Pulse Brain · Growing Health Evidence Index
Tier 3 — Observational / field trialPeer-reviewed

Improving the accuracy of two-sample summary-data Mendelian randomization: moving beyond the NOME assumption

Jack Bowden, Fabiola Del Greco M, Cosetta Minelli, Qingyuan Zhao, Debbie A. Lawlor, Nuala A. Sheehan, John R. Thompson, George Davey Smith

International Journal of Epidemiology · 2018

Read source ↗ All evidence

Summary

This methodological paper addresses fundamental statistical challenges in two-sample summary-data Mendelian randomization by critically examining how genetic variants are weighted when assessing causality. The authors demonstrate that commonly used 'first-order' weights risk false-positive heterogeneity detection, whilst 'second-order' weights risk false negatives, and they propose modified weights that improve accuracy in heterogeneity quantification and eliminate regression dilution bias. The contribution is illustrated using systolic blood pressure and coronary heart disease risk, with implications for the reliability of causal inference in epidemiological studies.

UK applicability

As a methodological advance in epidemiological causal inference techniques, this work is applicable to UK-based and international epidemiological research that uses Mendelian randomization to examine public health questions. The improved weighting approach enhances the statistical rigour of causal claims that may inform UK health policy and research priorities.

Key measures

Heterogeneity quantification, inverse-variance weights, regression dilution bias, type I error rate, estimate precision, power to detect causal effects, causal effect estimates

Outcomes reported

The study evaluated the accuracy of weighting schemes used in two-sample summary-data Mendelian randomization (MR) for detecting heterogeneity among genetic variants and quantifying causal effects. The authors developed modified weights to improve heterogeneity detection and reduce regression dilution bias compared to existing first- and second-order weighting approaches.

Theme
Measurement & metrics
Subject
Measurement methods & nutrient profiling
Study type
Research
Study design
Methodological study with Monte Carlo simulations and empirical application
Source type
Peer-reviewed study
Status
Published
Geography
International
System type
Other
DOI
10.1093/ije/dyy258
Catalogue ID
BFmokjo8sc-de6m3u

Topic tags

Pulse AI · ask about this record

Dig deeper with Pulse AI.

Pulse AI has read the whole catalogue. Ask about this record, its theme, or how the findings apply to UK farming and policy — every answer cites the underlying studies.