Meta-analysisindependent cross-checkfull text checked★ key study I-LEG-01 in-depth review
Legumes and death rates
Legume Consumption and Risk of All-Cause and Cause-Specific Mortality: A Systematic Review and Dose-Response Meta-Analysis of Prospective Studies.
Overview
Zargarzadeh N et al.. Adv Nutr 2023 · PMID 36811595 · DOI 10.1016/j.advnut.2022.10.009
People eating more legumes had about 6% lower risk of death from any cause.
Does the study support the specific claim? No · Methodological quality: moderate
Plain-language summary
Study explained simply
- What did the researchers want to find out?
Legumes and mortality
- How was the study done?
Meta-analysis
- How many people, animals or samples were studied?
1.14 million
- What was actually found?
Lower all-cause mortality
- How large was the observed effect?
HR 0.94
- What do the results mean in everyday life?
Legumes compatible with longevity.
- What are the study's limitations?
Observational
- What does it say about Elawady's specific claim?
Contradicts harm claims
- Do the results apply to Germany and Europe?
Moderate
- What can explicitly not be concluded from this study?
Cannot prove causation.
Scientific in-depth review
1. Full reference
- Authors
- Zargarzadeh N et al.
- Journal
- Adv Nutr
- Year
- 2023
- Identifier
- PMID 36811595 DOI 10.1016/j.advnut.2022.10.009
- Verification
- verified – PubMed efetch 2026-10-25
- Retracted
- no
- Corrections
- none
- Extent checked
- full text
- Conflicts of interest and funding
One author funded by Olav og Gerd Meidel Raagholt's Stiftelse and South-East Norway Regional Health Authority (grant 2017076); authors report no conflicts of interest.
2. Research question
Legume intake and mortality.
3. Study design and population
- Study design
- Meta-analysis – Systematic review and dose-response meta-analysis of prospective cohort studies (PROSPERO CRD42022296260; PRISMA)
- Population
General adult populations (aged >=18 y) without prior disease; 5 US, 14 European, 10 Asian, 1 Australian and 2 international cohorts
- Country
Multiple (USA, Europe, Asia, Australia, international)
- Period
Search from inception to December 2021, updated to September 2022; studies published 1995-2022; follow-up 3-26 years
- Sample size
32 cohorts (31 publications), 1,141,793 participants; 93,373 all-cause, 18,056 CVD, 2037 CHD, 2317 stroke and 12,890 cancer deaths
- Inclusion and exclusion criteria
Prospective cohorts of adults reporting total or subtype legume intake (soy foods excluded) and adjusted HR/RR with 95% CI for mortality; duplicate cohort publications de-duplicated
4. Methods and endpoints
- Exposure / intervention
Highest vs lowest legume intake; per 50 g/day increment (1 serving assumed 100 g)
- Comparator
Lower intake
- Endpoints
All-cause, CVD, CHD, stroke, cancer mortality
- Statistical methods
DerSimonian-Laird random effects; linear (Greenland-Longnecker) and non-linear one-stage mixed-effects dose-response with restricted cubic splines; I2; subgroup, leave-one-out and Egger tests; ROBINS-I for primary studies; GRADE
5. Actual results
- Results
High vs low: all-cause HR 0.94 (95% CI 0.91-0.98; 27 cohorts; I2 64.6%); stroke 0.91 (0.84-0.99; 5; not robust in leave-one-out); CVD 0.99 (0.91-1.09); CHD 0.93 (0.78-1.09); cancer 0.85 (0.72-1.01). Per 50 g/day: all-cause 0.94 (0.89-0.99; 19 cohorts), CVD 1.01 (0.94-1.08), CHD 0.90 (0.71-1.13), stroke 0.90 (0.76-1.06), cancer 0.82 (0.61-1.10); no non-linearity. Subgroups all-cause: Europe 1.01 (0.95-1.07; n=11), USA 1.00 (0.96-1.05; n=5), Asia/Australia 0.90 (0.85-0.95), international 0.78 (0.69-0.87). No publication bias (Egger P=0.15). Primary studies: 85% moderate, 15% serious risk of bias (ROBINS-I). GRADE moderate for all-cause and stroke, low for CVD, CHD, cancer.
- Absolute risks
not reported in full text
- Authors' conclusion
Higher legume intake was associated with lower all-cause and stroke mortality, with no association for CVD, CHD or cancer mortality; findings strongly support recommendations to eat more legumes.
