“Common preservatives used in many store-bought foods to kill bacteria and mold were linked to a 29% greater risk of elevated blood pressure and a 16% higher risk of heart attacks and stroke, according to a new study from France.”
From
I glanced at this paper and I thought it was not really useful to do an analysis of it.
https://academic.oup.com/eurheartj/advance-article/doi/10.1093/eurheartj/ehag308/8679203
I have, however, now shoved the PDF into chatGPT and asked for my usual prompt: do a summary, identify novelty and a critique of the uploaded paper
Summary
This paper examines whether chronic exposure to preservative food additives is associated with later hypertension and cardiovascular disease in the French NutriNet-Santé cohort. It uses a large prospective dataset: 112,395 participants, mostly women, followed from 2009–2024, with a median follow-up of about 7.9 years. Dietary exposure was estimated from repeated online 24-hour dietary records, including brand-level information, then linked to additive-composition databases, laboratory assays, and dynamic matching to account for product reformulation over time.
The main results were:
| Exposure | Main association found |
|---|---|
| Total non-antioxidant preservatives | Higher incidence of hypertension: HR 1.29; higher incidence of CVD: HR 1.16 |
| Total antioxidant preservatives | Higher incidence of hypertension: HR 1.22 |
| Total preservatives overall | Higher incidence of hypertension |
| Individual/group additives linked to hypertension | Sorbates, potassium sorbate E202, potassium metabisulphite E224, nitrites, sodium nitrite E250, ascorbates, ascorbic acid E300, sodium ascorbate E301, erythorbates, sodium erythorbate E316, citric acid E330, rosemary extract E392 |
| Individual/group additives linked to CVD | The strongest stated CVD associations included total non-antioxidant preservatives and ascorbic acid E300; the graphical abstract also lists total ascorbates, erythorbates, and sodium erythorbate E316 as CVD-associated |
The paper also reports that 16.2% of the association between total non-antioxidant preservative exposure and CVD was statistically mediated by hypertension, and a smaller proportion by type 2 diabetes.
The authors conclude that preservative additives widely used in industrial foods may be linked to higher cardiometabolic risk, but that experimental work is needed before making causal claims. They suggest the findings may justify re-evaluating additive regulations if confirmed.
Novelty
The most novel part is not simply the finding that processed foods correlate with cardiovascular risk. The novelty is that the paper tries to separate out specific preservative additive exposures from broader diet quality or ultra-processed food intake.
Key novel elements:
-
Brand-specific additive exposure
The study uses commercial brand information, additive databases, and laboratory assays, rather than just asking “how much processed food did people eat?” This is a more granular exposure model than most nutrition epidemiology studies.
-
Time-dependent exposure modelling
Additive intake was updated over follow-up, rather than treated as a fixed baseline variable. That matters because people’s diets and product formulations change.
-
Wide additive coverage
The study examines 58 preservative food additives, with 17 common enough to study individually. This is broader than the literature on nitrites/nitrates alone.
-
Distinction between additive sources and natural dietary sources
For compounds such as ascorbic acid, citric acid, nitrites, nitrates, and sulphites, the authors tried to distinguish additive-derived exposure from naturally occurring dietary exposure. This is important because, for example, vitamin C in fruit is not nutritionally equivalent to ascorbic acid used as an additive in industrial products.
-
Cardiovascular endpoints
The paper extends concern about additives beyond cancer, microbiome, or metabolic outcomes into hypertension and CVD incidence.
Strengths
The study has several real strengths.
First, the cohort is large, prospective, and long-running, which reduces some problems of reverse causation compared with cross-sectional work. There were 5,544 incident hypertension cases and 2,450 CVD cases, giving reasonable statistical power.
Second, exposure assessment is unusually detailed for food additive epidemiology. Repeated 24-hour records, brand-level information, dynamic reformulation matching, and laboratory assays are a much stronger design than simple food-frequency questionnaires.
Third, the models adjust for many relevant confounders: age, sex, BMI, education, smoking, physical activity, sodium, saturated fat, fibre, sugar, fruit and vegetables, dairy, red/processed meat, alcohol, and energy intake. The authors also ran sensitivity analyses, used false-discovery-rate correction for multiple testing, tested for nonlinearity, and used hip fracture as a negative-control outcome.
Fourth, the finding that sodium itself was associated with hypertension, while the additive associations remained after sodium adjustment, makes the result more interesting. It suggests the signal is not simply “more salty processed food.”
Critique
The biggest limitation is that this remains observational epidemiology. The paper can show association, not causation. People who consume more additive-containing foods may differ in many hard-to-measure ways: dietary pattern, convenience-food reliance, income-related food choices, stress, sleep, eating timing, food packaging exposure, emulsifiers, sweeteners, contaminants, or other unmeasured aspects of industrial food processing.
A second issue is residual confounding by ultra-processed food structure. Even though the authors adjust for ultra-processed food intake and major food groups, preservatives are not randomly assigned. They travel with products, processing methods, packaging, shelf life, storage, industrial fats, refined carbohydrates, and other additives. The study tries to separate these, but full separation is probably impossible in an observational cohort.
