Follow us on twitter Follow us on X

Image: Coloured TEM of an islet of Langerhans cell. Credit: Science Photo Library

Guidelines for Mendelian randomisation studies

Authors who plan to submit a Mendelian randomisation study to be considered for publication in Diabetologia, and reviewers and editors examining these submissions, should follow the guidelines below, which are outlined in more detail in our 2025 editorial, Corbin, L.J. et al Raising the bar for publication of Mendelian randomisation studies in Diabetologia. Diabetologia (2025).

In particular, the following guidance should be followed.

  1. All submissions should follow STROBE-MR guidelines [1] and be accompanied by a completed checklist that must be scrutinised by editors and reviewers for accuracy and completeness.
  2. A clear rationale should be provided for the MR analysis, including how the study aims to meaningfully advance existing understanding and the justification of the gene–environment equivalence [2].
  3. The strongest consideration will be given to those submissions in which MR is presented within a triangulation of evidence framework including at least one other approach with different key sources of potential bias. Comparisons may be made between results from MR and those from different study designs (e.g. RCTs, cohort or cross-sectional studies, wet lab experiments), and/or between results from MR and those from different analytical approaches within the same (cohort) study design (e.g. negative control studies, cross-context comparisons) [3, 4]. Whilst additional analyses such as colocalisation, reverse MR and functional enrichment may be used to support and contextualise MR findings, they are generally not considered to be independent approaches.
  4. Submissions that consist solely of MR analyses using previously published data (i.e. GWAS summary statistics) and existing methods are unlikely to meet the journal’s novelty requirements. Many of these results most likely already exist in the public domain [5, 6].
  5. As one of the most important aspects of any MR study, instrument selection should be clearly described, including the rationale for any pre-filtering of variants (e.g. linkage-disequilibrium based clumping). The exclusion of certain instruments based on specific properties such as F-statistics should generally be avoided [7].
  6. Instrumenting behavioural exposures such as diet and exercise is particularly problematic due to the time-varying, compositional and intercorrelated nature of these traits. To be considered for publication, studies involving such traits should begin with well-defined exposures, well-established links to the genetics and clear causal hypotheses [8].
  7. Where multiple MR methods are applied to the same data, authors should state a priori which is the primary analysis and why. Whilst concordance in effect estimates across multiple MR methods may increase confidence in the robustness of the analysis, it does not necessarily constitute independent evidence. Authors should discuss the extent to which results from different analyses can be considered statistically independent.
  8. We recognise that null effects from MR can be informative, in particular when they contrast with results from alternative analytical approaches, e.g. observational analyses. Whilst power calculations can be useful for providing additional context for null findings, unsupported and vague statements suggesting that ‘lack of power’ may be the reason for null results should be avoided.
  9. New MR methods should be rigorously evaluated before implementation including, for example, the use of negative and positive controls (e.g. [9]).
  10. All analytical code should be annotated and shared at the time of submission via GitHub or an equivalent platform, and details of how to access it should be included in a ‘Code availability statement’ in your article.

For further guidance on the robust application and evaluation of MR studies we recommend the following articles: [10, 11,12,13,14,15,16].

References

  1. Skrivankova VW, Richmond RC, Woolf BAR et al (2021) Strengthening the reporting of observational studies in epidemiology using Mendelian randomization: the STROBE-MR statement. JAMA 326(16):1614–1621. https://doi.org/10.1001/jama.2021.18236
  2. Ebrahim S, Davey Smith G (2008) Mendelian randomization: can genetic epidemiology help redress the failures of observational epidemiology? Hum Genet 123(1):15–33. https://doi.org/10.1007/s00439-007-0448-6
  3. Lawlor DA, Tilling K, Davey Smith G (2016) Triangulation in aetiological epidemiology. Int J Epidemiol 45(6):1866–1886. https://doi.org/10.1093/ije/dyw314
  4. Gutierrez S, Glymour MM, Smith GD (2025) Evidence triangulation in health research. Eur J Epidemiol https://doi.org/10.1007/S10654-024-01194-6
  5. Hemani G, Bowden J, Haycock P et al (2017) Automating Mendelian randomization through machine learning to construct a putative causal map of the human phenome. bioRxiv 173682. https://doi.org/10.1101/173682
  6. Liu Y, Elsworth B, Erola P et al (2021) EpiGraphDB: a database and data mining platform for health data science. Bioinformatics 37(9):1304–1311. https://doi.org/10.1093/bioinformatics/btaa961
  7. Burgess S, Thompson SG, CRP CHD Genetics Collaboration (2011) Avoiding bias from weak instruments in Mendelian randomization studies. Int J Epidemiol 40(3):755–764. https://doi.org/10.1093/IJE/DYR036
  8. Merino J, Tobias DK (2022) The unique challenges of studying the genetics of diet and nutrition. Nat Med 28(2):221–222. https://doi.org/10.1038/s41591-021-01626-w
  9. Hamilton FW, Hughes DA, Spiller W, Tilling K, Davey Smith G (2024) Non-linear Mendelian randomization: detection of biases using negative controls with a focus on BMI, Vitamin D and LDL cholesterol. Eur J Epidemiol 39(5):451–465. https://doi.org/10.1007/s10654-024-01113-9
  10. Sanderson E, Glymour MM, Holmes MV et al (2022) Mendelian randomization. Nat Rev Methods Primers 2(1):6. https://doi.org/ 10.1038/s43586-021-00092-5
  11. Skrivankova VW, Richmond RC, Woolf BAR et al (2021) Strengthening the reporting of observational studies in epidemiology using mendelian randomisation (STROBE-MR): explanation and elaboration. BMJ 375:n2233. https://doi.org/10.1136/bmj.n2233
  12. Lawlor DA, Wade K, Borges MC et al (2019) A Mendelian Randomization dictionary: useful definitions and descriptions for undertaking, understanding and interpreting Mendelian Randomization studies. OSF Preprints. https://doi.org/10.31219/osf.io/6yzs7
  13. Davies NM, Holmes MV, Davey Smith G (2018) Reading Mendelian randomisation studies: a guide, glossary, and checklist for clinicians. BMJ 362:k601. https://doi.org/10.1136/bmj.k601
  14. Burgess S, Davey Smith G, Davies NM et al (2023) Guidelines for performing Mendelian randomization investigations: update for summer 2023. Wellcome Open Res 4(186). https://doi.org/10.12688/wellcomeopenres.15555.3
  15. Stender S, Gellert-Kristensen H, Smith GD (2024) Reclaiming mendelian randomization from the deluge of papers and misleading findings. Lipids Health Dis 23(1):286. https://doi.org/10.1186/s12944-024-02284-w
  16. Kjaergaard AD, Smith GD, Stewart P (2023) Mendelian randomization studies in endocrinology: raising the quality bar for submissions and publications in the Journal of Clinical Endocrinology & Metabolism. J Clin Endocrinol Metab 109(1):1–3. https://doi.org/10.1210/CLINEM/DGAD569
Top