Relevance: genetic variant G is associated with the exposure of interest X”

  • The authors state this as an assumption.

    Independence: genetic variant G shares no unmeasured cause with outcome Y.

    “II. Independence: genetic variant G shares no unmeasured cause with outcome Y”

  • The authors state this as an assumption.

    Exclusion restriction: genetic variant G does not affect outcome Y except through its potential effect on the exposure of interest X.

    “III. Exclusion restriction: genetic variant G does not affect outcome Y except through its potential effect on the exposure of interest X.”

  • The authors state this as an assumption.

    The exposure has the same effects in everyone (homogeneity).

    “The homogeneity assumption requires that the exposure has the same effects in everyone, which is not directly verifiable.”

  • The authors state this as an assumption.

    An increase in the number of risk alleles does not lower the likelihood of exposure for any individuals (monotonicity).

    “The homogeneity assumption can also be replaced by imposing a monotonicity assumption that an increase in the number of risk alleles does not lower the likelihood of exposure for any individuals, typically leading to estimating an effect in a subgroup of the study population.61”

  • The authors state this as an assumption.

    The size of the direct effects of the genetic variants on the outcome that do not operate through the exposure are independent of the size of the genetic variants’ effects on the exposure (InSIDE).

    “MR Egger regression relaxes the exclusion restriction assumption but imposes an InSIDE assumption that the size of the direct effects of the genetic variants on the outcome that do not operate through the exposure are independent of the size of the genetic variants’ effects on the exposure.”

  • The authors state this as an assumption.

    The association between the genetic variants and the exposure is the same in the two samples (two sample MR).

    “Additionally, the two sample MR approach assumes that the association between the genetic variants and the exposure is the same in the two samples, which might not hold if samples are selected from different subpopulations (eg, by sex, age, ethnicity).”

  • The paper relies on this without support.

    Original data are needed by readers or researchers who wish to evaluate or replicate analyses.

    “Original data are needed by readers or researchers who wish to evaluate or replicate analyses.”

  • The paper relies on this without support.

    Many funders and journals encourage or require data sharing, and provide guidance to authors about the content of any explicit data sharing statement required by their journal.

    “Many funders and journals encourage or require data sharing, and provide guidance to authors about the content of any explicit data sharing statement required by their journal.”
12 of 12

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.

Each contradiction between two passages listed here was confirmed by an independent check.

  • Replicated odds ratio for rs4233949 falls outside its confidence interval

    Minor

    The replication-stage odds ratio is 1.04, but its stated 95% confidence interval runs from 1.05 to 1.05.

  • Confidence interval for the BMI allele score association changes between examples

    Minor

    The two examples report the same −0.06% association with 25(OH)D concentrations and P=0.004, but give different lower confidence limits: −0.20% and −0.10%.

  • Replicated odds ratio for rs7851693 falls outside its confidence interval

    Minor

    The replication-stage odds ratio is 1.05, but its stated 95% confidence interval runs from 1.06 to 1.06.

  • One SNP has more fracture cases than the stated combined total

    Minor

    Table 8 lists 185 085 combined fracture cases for rs1286083, exceeding the table note's combined total of 185 057 cases by 28.

4 of 4

Citations

Cites

Flagged first, then by citations

Research works printed but not linked: 1

  1. Conference MR. 2019. MRC Integrative Epidemiology Unit. University of Bristol. 2020. https://www.bristol.ac.uk/integrative-epidemiology/seminars/2019/mr-conference-2019.html

