Ask questions across claims linked to experiments, exact results, limitations, and sources.Know what falsifiable science actually supports before you trust the answer.
Press Enter β΅ to search evidence
Explore by Goal
"The diversity of the phenomena of nature is so great, and the treasures hidden in the heavens so rich, precisely in order that the human mind shall never be lacking in fresh nourishment."
- Johannes Kepler
Quick Explanation
Copied
This two-sample Mendelian randomization study reports that genetically predicted heavy manual work increases atrial fibrillation risk (IVW OR 1.787, 95% CI 1.082β2.853, P=0.023) while job satisfaction is protective (OR 0.719, 0.536β0.964, P=0.028), with null results for walking/standing work and both short and long sleep . Findings are suggestive only β borderline P values, modest instruments (11, 11, 5, 13, 3 SNPs), and conflicting prior MR evidence on sleep temper causal claims.
Long Explanation
Study Design and Reported Results
Wu et al. performed single-factor, two-sample MR with five genetically predicted exposures (three work factors from MRC-IEU/UK Biobank, two sleep traits from the Sleep Disorders Knowledge Portal) against atrial fibrillation in FinnGen (261,395 participants; 50,743 AF cases), all European ancestry . Only two of five associations reached nominal significance:
Strengths and Internal Consistency
For heavy manual work, sensitivity analyses supported the point estimate: Cochran's Q showed no heterogeneity (Q=14.979, P=0.133), MR-PRESSO outlier removal preserved the effect (pleiotropy P=0.146), MR-Egger intercept was null (P=0.336), and weighted median agreed (OR 1.879, P=0.031) . Job satisfaction showed the same pattern (OR 0.719, 95% CI 0.536β0.964, P=0.028; weighted median OR 0.817, non-significant). The authors exclude confounder-associated SNPs (blood pressure, coronary disease, obesity, diabetes, alcohol, heart failure, hyperthyroidism, valvular disease), a step many MR papers omit.
Critical Caveats
Nominal significance only. Five exposures tested at P<0.05 with no stated multiple-comparison correction; both βpositiveβ findings have CIs close to the null.
Few instruments. Only 3 SNPs for long sleep and 5 for walking/standing work; long-sleep CIs span 0.14β63, indicating near-zero power, so βno associationβ here means βundetected.β
Sleep results contradict prior MR. The authors themselves acknowledge Zhao et al. found short sleep increased AF risk, attributing divergence to LD thresholds, outcome databases, and IVW model choice . This methodological sensitivity itself is a red flag for fragile instruments.
Self-reported exposures (touchscreen questionnaire categories like βjob involves heavy manual workβ) may misclassify the actual exposure, and Winner's curse risk is elevated with weak, few instruments.
European-only generalizability; the authors note AF burden differs across ethnic groups. The null sleep finding conflicts with a Chinese prospective cohort reporting long sleep as an AF risk factor, underscoring ancestry/context dependence.
Job satisfaction interpretation is speculative. The depression/autonomic-dysfunction mediation chain is a hypothesis, not tested via mediation MR in this paper.
Verdict
The heavy-manual-work finding is biologically plausible (consistent with the Copenhagen City Heart Study observation cited by the authors) and internally coherent, but the evidence is suggestive, not definitive. A falsifying replication would be a well-powered MR using finer-grained occupational phenotypes and multiple AF consortia showing no heavy-work effect. What would strengthen the field: multivariable MR jointly modeling work and sleep factors, and non-European replication.
Feedback:
Updated: November 17, 2026
BGPT Paper Review
Study Novelty
60%
Extends the MR framework to occupational phenotypes (heavy manual work, job satisfaction) against AF, a relatively unexplored exposure-outcome pair; sleep-AF MR already existed, so the sleep portion is largely confirmatory and contradicts prior work.
Scientific Quality
50%
Standard TwoSampleMR workflow with reasonable sensitivity analyses and confounder-SNP screening, but nominal-only significance across five uncorrected tests, tiny instrument sets (3β13 SNPs), self-reported binary exposures, unstable sleep MR-Egger estimates (CI 0.003β45241), and no pre-registration stated. Exposure GWAS accessed βMay 2025β for a paper accepted March 2025 is an oddity.
Study Generality
60%
Addresses modifiable lifestyle/occupational risk factors for a common arrhythmia, relevant beyond the specific exposures, but limited by European ancestry and coarse categorical phenotypes.
Study Usefulness
50%
Adds genetic evidence that work characteristics may influence AF risk, potentially informing occupational health prevention, but effect sizes are imprecise and findings require replication before practical application.
Study Reproducibility
70%
All GWAS summary data are public (MRC-IEU, Sleep Disorders Knowledge Portal, FinnGen), software versions stated (TwoSampleMR 0.5.8, R 4.3.2), and thresholds reported, enabling independent replication; no code repository is shared.
Explanatory Depth
50%
Mechanistic discussion (left atrial dilation, atrial fibrosis, autonomic dysfunction via depression) is plausible but speculative and not formally tested (no mediation MR); the paper is primarily a causal-association exercise.
We'll email you the results when your analysis is finished.
Hypothesis Graveyard
Short sleep causally increases AF risk via sympathetic activation β not supported here despite prior MR support; conflicting results across studies with different SNP screens suggest the original finding may reflect pleiotropy or database heterogeneity, though it cannot be discarded.
Long sleep is a strong AF risk factor β the 3-SNP instrument (CI 0.14β63) is uninformative; observational Chinese cohort findings likely reflect reverse causation (illness extending sleep) rather than a causal effect.