← All articles
Nutrition

Ultra-Processed Foods and Mortality: Evidence from 18 Cohort Studies

A 2025 meta-analysis pooled 18 cohorts and 1.1 million participants. Highest consumption of ultra-processed foods is linked to a 15% increase in all-cause mortality (HR=1.15). Each additional 10% share of UPF in the diet adds another 10% to risk. Data are observational, but consistent across multiple independent cohorts.

6 min readNutrition22.09.2026
Quick answer

A meta-analysis of 18 cohort studies (Liang et al., Systematic Reviews, 2025; n=1,148,387; 173,107 deaths) reports HR=1.15 (95% CI: 1.09–1.22) for all-cause mortality at highest vs. lowest NOVA-defined ultra-processed food consumption. The dose-response relationship is linear and statistically significant. Data are observational.

What are ultra-processed foods under the NOVA classification?

The NOVA classification, developed by Monteiro's group at the University of Sao Paulo, divides all foods into four groups based on the degree of industrial processing. Group 4 — ultra-processed foods — are industrially manufactured mixtures containing ingredients not used in home cooking: emulsifiers, stabilizers, flavorings, colorings, preservatives, and sweeteners. Whole or minimally processed source ingredients are minimal or absent.

Typical examples include: carbonated soft drinks, sweetened breakfast cereals, industrial bread, packaged snacks, processed meats with fillers, instant noodles and soups, industrial desserts, and fast food. The classification is based not on nutrient composition but on the degree of processing and the presence of characteristic industrial additives.

How much of what we eat is ultra-processed?

In several high-income countries, ultra-processed foods account for more than half of daily caloric intake. In the US, UK, Canada, and Australia, this share exceeds 50% of total energy. This is not a marginal fraction of the diet — it is a structural component for the majority of the population.

This prevalence makes NOVA Group 4 a practically unavoidable subject for epidemiological research: any meaningful association with health outcomes concerns the majority of people in developed countries.

In the US, ultra-processed foods account for more than 50% of daily calories. This is not a dietary anomaly — it is the norm of modern eating, studied by the world's largest cohorts.

What does the largest 2025 meta-analysis show?

Liang and colleagues published in Systematic Reviews (2025) an updated systematic review and dose-response meta-analysis based on 18 prospective cohort studies with NOVA-classified data. Combined sample: 1,148,387 participants, 173,107 deaths, mean follow-up 14.5 years.

Key results for all-cause mortality:

  • Highest vs. lowest UPF consumption: HR=1.15 (95% CI: 1.09–1.22)
  • Per additional 10% UPF share in diet: HR=1.10 (95% CI: 1.04–1.16)
  • Dose-response relationship: linear and statistically significant (p<0.001)

Heterogeneity was high: I²=83.0% for highest vs. lowest and I²=91.0% for the incremental analysis. Meta-regression indicated that sex distribution partially explained between-study variability (stronger effect in males). The authors note that data remain observational and do not exclude residual confounding.

What does the EPIC Lancet cohort show?

Simultaneously, Gonzalez-Gil and colleagues published in The Lancet Regional Health — Europe (2025) an analysis of the EPIC cohort: 428,728 participants from 9 European countries, 40,016 deaths, median follow-up 15.9 years.

Association of ultra-processed foods (per 1 SD of consumption) with various causes of death:

  • All-cause mortality: HR=1.04 (95% CI: 1.02–1.05)
  • Cardiovascular mortality: HR=1.09 (95% CI: 1.07–1.12)
  • Cerebrovascular mortality: HR=1.11 (95% CI: 1.05–1.17)
  • Ischemic heart disease mortality: HR=1.10 (95% CI: 1.06–1.15)
  • Digestive disease mortality: HR=1.12 (95% CI: 1.05–1.20)
  • Parkinson's disease mortality: HR=1.23 (95% CI: 1.06–1.42)

No significant association was found for cancer or Alzheimer's disease mortality. The absolute HR values are smaller than in the global meta-analysis, reflecting the lower baseline UPF consumption in European cohorts.

What mechanisms are proposed?

