Cohort studyindependent cross-checkfull text checked★ key study I-WHT-07 in-depth review
Gluten and heart disease in people without coeliac disease
Long term gluten consumption in adults without celiac disease and risk of coronary heart disease: prospective cohort study.
Overview
Lebwohl B et al.. BMJ 2017 · PMID 28465308 · DOI 10.1136/bmj.j1892
Eating more gluten was not linked to more heart disease over 26 years.
Does the study support the specific claim? No · Methodological quality: high
Plain-language summary
Study explained simply
- What did the researchers want to find out?
Gluten and CHD
- How was the study done?
Cohort
- How many people, animals or samples were studied?
110,017
- What was actually found?
No association
- How large was the observed effect?
HR 0.95
- What do the results mean in everyday life?
Gluten is not a heart risk for non-coeliacs
- What are the study's limitations?
FFQ estimate
- What does it say about Elawady's specific claim?
Contradicts general gluten harm
- Do the results apply to Germany and Europe?
Moderate
- What can explicitly not be concluded from this study?
Says nothing about gut permeability, intestinal symptoms or whether gluten sensitivity exists; only about heart disease
Scientific in-depth review
1. Full reference
- Authors
- Lebwohl B et al.
- Journal
- BMJ
- Year
- 2017
- Identifier
- PMID 28465308 DOI 10.1136/bmj.j1892
- Verification
- verified – PubMed efetch 2026-10-25
- Retracted
- no
- Corrections
- none
- Extent checked
- full text
- Conflicts of interest and funding
AGA foundation research scholar award (BL); MGH research scholars award and NIH grants K24 DK098311, UM1 CA186107 (ATC); NIH UM1 CA167552, R01 HL035464, R01 HL060712, R01 HL034594 for other authors. ICMJE: no other support, no relevant financial relationships.
2. Research question
Gluten intake and CHD in people without coeliac disease
3. Study design and population
- Study design
- Cohort study – Prospective cohort (NHS, HPFS)
- Population
64,714 female nurses (NHS) and 45,303 male health professionals (HPFS) without CHD, stroke, CABG, cancer or self-reported coeliac disease, with plausible energy intake on the 1986 FFQ
- Country
USA
- Period
1986 to 2012 (June 2012 NHS; January 2012 HPFS); 2,273,931 person-years; diet assessed 1986-2010 every 4 years
- Sample size
110,017; 6529 CHD cases
- Inclusion and exclusion criteria
Excluded implausible energy (<600/>3500 kcal women; <800/>4200 kcal men), missing gluten data (48; 39), prior MI/angina/stroke/CABG (4015; 2647), cancer (4689; 1785), self-reported coeliac disease asked in 2014 (200; 160).
4. Methods and endpoints
- Exposure / intervention
Energy-adjusted estimated gluten intake (75% of protein from wheat, rye, barley; oats and condiments not counted) as cumulative average from repeated FFQs; highest fifth mean 7.5 g/day (women) and 10.0 g/day (men) vs lowest fifth 2.6 and 3.3 g/day at baseline
- Comparator
Lowest fifth
- Endpoints
Incident CHD: fatal or non-fatal myocardial infarction confirmed by medical records (WHO criteria), death records or autopsy (definite + probable)
- Statistical methods
Cox models conditioned on age and follow-up cycle, cohort-specific then pooled (stratified by cohort); a priori multivariable models incl. BMI, diabetes, medications, smoking, physical activity and dietary covariates; diet updates suspended after diabetes, CVD or cancer; secondary models adjusting for refined or whole grains; subgroup, tenths, time-period and post-hoc dietary sensitivity analyses; Bonferroni for smoking strata.
5. Actual results
- Results
CHD incidence 352 (lowest fifth) vs 277 (highest fifth) per 100,000 person-years; unadjusted rate difference 75 (51 to 98) fewer cases per 100,000 person-years. Age-adjusted pooled HR 0.87 (0.80-0.93); after non-dietary covariates 0.98 (0.91-1.06); full model 0.95 (0.88-1.02), P trend 0.29; per 1 g/day 0.99 (0.98-1.01). Further adjusted for whole grains 1.00 (0.92-1.09); for refined grains 0.85 (0.77-0.93), P trend 0.002. Current smokers: 1.34 (1.09-1.66), not significant after refined-grain adjustment (1.25, 0.95-1.64) or Bonferroni correction. Null for fatal and non-fatal MI separately.
