Do food preservatives raise heart disease risk?
A May 2026 European Heart Journal study of 112,395 French adults found that people with the highest intake of eight common food preservatives had 10–39% higher hazard ratios for hypertension and 16% higher CVD risk — but the study is observational, and the signal most likely reflects broad ultra-processed food consumption rather than preservatives themselves. The clearest action: cut processed meat, not the preservatives on the label.
A large French cohort study published May 21, 2026 in the European Heart Journal found that people with the highest dietary exposure to eight common food preservatives had meaningfully higher rates of high blood pressure and cardiovascular disease than those with the lowest exposure. The study is observational — it cannot prove the preservatives cause harm — and its methodology for measuring actual preservative consumption has drawn pointed criticism from independent researchers. The headline numbers are real; what they mean for the foods on your plate is a harder question.
The study is worth reading carefully, not dismissing and not panicking over.
What the study found
The NutriNet-Santé cohort enrolled 112,395 adults in France between 2009 and 2024 (78.7% female, mean age 42.8 years) and followed them for a median of 7.9 years. 1 Dietary intake was assessed through repeated 24-hour dietary recalls — up to 96 per participant — capturing brand-level food product information, which was then linked to multiple ingredient databases and laboratory analyses to estimate preservative exposure.
Across the follow-up period, the cohort accumulated 5,544 new hypertension cases and 2,450 cardiovascular events (myocardial infarction, stroke, angina). 2
Among 17 preservatives consumed by at least 10% of participants, 8 showed significant associations with incident hypertension after correcting for multiple comparisons: 3
| Preservative (E-code) | Role | HR for hypertension (95% CI) |
|---|---|---|
| Potassium sorbate (E202) | Mold inhibitor in dairy, baked goods | 1.39 (1.28–1.50) |
| Citric acid (E330) | Acidulant/preservative in drinks, canned goods | 1.25 (1.16–1.34) |
| Potassium metabisulfite (E224) | Sulfite in wine, dried fruit | 1.16 (1.08–1.25) |
| Sodium nitrite (E250) | Curing agent in processed meats | 1.16 (1.08–1.25) |
| Ascorbic acid (E300) | Antioxidant in processed foods | 1.14 (1.06–1.22) |
| Sodium ascorbate (E301) | Antioxidant in processed meats, beverages | 1.12 (1.04–1.20) |
| Sodium erythorbate (E316) | Antioxidant in processed meats | 1.14 (1.06–1.22) |
| Rosemary extract (E392) | Natural antioxidant in oils, snacks | 1.10 (1.02–1.18) |
For CVD endpoints, the picture was narrower. Non-antioxidant preservatives (as a group) were associated with a 16% higher CVD risk (HR 1.16, 95% CI 1.04–1.29) in those with the highest vs. lowest exposure. 1 Ascorbic acid (E300) specifically was associated with cardiovascular events (HR 1.15, 95% CI 1.04–1.28) and cerebrovascular events (HR 1.18, 95% CI 1.01–1.38). Sodium erythorbate (E316) showed similar CVD associations. 3
One more context point: 99.5% of participants consumed at least one food preservative in the first two years of the study. 2 These are not exotic chemicals; they are in a large fraction of packaged food.
Study design — this is not a clinical trial
The NutriNet-Santé study is a prospective observational cohort, not a randomized controlled trial. Participants were not assigned to eat more or fewer preservatives; the researchers observed what people already ate and tracked who developed hypertension or cardiovascular disease over time.
Rachel Richardson, methods support manager at the Cochrane Collaboration (a network specializing in systematic evidence review), reviewed the study and put it plainly: 4
"It's important to note that this is an observational study: this means that participants were not randomly allocated to consume more or fewer preservatives. As such we cannot be sure that the people who consumed more preservatives were not different from those who consumed fewer preservatives in ways that affected their risk of hypertension and CVD."
The researchers did adjust for a substantial list of confounders: age, sex, BMI, smoking, physical activity, alcohol intake, caloric intake, sugar, sodium, saturated fat, and dietary fiber. Richardson's assessment was that "the authors did a good job of controlling for confounders." 4 Residual confounding — from factors not measured or measured imprecisely — cannot be eliminated in any observational study design, however thorough.
Two additional population caveats apply: the cohort was 78.7% female and composed of French volunteers who tend to be more diet-conscious than the general population. French food consumption patterns — particular use of certain preserved meats and wines — also differ from diets in the UK, US, or East Asian contexts, which limits direct transferability.
No absolute risk figures (events per 1,000 person-years by exposure group) were available in the public abstract or press materials. 1 The study reports relative hazard ratios; readers cannot independently judge how large the absolute risk difference is between the highest and lowest consumers from publicly available data alone.
This is the third paper from the NutriNet-Santé group linking food additives to chronic disease endpoints: the same team previously reported preservative associations with cancer (BMJ, 2026) and type 2 diabetes (Nature Communications, 2025). 2 Replication of a direction across three independent disease endpoints in the same cohort adds some credibility to the signal — but does not change the observational design limitation.
The exposure measurement problem
The sharpest criticism of the study concerns not the design but the data used to estimate how much of each preservative participants actually consumed.
Prof. Gunter Kuhnle, a nutritional scientist at the University of Reading and a former member of the European Food Safety Authority's (EFSA) ANS (food additives) Panel, identified a structural limitation in the exposure model: 4
"Consumers should not be concerned by the findings of this study."
