The shopping rule that has taken hold over the past few years is simple. If it came out of a factory with a long ingredient list, put it back. New research presented this morning suggests that rule is doing some real work and also making some real mistakes.
Researchers at the Harvard T.H. Chan School of Public Health pooled three long-running U.S. cohort studies that together followed more than 200,000 adults from the 1980s into the 2020s. They set out to separate two things that normally travel together in nutrition data: how nutritious a person's diet was, and how much of it fell into the ultra-processed category under the NOVA classification system.
When the two were pulled apart, nutritional quality tracked with heart disease, type 2 diabetes and death. Processing level, considered on its own, did not.
That is a narrower claim than "processing does not matter," and it deserves to be read narrowly. It also cuts against a substantial body of other published work, including research this publication has covered.
Why the Two Measures Are So Hard to Separate
Most ultra-processed foods are also nutritionally poor. Soda, packaged desserts, salty snacks and processed meats are high in added sugar, sodium and refined carbohydrate. People who eat a lot of them tend to eat fewer vegetables, less fiber and less whole grain. In population data, the two measures rise and fall together, which makes it genuinely difficult to tell which one is driving the association with disease.
The NOVA system sorts foods by the extent and purpose of industrial processing rather than by nutrient content. That means the ultra-processed category is broader than its worst members. Canned beans, fortified breakfast cereal, unsweetened plant milk and packaged whole grain bread can land in the same bucket as candy and soft drinks, while a cake baked at home from refined flour, butter and sugar does not.
Xiaowen Wang, a postdoctoral research fellow in Harvard's Department of Nutrition, framed the distinction carefully rather than dismissively, saying that while many ultra-processed foods are nutritionally poor and should generally be limited, "processing itself is not inherently harmful."
Wang's practical takeaway from the analysis was to aim for more nutrient-dense foods to lower chronic disease risk, which turns a processing check into a nutrition check.
What This Analysis Cannot Settle
The limitations are significant, and they run in both directions.
This is a conference abstract. It was reviewed and selected by an expert committee but has not been through journal peer review, and the full methods are not yet available for independent scrutiny.
The cohorts are observational. They follow people who chose their own diets and report intake through periodic food frequency questionnaires, a method that captures broad patterns well and precise processing detail poorly. People who eat healthier diets also tend to smoke less, exercise more, and have more money, and while researchers adjust for those factors statistically, adjustment is never complete.
There is also a question about what the older data can measure. Cohorts that began in the 1980s were not designed around a classification system published in 2010, so the ultra-processed category has to be reconstructed retrospectively, and the products themselves have changed enormously across four decades.
Most importantly, other evidence points the other way. A controlled feeding trial conducted by researchers at the National Institutes of Health found that when adults were given nutrient-matched diets in a metabolic ward, they ate roughly 500 more calories a day and gained weight on the ultra-processed menu. That is a causal design, and observational cohorts cannot overturn it. MedicalDaily has separately reported on blood metabolite research linking higher ultra-processed intake to markers of inflammation and disrupted fat metabolism, and on cognitive findings that persisted after accounting for overall diet quality.
The honest reading is that processing and nutrient quality are two overlapping problems, that this analysis found nutrient quality doing more of the work in these cohorts, and that the case is not closed.
What This Changes in the Grocery Aisle
For a shopper, the useful shift is from one screening question to two.
The first stays where it was. Is this mostly sugar, refined starch, sodium and saturated fat, with little fiber or protein? That describes soda, packaged sweets, most chips, sweetened cereals and processed meats, and nothing in this research is a reason to buy more of them.
The second is the question this analysis supports adding. Does a long ingredient list, by itself, make this a bad choice? Canned beans, frozen vegetables, plain canned fish, higher-fiber fortified cereal, whole grain bread and unsweetened plant milk are convenient, affordable, shelf-stable and nutritionally useful. Treating them as equivalent to candy tends to push cost and preparation time up while pushing nutrition down.
That distinction matters most where a strict whole-foods-only approach is not realistic. A separate analysis presented to the same society in 2024 found that building a menu around minimally processed foods cost more than twice as much and reached its expiration date about three times faster, without delivering added nutritional value.
The Nutrition Facts panel and the ingredient list together answer the first question. Added sugar, sodium, saturated fat, fiber and protein are the lines that carry information. The number of ingredients, by itself, is not.
Who Should Pay the Most Attention
People managing high blood pressure, high cholesterol, prediabetes or type 2 diabetes get the most from focusing on nutrient content, because the targets their clinicians set are expressed in sodium, saturated fat, fiber and added sugar rather than in NOVA groups.
Families on tight food budgets, people without reliable kitchen access, and caregivers cooking for older adults with limited appetite or chewing difficulty are the groups most penalized by a blanket avoid-all-packaged-food rule.
Nobody should read this as permission to increase soda or packaged desserts. Both the processing literature and the nutrient literature agree on those.
What Happens Next
The research team's next step is peer-reviewed publication, which would open the category definitions and statistical methods to independent review. Larger controlled feeding trials examining processing independent of nutrient content are underway in several countries and will do more to resolve the question than any observational analysis can.
Federal dietary guidance in the United States currently addresses nutrients and food groups rather than processing categories, and whether that changes is an open policy question.
The newest finding is that in three large American cohorts, nutritional quality predicted disease risk while processing level alone did not. The most reasonable response is to keep limiting the foods both camps agree on and stop treating a can of beans like a candy bar. The central uncertainty is whether this holds up in peer review and in controlled trials.
Frequently Asked Questions
What did the researchers find? Across more than 200,000 adults followed for decades, nutritional quality tracked with heart disease, diabetes and death, while the level of food processing considered on its own did not.
Does this mean ultra-processed food is fine? No. Many ultra-processed products are nutritionally poor, and the researchers say those should still be limited. The finding is that the processing label alone is a weaker guide than nutrient content.
Has it been peer reviewed? No. It is a conference abstract presented at NUTRITION 2026 and should be treated as preliminary.
Does other research disagree? Yes. A controlled NIH feeding trial found people ate about 500 more calories a day on an ultra-processed diet, and other work links higher intake to inflammation markers and cognitive decline.
What is the NOVA system? A classification that sorts foods by the extent and purpose of industrial processing rather than by nutrient content, which is why canned beans and candy can share a category.
What should I look at on the label? Added sugar, sodium, saturated fat, fiber and protein. Ingredient count by itself carries little information.
Which packaged foods are worth keeping? Canned beans, frozen vegetables, plain canned fish, unsweetened plant milk and higher-fiber whole grain products are inexpensive, keep well and are nutritionally useful.