Cited by

Sorted by citations

See all 1,648

Related

See all

Metadata

Authors
Skrivankova, V. W., Richmond, R. C., Woolf, B. A. R., Davies, N. M., Swanson, S. A., VanderWeele, T. J., Timpson, N. J., Higgins, J. P. T., Dimou, N., Langenberg, C., Loder, E. W., Golub, R. M., Egger, M., Davey Smith, G., & Richards, J. B.
Affiliations
  • University of Bern (CH)
  • University of Bristol (GB)
  • Norwegian University of Science and Technology (NO)
  • Erasmus MC (NL)
  • Harvard University (US)
  • Centre international de recherche sur le cancer (FR)
  • University of Cambridge (GB)
  • Northwestern University (US)
  • University of London (GB)
Corresponding author
Richards, J Brent
ORCID
Venue
BMJ
Publisher
BMJ
Published
2021-10-26
Volume
375
Pages
n2233
ISSN
1756-1833 (ISSN-L)
DOI
10.1136/bmj.n2233
PMC
PMC8546498
OpenAlex
W3210406414
Semantic Scholar
99c3b9a1837c0cd16d8ee30fe55332a366bfbbf6
Cited by
1,648
References
137
Citations by year
2026: 217 · 2025: 484 · 2024: 599 · 2023: 252 · 2022: 91 · 2021: 4
FWCI
140.35
Type
journal article
Language
en
Open access
Licence
Locations
  • BMJ (publishedVersion, cc-by)
  • Europe PMC (PubMed Central) (publishedVersion, other-oa)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • EUR Research Repository (Erasmus University Rotterdam) (publishedVersion, cc-by)
  • EUR Research Repository (Erasmus University Rotterdam) (submittedVersion, cc-by)
  • BORIS (University Library Bern) (submittedVersion, cc-by)
  • Apollo (University of Cambridge) (submittedVersion, cc-by)
  • Apollo
  • Apollo (cc-by)
Funders
  • National Science Foundation (189498)
  • World Health Organization
  • Wellcome Trust (WT 217065/Z/19/Z, 217065/Z/19/Z, 202802/Z/16/Z, 217065, 202802)
  • Public Health Agency of Canada
  • University Hospitals Bristol NHS Foundation Trust
  • Compute Canada
  • NIHR Bristol Biomedical Research Centre (MC_UU_00011/1, BRC‐1215‐2001, NF-SI-0617-10145)
  • Cancer Research UK (C18281/, 29019, C18281 / A19169, 10710, MC_UU_00011/1, 10118, 16563, A19169, 19169)
  • National Institute for Health and Care Research (BRC-1215-2001, C18281/A19169, 217065/Z/19/Z, NF-SI-0617-10145, MC_UU_00011/1)
  • Department of Health and Social Care
  • King's College London
  • European Commission (633784)
  • University of Bristol (C18281/A19169, 217065/Z/19/Z, MC_UU_00011/1)
  • Public Health Agency
  • Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung (189498, 10145)
  • ZonMw (91617066)
  • Nederlandse Organisatie voor Wetenschappelijk Onderzoek (91617066)
  • Jewish General Hospital
  • Norges Forskningsråd (189498, 295989)
  • National Institutes of Health (217065/Z/19/Z)
  • Canadian Institutes of Health Research (365825; 409511, 40951_1, 365825)
  • Medical Research Council (MC_UU_00011/6, 217065/Z/19/Z, MR/M005070/1, 202802/Z/16/Z, 217065, WT 217065/Z/19/Z, MC_UU_00011/7, MC_UU_00011/5, MC_UU_00011/1, MC_UU_00011, MC_UU_00011/2)
  • Economic and Social Research Council (+3 PhD Studentship Award, ES/P000630/1, 1+3 PhD Studentship, ES/N000498/1)
  • National Cancer Institute (R01CA222147)
Topics
  • Social Sciences › Decision Sciences › Statistics, Probability and Uncertainty › Meta-analysis and systematic reviews
  • Physical Sciences › Mathematics › Statistics and Probability › Statistical Methods in Epidemiology
  • Physical Sciences › Mathematics › Statistics and Probability › Advanced Causal Inference Techniques
Keywords
reporting guidelines, Mendelian randomization, observational studies, causal inference, reporting checklist, transparency in research reporting, study quality assessment
MeSH terms
Humans, Epidemiologic Research Design, Guidelines as Topic, Mendelian Randomization Analysis, Observational Studies as Topic

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.

A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

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