Third, some of the associations are biologically harder to interpret. For example, ascorbic acid/citrates as additives were associated with higher risk, whereas naturally vitamin-C-rich diets are usually associated with better cardiovascular health. That may mean additive source matters, but it may also mean these additives are markers of particular industrial foods rather than causal agents.
Fourth, the cohort is not fully representative. NutriNet-Santé participants are volunteers, mostly women, more educated, and likely more health-conscious than the general population. The authors acknowledge this. Generalising to men, less health-conscious populations, non-French populations, or groups with different food systems should be done cautiously.
Fifth, hypertension ascertainment may miss undiagnosed cases. This matters because hypertension is often silent, and if detection is related to health awareness or healthcare contact, bias could occur. Since the cohort is majority female and likely more health-engaged, diagnosis patterns may not reflect the wider population.
Sixth, multiple testing remains an issue despite false-discovery-rate correction. The authors examine many additives, groups, outcomes, sex strata, nonlinear models, and sensitivity analyses. Some findings are likely robust, especially the broad non-antioxidant preservative signal for hypertension, but isolated additive-specific findings should be treated as hypothesis-generating.
Overall assessment
This is a strong and important epidemiological paper, mainly because it moves beyond “ultra-processed foods are bad” toward specific additive exposure modelling. The most convincing finding is the association between total non-antioxidant preservative exposure and hypertension, with a weaker but still notable association with CVD.
However, the paper should not be read as proving that potassium sorbate, nitrite, ascorbic acid, citric acid, or rosemary extract directly cause hypertension. A more cautious interpretation is:
People with higher long-term exposure to several preservative additives, especially non-antioxidant preservatives, had higher subsequent hypertension risk in this cohort, even after extensive adjustment. The findings justify mechanistic and replication studies, but causal attribution to individual additives remains uncertain.
The policy implication is not yet “ban these additives,” but rather: reassess their cardiovascular safety using experimental studies, better mixture models, and replication in other cohorts, while continuing to favour freshly prepared or minimally processed foods where practical.
I asked chatGPT what the paper said about citrates and:
The paper treats citric acid / citrates as preservative antioxidant food additives, not as naturally occurring citrate in fruit or as citrate supplements.
What it studied
It grouped citrates as additives with European codes:
| Additive group | E numbers |
|---|---|
| Citrates | E330, E332, E333 |
Specifically, citric acid is E330. The paper says potassium citrate E332 and calcium citrate E333 had no consumers in this cohort, so the citrate signal appears to be essentially driven by citric acid E330, not citrate salts.
Exposure levels
Citric acid/citrates were very common. The paper says citric acid was consumed by 91.3% of participants, making it one of the most common preservative additives detected.
For total citrates, the mean intake was about 322.5 mg/day across all participants, and 353.4 mg/day among consumers only. Median intake among all participants was 199.6 mg/day.
The paper also says that for compounds that occur both naturally and as additives, the additive contribution varied: for citric acid, about 16% of total dietary citric acid came from additive sources on average. It also notes that processed fruits and vegetables were a major source of citrate additive exposure.
Association with hypertension
The main result for citric acid was an association with higher hypertension incidence:
| Additive | Outcome | HR higher vs lower exposure | Interpretation |
|---|---|---|---|
| Citric acid E330 | Hypertension | 1.25 | 25% higher hazard in higher consumers |
| 95% CI | 1.16–1.34 | Statistically significant | |
| P value | < .001 | Remained significant after FDR correction |
So, in the study, citric acid E330 was one of the additives associated with higher hypertension risk after multiple-testing correction.
Association with CVD
The paper does not highlight citric acid/citrates as clearly associated with overall cardiovascular disease in the same way it highlights ascorbates and erythorbates. The graphical abstract lists citric acid E330 under hypertension-associated additives, not under CVD-associated additives.
Nonlinear dose-response
The authors report that the spline analysis suggested a possible plateau-type relationship between citric acid E330 and hypertension, with a turning point around 366.87 mg/day. In other words, the association may not simply rise linearly forever; it may rise and then flatten. They say this possible nonlinearity needs further study.
Mechanistic comment
The paper does not give a strong established mechanism for citric acid causing hypertension. It notes that some experimental evidence has suggested antihypertensive properties for citric acid, which makes the observed positive association somewhat counterintuitive. It also mentions an in vitro study where citric acid enhanced cell proliferation in some contexts, including liver-related models, but this is not presented as a proven cardiovascular mechanism.
Bottom line
The paper’s finding is:
Additive citric acid E330 was common and was associated with higher subsequent hypertension risk, but not clearly highlighted as associated with CVD.
My reading is that this should be treated cautiously. Citric acid may be acting as a marker of certain industrial food matrices, rather than being proven causal itself. The paper’s own design adjusts for many confounders, but it remains observational.
Personally I am not a fan of citric acid as a supplementary substance because it is acidic.