6. Strengths
- Registered protocol; four databases without language limits
- Duplicate selection, extraction and ROBINS-I assessment
- Large number of cohorts and deaths; linear and non-linear dose-response
- GRADE certainty rating; subgroup and influence analyses
7. Weaknesses
- Observational data: residual confounding and healthy-user effects possible
- Substantial heterogeneity (I2 ~65% for all-cause)
- No association in European (HR 1.01) or US (HR 1.00) cohorts; inverse association driven by Asian and international cohorts
- Single baseline dietary assessment; legume type and preparation rarely reported
- Stroke finding not robust; few studies for CHD, stroke, cancer
- Substitution (what legumes replace) not addressed
- Authors' 'strongly support' wording is stronger than moderate/low GRADE certainty
8. Risk of bias
- Orienting tool
- AMSTAR 2
- 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
Single-reviewer structured AMSTAR 2 appraisal from full text; not formal (needs two independent reviewers). Review conduct is good (moderate-to-high confidence); remaining concerns stem from the observational primary evidence and heterogeneity.
| Domain | Judgement | Rationale |
|---|---|---|
| Item 1 PICO components | low | Clearly defined. |
| Item 2 Protocol registered a priori (critical) | low | PROSPERO CRD42022296260. |
| Item 3 Explanation of included designs | low | Prospective cohorts justified. |
| Item 4 Comprehensive search (critical) | low | Four databases, reference lists, key journals, no language limits. |
| Item 5 Duplicate selection | low | Two independent reviewers. |
| Item 6 Duplicate extraction | low | Two independent extractors. |
| Item 7 Excluded studies justified (critical) | low | Supplemental Table 2 lists exclusions and reasons. |
| Item 8 Description of included studies | low | Supplemental tables 3-7. |
| Item 9 RoB assessment (critical) | low | ROBINS-I by two reviewers. |
| Item 10 Funding of included studies | high | Not reported. |
| Item 11 Appropriate meta-analytic methods (critical) | low | Random effects and established dose-response methods. |
| Item 12 Impact of RoB on results | some concerns | No sensitivity analysis by RoB category reported. |
| Item 13 RoB considered in interpretation (critical) | some concerns | GRADE downgraded for confounding, but conclusion wording ('strongly support') is assertive. |
| Item 14 Heterogeneity explained | low | Extensive subgroup analyses. |
| Item 15 Publication bias (critical) | low | Funnel plots and Egger tests. |
| Item 16 Conflicts of interest | low | Declared. |
Full-text review (25 October 2026): Added protocol registration, region subgroups (Europe HR 1.01, USA 1.00 null; Asia 0.90), dose-response for all outcomes, primary-study ROBINS-I distribution, GRADE ratings, soy exclusion, funding/COI; AMSTAR 2 item-level appraisal; transferability to EU revised to moderate.
9. Assessment of the statistical interpretation
Modest relative association (6% lower all-cause mortality) with moderate GRADE certainty; heterogeneous by region, null in European and US cohorts. Importantly for harm claims, no subgroup or outcome shows increased mortality with higher legume intake (upper CI bounds around 1.05-1.09).
10. Relevance to the specific claim
Contradicts claims of general legume harm.
11. Independent contrary findings
None for harm.
12. Transferability to Germany and Europe
Moderate: European cohorts (n=11) showed no association with all-cause mortality (HR 1.01, 0.95-1.07), i.e. neither benefit nor harm.
13. Overall methodological judgement
Key evidence against general harm.
| Multidimensional assessment | |
|---|---|
| Authenticity of the source | verified |
| Methodological quality | moderate |
| Risk of bias | some concerns |
| Precision of results | high |
| Directness | direct |
| Reproducibility | replicated |
| Consistency with the overall evidence | mixed |
| Regional transferability | moderate |
| Adequacy of the interpretation (by the archive) | not applicable |
14. Open questions
Causality; effects of specific legume types and preparation; what legumes replace in the diet; reasons for regional differences.
15. Linked fact checks
- Fava beans and lentils (together with flour products) are toxins that cause inflammation and deteriorate health; the fava bean hull is indigestible and the bean contains toxins that ferment 'dirtily'. – Studies contradicting or limiting the claim
- Legumes, and fava beans in particular, cause autoimmune diseases. – Studies contradicting or limiting the claim
- Because of favism (and the other narrow risks), legumes should be excluded from everyone's diet. – Studies contradicting or limiting the claim
- Plant lectins in legumes survive digestion, bind to intestinal cells and harm human health, so legumes are harmful even when eaten normally. – Studies contradicting or limiting the claim
- Legumes as a whole (fava beans, lentils, chickpeas, beans, cowpeas, peas, lupin, peanuts) are harmful and should be fully forbidden. – Studies contradicting or limiting the claim
- Avoiding (nearly) all vegetables, salads and legumes is compatible with good health. – Studies contradicting or limiting the claim