Cohort studyindependent cross-checkfull text checked★ key study I-UPF-04 in-depth review
Emulsifiers and heart disease in France
Food additive emulsifiers and risk of cardiovascular disease in the NutriNet-Santé cohort: prospective cohort study.
Overview
Sellem L et al.. BMJ 2023 · PMID 37673430 · DOI 10.1136/bmj-2023-076058
Higher intake of some emulsifiers, including CMC, was linked with slightly more heart disease.
Does the study support the specific claim? Partly · Methodological quality: moderate
Plain-language summary
Study explained simply
- What did the researchers want to find out?
Are emulsifier intakes associated with CVD?
- How was the study done?
Prospective cohort with additive exposure estimation.
- How many people, animals or samples were studied?
95,442; 1,995 CVD events; Adults without CVD, 79% women
- What was actually found?
Celluloses (E460-E468) CVD HR 1.05 (1.02-1.09), CHD 1.07; CMC (E466) CVD 1.03 (1.01-1.05), CHD 1.04; E471/E472 associated with all outcomes; E472b cerebrovascular 1.11; trisodium phosphate CHD 1.06.
- How large was the observed effect?
HR 1.03-1.11 per SD
- What do the results mean in everyday life?
Small possible effect.
- What are the study's limitations?
One French cohort of mostly women volunteers; associations are very small and some disappear after correction for multiple testing; most emulsifiers studied showed no link; additive intake was estimated, not measured in blood or urine.
- What does it say about Elawady's specific claim?
Supports C-UPF-06 weakly.
- Do the results apply to Germany and Europe?
High.
- What can explicitly not be concluded from this study?
It cannot show that emulsifiers cause heart disease, nor that they erode the gut mucus in people; it does not provide absolute risks.
Scientific in-depth review
1. Full reference
- Authors
- Sellem L et al.
- Journal
- BMJ
- Year
- 2023
- Identifier
- PMID 37673430 DOI 10.1136/bmj-2023-076058
- Verification
- verified – PubMed efetch 2026-10-25
- Retracted
- no
- Corrections
- none
- Extent checked
- full text
- Conflicts of interest and funding
Public and charitable funding (French Ministry of Health, Santé Publique France, INSERM, INRAE, CNAM, University Sorbonne Paris Nord, ERC Horizon 2020, INCa, IdEx Université de Paris, Bettencourt-Schueller Foundation). ICMJE: no support from for-profit organisations, no relevant financial relationships. B. Chassaing (author of A-UPF-15) is a co-author.
2. Research question
Are emulsifier intakes associated with CVD?
3. Study design and population
- Study design
- Cohort study – Prospective web-based volunteer cohort (NutriNet-Santé) with brand-level additive exposure estimation from repeated 24-h dietary records matched to OQALI, Open Food Facts and Mintel databases plus laboratory assays.
- Population
French adults without prevalent CVD; mean age 43.1 (SD 14.5); 79.0% women; more educated and health-conscious than the general population
- Country
France
- Period
2009 to 5 October 2021; median follow-up 7.4 years (maximum 12.4)
- Sample size
95,442; 1,995 incident CVD, 1,044 coronary heart disease, 974 cerebrovascular disease events
- Inclusion and exclusion criteria
>=3 valid 24-h records in the first 2 years, no prevalent CVD; energy under-reporters (~17% of cohort) excluded; individual emulsifiers analysed only if consumed by >=5% of participants.
4. Methods and endpoints
- Exposure / intervention
Intake (mg/day) of total and individual food-additive emulsifiers, per 1 SD increment (e.g. SD 32.0 mg/day for E466, 3,170.8 mg/day for total emulsifiers)
- Comparator
Lower intake
- Endpoints
CVD, CHD, cerebrovascular
- Statistical methods
Cause-specific Cox models (age timescale) adjusted for sex, BMI, activity, smoking and pack-years, education, number of records, family history of CVD, energy, alcohol, saturated fat, sodium, fibre, sugars, fruit/vegetables, red/processed meat, whole grains and proportion of ultra-processed food; multiple imputation; false discovery rate correction; sensitivity models (dietary patterns, metabolic disorders, weight change, 2-year lag, whole-follow-up exposure, other emulsifiers and sweeteners).
