Ultra-processed foods keep turning up in studies of earlier death

A growing shelf of nutrition research keeps arriving at the same uncomfortable place. When scientists sort diets by how heavily they lean on industrially manufactured food and then follow people for years, the group eating the most of it tends to die somewhat sooner. The pattern has surfaced across countries, cohorts, and research teams often enough that it has become one of the more discussed findings in modern nutrition science.

What makes the topic slippery is the category at its center. The foods in question are not defined by a single nutrient like sugar or salt but by how extensively they have been transformed and formulated, a classification that groups together an unusually broad range of products. Sorting out what the evidence really shows, and where it stops short, requires first understanding what researchers mean when they use the term.

What counts as ultra-processed

The label comes from a framework known as the NOVA classification, which sorts foods by degree of processing rather than by nutrient content. Under that system, described in overviews of ultra-processed food, the most heavily manufactured tier covers industrial formulations built largely from substances extracted or synthesized from foods — refined starches, protein isolates, hydrogenated oils — combined with additives such as emulsifiers, artificial colors, and flavor enhancers.

In practice that captures a familiar aisle: packaged snacks, sugary drinks, mass-produced breads, reconstituted meat products, instant noodles, and many ready-to-eat meals. Critics of the framework point out that it lumps a diet soda in with a frozen dinner and a supermarket loaf, products that differ enormously in nutrition, which is one reason the category remains contested even as the mortality signal attached to it strengthens.

What the mortality research shows

The consistency across studies is what draws attention. A large analysis drawing on the UK Biobank, published in the peer-reviewed literature, linked higher consumption of ultra-processed foods to greater risk of both cardiovascular disease and death from any cause, tracking a population of tens of thousands of adults over years of follow-up.

Pooling many such studies sharpens the picture. An updated systematic review and meta-analysis reported in a 2025 journal article combined 18 cohort studies covering more than a million participants and roughly 173,000 deaths, finding that those with the highest ultra-processed intake carried about a 15 percent higher risk of all-cause mortality than those with the lowest. The size of that combined dataset is part of why the association is treated as robust rather than a one-off result.

The dose-response pattern

Beyond the high-versus-low comparison, researchers have looked for a graded relationship, and they tend to find one. The 2025 meta-analysis reported that each 10 percent increase in the share of the diet made up of ultra-processed food was associated with roughly a 10 percent higher risk of death, following a positive linear trend rather than a threshold that only kicks in at extreme intake.

A dose-response gradient of this kind is one of the classic features scientists look for when weighing whether an association might reflect cause and effect, since a steadily rising risk is harder to explain away than an isolated jump. Earlier pooled work, including a dose-response meta-analysis of more than 200,000 participants summarized in the research record, pointed in the same direction, lending the pattern additional weight across different sets of studies.

Why the link is not proof

The evidence, however, has a fundamental limitation that no amount of replication removes. Nearly all of it comes from observational cohorts, which follow people who chose their own diets rather than being randomly assigned to eat one way or another. People who consume the most ultra-processed food differ from those who eat the least in many ways — income, education, physical activity, smoking, overall diet quality — and those differences can independently affect how long they live.

Researchers adjust statistically for the factors they can measure, but residual confounding always remains a live possibility, and the results are not uniform. Some analyses have found no association between total ultra-processed intake and death once specific subgroups are separated out, with the risk concentrated in particular products such as processed meats rather than spread evenly across the whole category. That heterogeneity is a caution against treating every item under the label as equally hazardous.

Reading the evidence sensibly

The mechanisms proposed to explain the link are plausible but not fully proven. Ultra-processed foods tend to be energy-dense, high in refined carbohydrates, sugar, and salt, and low in fiber, and their engineered textures may encourage overeating; some additives and processing byproducts have drawn scrutiny of their own. A small number of controlled feeding experiments have suggested that people eat more calories when offered ultra-processed meals, offering one concrete pathway, though such studies are short and small.

Taken together, the research supports a measured conclusion rather than a panic. The signal linking diets dominated by ultra-processed food to earlier death is consistent, graded, and reproduced across large populations, which is enough for many public-health bodies to advise shifting toward whole and minimally processed foods where practical. What the evidence does not establish is that any single packaged item is dangerous, or that processing itself, independent of the nutrition profile it usually carries, is the culprit. The honest summary is that the pattern keeps appearing and deserves attention, while the precise cause remains an open scientific question rather than a settled verdict.

This article was produced with the assistance of AI and reviewed by Morning Overview editors prior to publication.

More from Morning Overview