- Absolute risks
352 vs 277 CHD events per 100,000 person-years (lowest vs highest fifth, unadjusted); rate difference 75 (95% CI 51 to 98) per 100,000 person-years
- Authors' conclusion
Gluten not associated with CHD; gluten avoidance may reduce whole grain intake; gluten-free diets should not be promoted for people without coeliac disease.
6. Strengths
- Large cohorts with 26 years of follow-up and repeated diet assessment (cumulative averages)
- Medically confirmed CHD outcomes
- Diet updates suspended after major diagnoses to limit reverse causation
- A priori secondary analyses separating refined and whole-grain variance; multiple-testing correction for subgroup
7. Weaknesses
- Gluten estimated indirectly from FFQ with a fixed 75% protein conversion; not validated directly
- Could not identify people following gluten-free diets or eating gluten-free substitutes; undiagnosed coeliac disease not excluded
- Residual confounding possible (observational)
- US health professionals, mostly white; generalisability to other diets limited
- Outcome limited to CHD; no data on gut, permeability, symptoms or NCGS
8. Risk of bias
- Orienting tool
- ROBINS-E
- Type of assessment
- structured assessment based on the full text by a single reviewer (no formal double review)
- Full formal assessment
- no
- Overall risk of bias
- some concerns
- Rationale
Overall unchanged. Simplified single-reviewer judgement from the full text; a formal ROBINS-E assessment would need two independent reviewers.
| Domain | Judgement | Rationale |
|---|---|---|
| Confounding | some concerns | Extensive adjustment; gluten tracks whole and refined grain intake, making results model-dependent (0.85 vs 1.00); residual confounding acknowledged. |
| Measurement of exposure | some concerns | Gluten derived from food composition with fixed conversion; FFQ correlations for gluten sources 0.35-0.86; repeated measures reduce error. |
| Selection of participants | low | Prospective cohorts; exclusions based on baseline disease. |
| Post-exposure interventions | low | Diet updating suspended after diabetes, CVD or cancer; sensitivity analysis without suspension unchanged. |
| Missing data | low | Missing diet carried forward; few missing gluten data. |
| Measurement of outcome | low | Medical record confirmation of MI and CHD death. |
| Selection of reported result | low | A priori secondary analyses; full set of models reported. |
Full-text review (25 October 2026): Added exclusions (incl. self-reported coeliac disease), gluten estimation method and intake levels, rate difference 75 (51-98)/100,000 person-years, age-adjusted and stepwise HRs, smoker subgroup with Bonferroni, funding details. Bias overall unchanged with full ROBINS-E domains; directness set to indirect for the linked claims (CHD outcome only).
9. Assessment of the statistical interpretation
Null association of estimated gluten with CHD in the main model (HR 0.95, 0.88-1.02). The inverse association after refined-grain adjustment reflects whole-grain-related variance, not a gluten effect. Absolute crude difference 75 per 100,000 person-years favours higher gluten intake but is confounded.
10. Relevance to the specific claim
Counter-evidence for general gluten harm (C-WHT-09).
11. Independent contrary findings
12. Transferability to Germany and Europe
Moderate.
13. Overall methodological judgement
Good-quality cohort evidence that long-term gluten intake is not associated with CHD in adults without diagnosed coeliac disease; does not address gut permeability, intestinal symptoms or gluten sensitivity.
| Multidimensional assessment | |
|---|---|
| Authenticity of the source | verified |
| Methodological quality | high |
| Risk of bias | some concerns |
| Precision of results | high |
| Directness | indirect |
| Reproducibility | unknown |
| Consistency with the overall evidence | unknown |
| Regional transferability | moderate |
| Adequacy of the interpretation (by the archive) | not applicable |
14. Open questions
Effects of gluten intake on outcomes other than CHD (e.g. gastrointestinal symptoms, permeability) in non-coeliac populations; populations outside the US.
15. Linked fact checks
- Gliadin (a wheat gluten protein) increases intestinal permeability via zonulin, which the archive presents (by listing the studies under the white-flour rule) as relevant to everyone eating white flour. – Studies contradicting or limiting the claim
- 'There is no such thing as gluten sensitivity'; the real problem is only that white flour turns into a glue-like substance. – Studies with mixed or unclear results