A new audit of the scientific literature on ultra-processed foods has found that around nine in ten papers disputing the standard definition of the category were authored by people with food industry connections. The study turns the spotlight from the foods themselves to the debate about them — and suggests that scepticism toward the definition has been far from evenly funded.
The researchers behind the audit are measured in their conclusions. An industry connection does not invalidate a scientist’s argument, and the classification system under discussion attracts genuine scholarly criticism. Their point is narrower and harder to dismiss: when readers meet this debate, they should know that one side of it has been financed overwhelmingly by the industry the classification most affects.
Understanding NOVA
NOVA, developed by researchers led by Carlos Monteiro at the University of São Paulo, classifies foods by how they are made rather than by nutrients alone. Group one holds unprocessed and minimally processed foods. Group four, the ultra-processed category, covers industrial formulations — soft drinks, packaged snacks, confectionery, instant noodles and many ready meals — built from refined substances and cosmetic additives.
The classification’s critics contend that it can be imprecise, placing products with quite different nutritional value in the same group and sometimes penalising convenient foods that do little harm in context. Its defenders argue that processing itself shapes eating behaviour and health, and that repeated observational associations with obesity and chronic disease deserve to be taken seriously. Global health guidance, including that of the World Health Organization, generally steers consumers toward minimally processed foods while research continues.
The pattern in the papers
What the new study adds is a map of who has been making the critical case. Reviewing papers that challenged the definition, the authors found industry ties — employment, consultancy or research funding — attached to roughly ninety percent of them. Comparable concentrations have been documented in earlier controversies over tobacco and sugar, where sponsored science functioned less to falsify than to prolong uncertainty.
That history is why disclosure rules exist, and why the Harvard T.H. Chan School of Public Health and other academic centres advise weighing whole literatures rather than single studies. A definition can still be wrong even if its critics are funded; but a shopper or a policymaker is entitled to know the funding before deciding how decisive a critique really is.
What nutrition scientists agree on
Amid the argument, a core of agreement is easy to miss. Almost nobody in nutrition science defends a diet built mainly on soft drinks, confectionery and instant snacks, whatever classification they prefer. Almost everybody accepts that observational studies cannot prove causation on their own, that overall dietary pattern matters more than any single product, and that price, time and availability shape what people actually eat at least as powerfully as information does.
The definition fight matters because policy needs categories, and categories need definitions that can survive scrutiny — including scrutiny of who funded the scrutiny. On the evidence of this latest audit, that second-order check is becoming a standard part of the process, and food policy debates from here on will be conducted with the funding map on the table.
What follows for labelling
Definitions decide labels. Several countries already print warnings on ultra-processed products, and others are weighing similar schemes; in the US, labelling sits with the US Food and Drug Administration. If the scientific definition is publicly framed as unsettled, regulation slows — an outcome with obvious commercial value, and the reason funding audits like this one attract attention beyond the journals.
For now, the consumer takeaway is restrained. The audit does not prove that ultra-processed foods cause particular diseases, and it does not absolve anyone from reading a label. It simply restores a missing piece of information to a public argument. As our recent coverage in this recent article noted, and as this related explainer also explored, evidence in nutrition builds slowly — and it builds more honestly when everyone’s interests are on the table.