5. Actual results
- Results
Per 1 SD: total celluloses (E460-E468) CVD HR 1.05 (1.02-1.09), CHD 1.07 (1.02-1.12); E460 CVD 1.05 (1.01-1.09), CHD 1.07 (1.02-1.12); carboxymethylcellulose E466 CVD 1.03 (1.01-1.05), CHD 1.04 (1.02-1.06); total mono- and diglycerides (E471/E472) CVD 1.07 (1.04-1.11), CHD 1.08 (1.03-1.14), cerebrovascular 1.07 (1.01-1.13); E472b CVD 1.06 (1.02-1.10), cerebrovascular 1.11 (1.06-1.16); E472c CVD 1.04 (1.02-1.07), CHD 1.06 (1.03-1.09); trisodium phosphate E339 CHD 1.06 (1.00-1.12). After FDR correction, E460 (adjusted P=0.06), E471/E472 for cerebrovascular disease (P=0.1) and E339 (P=0.3) were no longer significant. No associations for other emulsifiers, including carrageenans and lecithins. Results consistent in sensitivity analyses and for severe CVD (excluding TIA and angina).
- Absolute risks
not reported in full text
- Authors' conclusion
Positive associations were found between CVD risk and intakes of total celluloses, E460, E466, total mono- and diglycerides of fatty acids, E472b and E472c; these findings need replication in other populations and should be complemented by experimental studies; they provide new insights for debate on regulation of emulsifier use.
6. Strengths
- Brand-level, repeated, quantitative estimation of individual additive intakes
- Adjustment for diet quality and proportion of UPF, reducing confounding by overall processed-food intake
- FDR correction and extensive sensitivity analyses
- Public funding; no for-profit COI
7. Weaknesses
- Single volunteer cohort (79% women, health-conscious); limited generalisability (authors)
- Very small per-SD HRs (1.03-1.11); absolute risks not reported
- Many emulsifiers and outcomes tested; several associations lost significance after FDR correction
- Additive exposure not validated by biomarkers; under-estimation for unlabelled foods (authors)
- Exclusion of ~17% energy under-reporters
- Residual confounding cannot be ruled out (authors)
- Self-reported diet
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
Structured single-reviewer judgement from the full text; a complete formal ROBINS-E assessment requires two independent reviewers and a protocol.
| Domain | Judgement | Rationale |
|---|---|---|
| Confounding | some concerns | Extensive adjustment including UPF share and nutrients; emulsifier intake still tracks specific food groups (cakes, dairy desserts, soups). |
| Exposure measurement | some concerns | Detailed brand-level matching but no biomarker validation; some products unlabelled; baseline-period exposure. |
| Selection of participants | some concerns | Volunteer cohort; exclusion of under-reporters. |
| Post-exposure interventions | low | Sensitivity analysis using whole-follow-up exposure gave consistent results. |
| Missing data | low | Multiple imputation of covariates. |
| Outcome measurement | low | Physician-validated events, health-insurance and mortality-registry linkage; results similar for severe CVD. |
| Selection of reported result | some concerns | Many exposures tested; FDR correction applied and null results provided in supplement. |
Full-text review (25 October 2026): Headline HRs confirmed. Added: follow-up to October 2021, exclusion criteria (~17% under-reporters), SDs of exposure, adjustment for UPF proportion and nutrients (previous concern about confounding by UPF partly addressed), FDR results (E460, E339 and E471/E472-cerebrovascular no longer significant), null results for carrageenans and lecithins, severe-CVD sensitivity, funding/COI. Domain-level ROBINS-E (overall some_concerns, unchanged).
9. Assessment of the statistical interpretation
Per-SD HRs of 1.03-1.11 are small (e.g. +3% CVD risk per 32 mg/day of E466); several associations do not survive multiple-testing correction. Hypothesis-generating rather than confirmatory.
10. Relevance to the specific claim
Supports C-UPF-06 weakly.
11. Independent contrary findings
None.
12. Transferability to Germany and Europe
High.
13. Overall methodological judgement
Signal of small associations between a few specific emulsifiers (celluloses, mono- and diglycerides) and CVD, independent of overall UPF intake, in one French volunteer cohort; most emulsifiers, including carrageenans and lecithins, showed no association. Needs replication.
| Multidimensional assessment | |
|---|---|
| Authenticity of the source | verified |
| Methodological quality | moderate |
| Risk of bias | some concerns |
| Precision of results | high |
| Directness | direct |
| Reproducibility | not replicated |
| Consistency with the overall evidence | unknown |
| Regional transferability | high |
| Adequacy of the interpretation (by the archive) | not applicable |
14. Open questions
Replication in other cohorts; experimental human data; dose relevance; mechanism (microbiota, inflammation).
15. Linked fact checks
- Industrial emulsifiers such as carboxymethylcellulose erode the gut's protective mucus layer and drive inflammation and metabolic disease. – Studies supporting the claim
- Elawady: preservatives in industrial products are of little health concern (sodium benzoate is present in trivial amounts; preservatives are preferable to pesticide residues; 'returning to nature' is a trap). – Studies contradicting or limiting the claim