His reasoning: the food industry is not required to report the exact quantity of each additive used in a specific product — only upper limits are publicly available. Food formulations change regularly, and the study did not conduct chemical analysis of individual food products. This means the individual-level preservative exposure estimates are built on imprecise inputs, and any associations derived from those estimates carry an additional layer of measurement uncertainty beyond what standard dietary recall methods already entail.
Kuhnle also raised a food matrix argument: some of the preservatives with the strongest associations — sodium nitrite (E250) and sulfites (E224) — are found predominantly in processed meats and wine, respectively. Both of those food categories already carry independent cardiovascular risk signals from multiple other studies. The association could reflect the food itself rather than the preservative. 4
The ascorbic acid finding makes this tension visible. Lead author Anaïs Hasenböhler and corresponding author Mathilde Touvier (INSERM research director, Université Sorbonne Paris Nord) both stressed that the E300 association does not apply to vitamin C from fruit and vegetables: 5
"Naturally occurring ascorbic acid and added ascorbic acid — which may be chemically manufactured — may have different impacts on health. Thus, the results observed here for these food additives are not true for natural substances found in fruits and vegetables."
This is a meaningful caveat: E300-heavy foods are typically ultra-processed products (processed meats, soft drinks, industrial baked goods). An association between E300 consumption and CVD risk may well index ultra-processed food intake more broadly, not the ascorbic acid molecule per se.
What the signal most plausibly means
Taken together, the study is most credible as evidence that frequent consumption of ultra-processed foods is associated with hypertension and CVD risk — a conclusion supported independently by at least a dozen other prospective cohort studies and several systematic reviews. The preservatives are correlated with that consumption pattern; whether they contribute to risk mechanistically, or whether they are markers of a dietary pattern that causes harm through other pathways (excess sodium, refined carbohydrates, saturated fat, low fiber), the study cannot determine.
The two preservatives with the most plausible independent mechanisms — sodium nitrite (E250) and sulfites (E224) — do have experimental data suggesting vascular effects: nitrites generate reactive nitrogen species that may impair endothelial function; sulfites have been studied in the context of oxidative stress. Neither mechanism is conclusively established in humans at typical dietary exposures. 3
Touvier's call for EFSA and FDA to re-evaluate preservative risk-benefit assessments is defensible on precautionary grounds. 6 Kuhnle's counterpoint — that regulators already review additives on an ongoing basis and have removed additives from market when warranted — is also true. 4 These are not irreconcilable positions.
The dietary takeaway
This study does not give you a new mechanism to avoid — it gives you a cleaner marker for an old recommendation.
The most defensible dietary action from these findings is to reduce overall ultra-processed food intake, particularly processed meats (the primary sources of E250, E316, E301) and sulfite-containing products (wine, certain dried fruits, commercial vinegars). This recommendation would hold even if every preservative association turned out to be entirely driven by the food matrix rather than the preservative itself.
More specifically:
- Processed meats (deli meats, hot dogs, sausages, bacon) carry sodium nitrite (E250) and sodium erythorbate (E316) and are already the strongest single food-category signal in cardiovascular epidemiology. The NutriNet-Santé finding adds one more observational data point to an already substantial pile. If you eat processed meats daily, reducing frequency to two or fewer servings per week is consistent with both the prior evidence and this new study.
- Sulfite exposure from wine at moderate consumption is unlikely to be a meaningful target unless you have a known sulfite sensitivity. Dried fruits with added sulfites (listed as E220–E228 on European labels, "sulfur dioxide / sodium bisulfite" on US labels) are a realistic reduction point if you consume them in large quantities.
- E330 (citric acid) and E392 (rosemary extract) showed associations in this study, but both are found in a wide range of foods including relatively unprocessed ones (canned tomatoes, olive oil). The food matrix argument for these two is particularly strong — reducing your intake of all citric acid-containing products is not a coherent or useful target.
- Do not interpret this study as evidence that vitamin C from food is harmful. The authors are explicit on this point. Whole fruit, vegetables, and 100% fruit juice contain ascorbic acid; the E300 association reflects packaged food consumption patterns, not the molecule in its natural context.
The practical threshold question — how much ultra-processed food is low enough risk — remains unanswered by this study. The data report only high vs. low tertiles; no dose-response curve or safe intake level was published in the available materials.
For dietitians advising clients: this study strengthens the evidence base for reducing processed meat frequency as a cardiovascular risk-reduction strategy. It does not create an independent clinical action for the other listed preservatives beyond what a general "minimize ultra-processed food" recommendation already covers.
Study label: Observational cohort. The findings are associative, not causal. The full text of the European Heart Journal article was not accessible (Oxford Academic paywall); this article is based on the structured PubMed abstract, a pre-publication conference abstract, the INSERM press release, the Science Media Centre expert reactions, and ESC press materials. Funding and conflict-of-interest disclosures for the full paper could not be verified. Prof. Gunter Kuhnle disclosed prior funding from Mars (flavanol research, 2010–2020), prior EFSA ANS panel membership (2018–2019), and UK Committee on Toxicity membership (2019–2025).
Cover image: heart-shaped red bell pepper and fresh produce on a wooden board. Image from bioengineer.org.
References
- 1
- 2
- 3
- 4
- 5
- 6
This story was produced automatically by a channel. One sentence is all it takes for Neodrop to keep producing for you.