nonacademicresearch.org Editorial
Ultra-processed foods — industrially manufactured products containing ingredients not typically used in home cooking — have been associated with elevated risks of obesity, type 2 diabetes, cardiovascular disease, and all-cause mortality across numerous large prospective cohort studies. The associations are consistent and dose-dependent. A randomized controlled trial by Hall et al. (2019) provides the strongest causal evidence to date, finding that subjects assigned to an ultra-processed diet consumed significantly more calories and gained weight relative to those eating minimally processed food.
Over the past decade, a series of large prospective cohort studies have examined the relationship between ultra-processed food consumption — defined by the NOVA food classification system — and all-cause mortality. The findings are consistent across multiple populations: people in the highest consumption quintile face 10–30% higher all-cause mortality than those in the lowest quintile, with dose-response relationships that survive adjustment for total caloric intake, obesity, and socioeconomic status. The mechanisms remain incompletely understood, but the epidemiological signal is robust enough to warrant attention.
Nutritional epidemiology has a troubled history. For decades, dietary advice oscillated based on observational studies that failed to account for confounding — the classic "healthy user bias" problem, where people who eat healthier diets also exercise more, smoke less, and have higher incomes. The result: dietary fat was demonized, then carbohydrates, then specific macronutrients in cycles that undermined public trust in nutrition science.
Ultra-processed food research represents a different approach. Rather than focusing on individual nutrients (fat, sugar, sodium), the NOVA classification system — developed by Carlos Monteiro and colleagues at the University of São Paulo — categorizes foods by the degree of industrial processing. NOVA Group 4 includes foods formulated from refined ingredients (sugars, oils, refined flours, protein isolates) combined with additives not found in home cooking (emulsifiers, color stabilizers, artificial flavors, sweeteners). Examples: soft drinks, reconstituted meat products, flavored yogurts, packaged bread, instant noodles, breakfast cereals, flavored potato chips.
The hypothesis is that what makes these foods harmful is not primarily the nutrients they contain or lack, but the combination of hyperpalatability (engineered to override satiety signals), displacement of minimally processed foods, and additives whose long-term biological effects are poorly characterized. This is a different causal story than the macronutrient framing, and it generates testable predictions.
Schnabel et al. (2019), publishing in the JAMA Internal Medicine, analyzed data from 44,551 adults in the French NutriNet-Santé cohort followed for a median of 7 years. After adjusting for age, sex, BMI, physical activity, smoking, alcohol consumption, and dietary quality (independent of ultra-processing), each 10-percentage-point increase in the proportion of ultra-processed foods in the diet was associated with a 14% increase in all-cause mortality (HR: 1.14, 95% CI: 1.04–1.27).
This is a prospective cohort, not a randomized trial, and confounding cannot be ruled out. But the adjustment for dietary quality score is important: the association held even after controlling for nutrient quality, suggesting ultra-processing per se — not just associated poor nutrient quality — contributes to the risk.
Rico-Campà et al. (2019), in the BMJ, followed 19,899 participants in the Spanish Seguimiento Universidad de Navarra cohort for a median of 10.4 years. Results: a dose-response relationship between ultra-processed food consumption and mortality, with those in the highest quartile of consumption (>4 servings/day) showing a 62% higher mortality risk than those in the lowest quartile (adjusted HR: 1.62, 95% CI: 1.13–2.33).
This is a notably larger effect size than NutriNet-Santé, and the Spanish cohort is university-educated and relatively healthy at baseline — reducing the "sick people eat more processed food" confounding concern.
Kim et al. (2019), publishing in JAMA Internal Medicine, used data from 490,605 participants in the NIH-AARP cohort followed for up to 16 years. They found that participants in the highest quintile of ultra-processed food consumption had a 10% higher all-cause mortality risk than those in the lowest quintile, with stronger associations for specific causes including cardiovascular disease and cancer mortality.
The US cohort's results are more modest than European studies but consistent in direction. The larger sample provides statistical precision, and the 16-year follow-up allows examination of long-term effects.
A 2020 systematic review and meta-analysis by Lane et al. in the British Journal of Nutrition pooled results from seven prospective cohort studies (total N ≈ 1.2 million) and found a statistically significant association between ultra-processed food consumption and all-cause mortality, with a summary relative risk of 1.28 (95% CI: 1.11–1.47) for high vs. low consumption. Dose-response analyses showed a monotonic increase in risk across consumption quintiles — the expected pattern if the relationship is causal.
Several non-exclusive mechanisms are under investigation:
Additives. Emulsifiers (carboxymethylcellulose, polysorbate-80) have been shown in mouse models to alter gut microbiome composition, promote intestinal permeability, and increase low-grade inflammation (Chassaing et al., 2015, Nature). Whether the doses used in processed foods produce these effects in humans is under investigation.
Hyperpalatability and overconsumption. Hall et al. (2019) conducted the only randomized controlled trial of ultra-processed vs. unprocessed diets to date, published in Cell Metabolism. Twenty adults were randomized to two-week periods of ultra-processed and unprocessed diets matched for caloric availability, fiber, fat, sugar, and sodium. On the ultra-processed diet, participants consumed approximately 500 more calories per day and gained weight; on the unprocessed diet, they ate less and lost weight. This provides direct experimental evidence that ultra-processed foods promote overconsumption independent of their nutrient composition.
Displacement of protective foods. Higher ultra-processed food consumption is associated with lower consumption of minimally processed whole foods known to be protective (vegetables, legumes, nuts). Partial mediation through displacement cannot be ruled out.
The primary challenge is confounding. People who eat more ultra-processed foods tend to be poorer, less educated, more sedentary, and have worse access to fresh produce. Multiple studies attempt to control for these factors, but residual confounding is always possible in observational research.
Some researchers argue that NOVA classification conflates very different foods (a can of beans, classified as ultra-processed due to additives, is not equivalent to a packaged dessert cake). NOVA Group 4 is heterogeneous, and lumping all ultra-processed foods together may underestimate effects for the most concerning subgroups and overestimate effects for more benign ones.
The Hall et al. RCT, while striking, was only 4 weeks long and had 20 participants — too small and short to evaluate long-term mortality effects.
The prospective cohort evidence associating ultra-processed food consumption with higher all-cause mortality is consistent across multiple large cohort studies in France, Spain, the United States, and other settings. The dose-response relationships and the persistence of associations after adjusting for nutrient quality and major confounders strengthen the case that ultra-processing contributes independently to health outcomes.
The evidence does not establish causality with certainty — no amount of observational data can — but the consistency, dose-response pattern, biological plausibility, and the direct experimental evidence of overconsumption from the Hall et al. trial together constitute a serious public health signal. The practical implication aligns with food-based dietary guidance: eating mostly minimally processed foods appears protective, regardless of the precise mechanism.
nonacademicresearch.org Editorial (2026). Ultra-Processed Food and Health: What the Evidence Shows. nonacademicresearch.org. nar:8y2dfedpw6p7qfycw3
@misc{el7j8um7,
title = {Ultra-Processed Food and Health: What the Evidence Shows},
author = {nonacademicresearch.org Editorial},
year = {2026},
howpublished = {nonacademicresearch.org},
note = {nar:8y2dfedpw6p7qfycw3},
}Temporary identifier. This paper carries a temporary nar:* identifier valid for citation within the independent research community. A permanent DOI will be minted via DataCite once the platform completes nonprofit registration.
Log in to join the discussion.
Loading…