The mechanisms linking ultra-processed foods to health outcomes remain under investigation. Several non-mutually exclusive hypotheses are considered:

  • Nutrient displacement: high UPF intake reduces the share of whole foods rich in fiber, antioxidants, polyphenols, and essential nutrients.
  • Food additives: emulsifiers (polysorbate 80, carboxymethylcellulose) in animal experiments disrupted gut mucin structure and shifted microbiome composition toward pro-inflammatory patterns. Human data are limited.
  • Nitrites: in processed meats, associated with colorectal cancer (WHO Group 1 — proven carcinogen).
  • Matrix effect: industrial processing alters the physical structure of food, affecting digestion speed, glycemic index, and satiety independent of macronutrient composition.

None of these hypotheses has been confirmed at the level of evidence sufficient to establish causality in humans.

What this means in practice
  • Assess the share of ultra-processed foods in your diet. If carbonated drinks, sweetened cereals, fast food, and packaged snacks make up a significant part of daily eating — that is NOVA Group 4 by definition.
  • The association data do not imply prohibition. The linear nature of the dose-response means: any reduction in the UPF share of the diet is associated with a proportional reduction in associated risk.
  • Replacing one ultra-processed product with a minimally processed equivalent is a practically achievable step. Example: flavored industrial yogurt → plain yogurt; breakfast cereals → whole-grain oatmeal.
  • The observational nature of data means one cannot exclude that people who consume fewer ultra-processed foods lead healthier lifestyles overall. However, the association persists after adjustment for calories, physical activity, BMI, and socioeconomic factors.
  • In a sports nutrition context: industrial sports nutrition products (protein bars with long additive lists, ultra-sweet mass gainers) also fall into NOVA Group 4. This does not mean avoiding them entirely, but warrants deliberate choice.

Frequently asked questions

What are ultra-processed foods under the NOVA classification?
NOVA is a processing-degree classification (University of Sao Paulo, Monteiro). Group 4 — industrially manufactured products with additives not used in home cooking: emulsifiers, stabilizers, flavorings, colorings. Typical examples: soft drinks, fast food, industrial desserts, packaged snacks.
How strong is the link between ultra-processed foods and mortality?
Meta-analysis by Liang et al. (Systematic Reviews, 2025; 18 cohorts, n=1,148,387, 173,107 deaths): HR=1.15 (95% CI: 1.09–1.22) comparing highest vs. lowest consumption; HR=1.10 per +10% share. Dose-response is linear (p<0.001). Data are observational.
What share of the diet are ultra-processed foods?
In the US, UK, Canada, and Australia they account for more than 50% of daily calories. In European cohorts (EPIC) consumption is lower, matching a smaller absolute effect (HR=1.04 per 1 SD). The higher the baseline consumption, the greater the opportunity to reduce risk through dietary changes.
Why might ultra-processed foods be harmful?
The exact mechanism is unknown. Working hypotheses: nutrient displacement, emulsifier effects on the microbiome, nitrite effects in processed meats, food matrix changes from industrial processing. None has been proven as a single cause-and-effect relationship in humans.

Sources

  1. Liang X., Liang Z., Chen X. et al. «Ultra-processed foods and risk of all-cause mortality: an updated systematic review and dose-response meta-analysis of prospective cohort studies». Systematic Reviews. 2025;14:67. DOI: 10.1186/s13643-025-02800-8. link.springer.com/article/10.1186/s13643-025-02800-8
  2. González-Gil E.M., Matta M., Berstein M. et al. «Associations between degree of food processing and all-cause and cause-specific mortality: a multicentre prospective cohort analysis in 9 European countries». The Lancet Regional Health — Europe. 2025;50:101208. DOI: 10.1016/j.lanepe.2024.101208. thelancet.com/journals/lanepe/article/PIIS2666-7762(24)00377-6/fulltext
  3. Suksatan W., Moradi S., Naeini F. et al. «Ultra-Processed Food Consumption and Adult Mortality Risk: A Systematic Review and Dose-Response Meta-Analysis of 207,291 Participants». Nutrients. 2022;14(1):174. DOI: 10.3390/nu14010174. pmc.ncbi.nlm.nih.gov/articles/PMC8747520/
  4. Monteiro C.A., Cannon G., Levy R.B. et al. «Ultra-processed foods: what they are and how to identify them». Public Health Nutrition. 2019;22(5):936–941. DOI: 10.1017/S1368980018003762. cambridge.org/.../public-health-nutrition/...
This material is for educational purposes only and does not constitute medical advice.

Build your body on data, not hype

Anvil builds your program around your metrics — load, nutrition, and recovery in one system.

Open in Telegram