Across the United States, Australia, Brazil, Canada, Chile, Colombia, Mexico, and the United Kingdom, researchers now estimate that a substantial share of premature deaths each year can be linked to ultra-processed food consumption. The study, published in the American Journal of Preventive Medicine and analyzed here in June 2026, found that the risk does not simply exist or not exist. It scales: for every additional 10 percent of daily calories that come from ultra-processed products, the likelihood of dying early climbs measurably higher. In countries where those products already supply more than half of the average person’s energy, the projected toll is steepest.
The findings arrive at a moment when ultra-processed foods, a category that includes soft drinks, packaged snacks, instant noodles, reconstituted meat products, and many frozen ready meals, are not a fringe part of the Western diet. They are the center of it. In the United States, ultra-processed items account for roughly 57 percent of adult caloric intake, according to nationally representative survey data. The new research suggests that dominance carries a measurable cost in human life.
What the eight-country study actually found
The research team, led by Eduardo A.F. Nilson of the University of São Paulo, built a dose-response model using data from prospective cohort studies and national dietary surveys. Rather than asking whether ultra-processed food is harmful in a binary sense, the model quantified how much additional mortality risk accompanies each 10 percent increase in ultra-processed calories as a share of total energy intake. Countries with higher average consumption, notably the United States and the United Kingdom, showed larger estimated death tolls. Countries with lower intake, such as Colombia and Brazil, registered smaller but still meaningful burdens.
The approach draws on real-world dietary patterns tracked over years in large populations, not a single controlled experiment. That distinction matters: the model captures how people actually eat, but it cannot isolate ultra-processed food as the sole cause of any individual death. What it can do is estimate, with statistical rigor, how many fewer deaths might occur if populations shifted even modestly toward less processed diets.
The pattern holds across multiple independent datasets
The eight-country estimate did not emerge in isolation. An updated meta-analysis published in Systematic Reviews pooled results from multiple prospective cohorts worldwide, harmonizing how ultra-processed food was defined using the NOVA classification system, which groups foods by degree of industrial processing rather than nutrient content alone. After adjusting for confounders including body mass index, smoking, physical activity, and socioeconomic status, the review confirmed a statistically significant, graded association: more ultra-processed food, higher all-cause mortality.
U.S.-specific data reinforces the global picture. A cohort analysis using National Health and Nutrition Examination Survey data from 2003 through 2018, linked to National Death Index mortality records, found that American adults who consumed more ultra-processed food faced higher all-cause mortality risk even after researchers controlled for overall diet quality and total calorie intake. That last detail is important: it suggests the degree of processing itself carries risk information beyond what a simple nutrient breakdown would predict. A separate, earlier analysis using NHANES III data from 1988 through 1994 had already identified a similar mortality link over long follow-up, indicating the pattern predates the current wave of public attention by decades.
European cohorts tell a consistent story. France’s NutriNet-Santé study, published in JAMA Internal Medicine, reported increased mortality risk per 10 percent rise in ultra-processed food proportion among middle-aged adults. Spain’s SUN prospective cohort, published in The BMJ, found that participants in the highest consumption group faced a significantly greater hazard of death from any cause, with a clear dose-response gradient. A separate multicentre analysis covering nine European countries, coordinated by the International Agency for Research on Cancer (part of the World Health Organization), documented associations between the degree of food processing and deaths from cardiovascular disease and certain cancers.
Tying these threads together, an umbrella review published in The BMJ in 2024 aggregated dozens of epidemiological meta-analyses on ultra-processed food and adverse health outcomes. All-cause mortality was among the endpoints with the most consistent positive associations, meaning higher consumption repeatedly tracked with higher death rates across study designs, countries, and demographic groups. The review also flagged links with cardiometabolic disease, mental health disorders, and several site-specific cancers, suggesting ultra-processed diets may contribute to a broad spectrum of disease rather than a single condition.
Where the science still has gaps
No randomized controlled trial has ever assigned thousands of people to eat ultra-processed food for years and then measured how many died. That kind of study would be neither ethical nor practical, which means the entire evidence base rests on observational research. Prospective cohort studies can show that people who eat more ultra-processed food die at higher rates, but they cannot definitively prove the food itself, rather than some unmeasured factor traveling alongside it, is the direct cause. Researchers adjust for every confounder they can measure, and the association still holds. But residual confounding, the possibility that some untracked variable explains part of the link, cannot be fully eliminated.
The eight-country modeling study has not released individual-level dietary records or verified death certificates for independent re-analysis. National statistical agencies have also not publicly confirmed the ultra-processed food sales volumes used to scale the mortality projections. Until those details are disclosed, the precise numerical estimates should be treated as best current approximations rather than final counts.
Defining “ultra-processed” itself introduces variability. The NOVA system, used by most of the cited studies, classifies foods by the extent and purpose of industrial processing. Critics, including some nutrition scientists, argue that NOVA lumps together products with very different health profiles. A fortified whole-grain breakfast cereal and a sugar-laden soda both land in the same ultra-processed category. The systematic review noted heterogeneity across included studies, partly because researchers applied NOVA criteria in slightly different ways depending on available dietary data. Until classification methods are standardized and validated against biomarkers of processing, some measurement noise will persist.
The question of mechanism also remains open. Ultra-processed foods tend to be high in added sugars, sodium, and saturated fat while low in fiber and micronutrients. They often contain emulsifiers, artificial colorings, and other industrial additives whose long-term biological effects are not fully characterized. Experimental work has suggested that ultra-processed products may alter satiety signaling, promote excess calorie intake, and shift the composition of the gut microbiome. But those hypotheses have not yet been connected directly to mortality outcomes in large human trials. Whether the observed risk stems from poor nutrient profiles, specific additives, the displacement of whole foods, or some combination is not settled.
Generalizability is another concern. Most large cohorts involve middle-aged or older adults in high-income countries with relatively stable healthcare access. Ultra-processed food markets are expanding rapidly in low- and middle-income nations where baseline diets, disease burdens, and healthcare infrastructure differ in ways that could amplify or attenuate the risks seen in wealthier populations.
What this means at the grocery store and in policy
The convergence of evidence across continents, decades, and study designs points in one direction: people who derive a larger share of their calories from ultra-processed products tend to die earlier than those who eat more whole or minimally processed foods, and the relationship strengthens as intake rises. The eight-country model, the pooled cohort analyses, and the national studies in the U.S. and Europe all land on a graded risk increase with higher consumption.
For individuals, the most practical takeaway is not that every packaged food is poison. It is that the cumulative effect of a diet dominated by ultra-processed products appears to carry real, measurable risk. Shifting even a portion of daily calories toward whole grains, fruits, vegetables, legumes, and minimally processed proteins is unlikely to cause harm and may reduce the chance of premature death, particularly when those changes also improve overall diet quality.
For policymakers, the evidence base now supports closer scrutiny of ultra-processed food availability and marketing, especially in communities where these products already dominate energy intake. Several Latin American countries have already implemented front-of-package warning labels targeting ultra-processed products; Canada adopted similar labeling rules in recent years. Whether the evidence justifies specific taxation schemes or advertising restrictions will require further work to clarify mechanisms, validate classification systems, and quantify the benefits of realistic dietary shifts at the population level.
As additional cohort data accrue and more transparent modeling efforts emerge, estimates of deaths attributable to ultra-processed food will almost certainly be updated. For now, the weight of independent evidence justifies treating high ultra-processed consumption as a credible, dose-dependent marker of elevated health risk. The science is still refining exactly how, and how much, these products shorten lives. But the direction of the evidence has not wavered.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
