A new study has found that approximately nine in ten research papers disputing the widely used definition of ultra-processed foods were written by authors with ties to the food industry. The analysis reviewed papers that challenged the classification and recorded their authors’ connections to food companies, whether through employment, consultancy work or research funding.
The finding is about the conduct of a debate rather than the safety of any single product. The authors say industry-linked scientists may still make valid arguments, and that the classification they examined has genuine scientific critics. Their concern is proportion: criticism of the definition has been produced overwhelmingly from within the industry it most directly affects.
The NOVA classification explained
The definition in dispute belongs to the NOVA system, developed by researchers at the University of São Paulo in Brazil. NOVA groups foods according to the nature and extent of their processing. Its fourth group, ultra-processed foods, includes soft drinks, packaged snacks, confectionery, instant noodles and many ready meals — industrial products typically assembled from refined ingredients and additives.
Opponents of the system argue that it can be imprecise, sometimes placing nutritionally different foods in the same category. Supporters respond that processing itself influences how food is consumed and that large observational studies repeatedly associate ultra-processed-heavy diets with poorer health outcomes. International agencies such as the World Health Organization continue to assess this evidence, and none of the major bodies treats the question as fully settled.
Reading the result carefully
Sponsor influence in science usually works quietly. It shapes which questions are studied, which results are promoted and how confidently conclusions are stated, rather than producing openly false findings. In earlier controversies involving tobacco and sugar, similar audits found that industry-linked literature leaned consistently toward doubt — a pattern that delayed regulation without ever winning the scientific argument outright.
That is why disclosure rules matter and why synthesising bodies, including the Harvard T.H. Chan School of Public Health, base guidance on the whole evidence base rather than selected papers. A reader who checks the funding statement of a striking headline is doing exactly what journal editors intend.
How such audits are done
Studies of this kind are built by hand. Researchers define the set of papers that dispute the classification, retrieve each paper’s disclosure statements and author affiliations, and code the connections they find: employment by a food company, consultancy roles, grants, or membership of industry-funded institutes. Disagreements between coders are reconciled, and the results are reported as proportions — which is where the nine-in-ten figure comes from. The method cannot capture undisclosed ties, so if anything it tends to understate the network it maps.
The approach has a track record. Bibliometric audits were among the tools that clarified the tobacco industry’s role in manufacturing doubt about smoking, decades after the fact. Applying them in real time, while a policy debate is still live, is the newer development — and nutrition, where commercial stakes are immediate and the science is genuinely complex, is becoming their main testing ground.
Consequences for labels and consumers
The definition matters commercially because it is becoming law in some places. Several countries have introduced warning labels on packaged products, and in the United States labelling policy is led by the US Food and Drug Administration. Where ultra-processed becomes an official category, manufacturers face reformulation and packaging costs, which explains the intensity of the scientific argument around it.
For consumers, the study’s message is one of informed caution. It does not demonstrate that a particular food causes a particular disease, and it does not end reasonable debate about NOVA’s boundaries. It does show who has financed most of the opposition to the definition — information that belongs in any honest account of the controversy, as our earlier features this recent article and this related explainer have argued in related contexts.
The authors also note a limitation of their own work that deserves repeating: disclosure statements are an imperfect record, practices differ between journals and countries, and some relevant ties will inevitably have been missed by any audit of published declarations alone.
