Papers arguing against the standard definition of ultra-processed foods overwhelmingly came from authors with food-industry connections, according to a new analysis that found industry ties in nine out of ten such papers. The result is likely to intensify an already heated scientific and policy debate. At stake is more than terminology: the definition in question underpins research on diet and health, informs labelling proposals in several countries, and could eventually affect how products are marketed. The analysis does not claim that linked authors wrote anything false. Its point is about the composition of the published opposition, and whether that opposition reflects the wider scientific community or a narrower, commercially motivated slice of it.
Counting the connections
The definition comes from the NOVA classification, created by researchers at the University of Sao Paulo, which sorts foods into four groups according to processing. Group four, ultra-processed foods, covers industrial formulations such as fizzy drinks, confectionery, packaged snacks, instant noodles and many ready-to-heat products. Epidemiological studies using NOVA have repeatedly associated diets high in these foods with higher risks of obesity and chronic disease, though such studies show association rather than proof of cause. Critics say the grouping is imprecise, lumping together products with different ingredients and effects, and that it can confuse consumers who might reject nutritious processed foods alongside genuinely poor ones.
The classification under dispute
The new study examined papers that disputed the definition and checked their authors for industry relationships, including funding, board roles, consultancy and employment. Nine in ten had at least one such tie. Researchers in science-policy debates will find the pattern familiar: analyses of past controversies have found similar concentrations of interested authorship. The authors of the analysis argue that journals and readers should treat disclosure as essential context, not as a footnote. They also note an asymmetry: papers supporting the NOVA definition were far less likely to report commercial backing, since there is no obvious industry that profits from a stricter classification of processed products.
Disclosure and the scientific record
Why does any of this matter beyond academia? Because definitions migrate into law. Chile, Mexico and other countries already use processing-related criteria in warning-label systems, and international guidance from the World Health Organization informs national debates. In the United States, federal agencies have been examining how ultra-processed food might feature in dietary guidelines, and the Harvard T.H. Chan School of Public Health is among the institutions publishing accessible explainers on the evidence. If a definition is adopted, billions of dollars in product sales can be affected by reformulation, labelling or advertising restrictions, which explains why the academic argument attracts sustained, well-funded attention from industry.
Reading research as a consumer
None of this means consumers should dismiss every critical paper. Some criticisms of NOVA are substantive: the system can classify a wholemeal packaged bread and a sugary drink as related, and researchers across the spectrum agree classification should keep improving. Nor does an industry tie prove bad faith; companies employ capable scientists, and collaboration can produce useful data. The analysis instead supports a habit of informed reading. Check who funded a study, whether ties were disclosed, and whether independent groups have replicated the claims. We made a similar point in this earlier report: evidence deserves trust in proportion to its transparency and its fit with the rest of the literature, not in proportion to how loudly it is promoted.
The study also lands amid a broader reckoning with conflicts of interest in nutrition science, a field where research budgets are small compared with the commercial stakes and where industry funding has historically been common. Some journals now require detailed competing-interest statements, and funders increasingly expect them. Even so, enforcement varies, and older papers in the debate were published under looser norms. The authors suggest that systematic checks, like the one they performed, should become routine when a literature appears polarised, so that apparent scientific disagreement can be distinguished from organised advocacy wearing scientific dress. That recommendation is likely to be debated as vigorously as NOVA itself.
For everyday shoppers, the takeaway is calm rather than alarmist. The analysis changes how much confidence to place in claims that ‘scientists are divided’ over the ultra-processed definition; it appears the division in print overstates the division in the wider field. It does not, by itself, prove what any individual food does to health, and it should not be used to frighten anyone about a particular product. The mainstream advice remains unglamorous: build meals mostly from minimally processed ingredients, treat ultra-processed products as occasional rather than foundational, and be sceptical of dramatic claims from any direction, especially when money is attached to them.
