A secondary analysis of a 62-person crossover trial found that minimally processed animal foods were associated with weight gain — the only association that survived statistical correction.

Ultra-processed food has become the organising idea in nutrition research. The NOVA classification system, which sorts foods by how industrially processed they are rather than by what nutrients they contain, has produced a decade of findings linking the most processed category to weight gain and chronic disease, and it now shapes dietary guidelines in several countries.

An analysis in Nutrition Research asks whether that single axis is the right one. Reclassifying the food records from an earlier randomised trial by both processing level and biological origin, it finds that where a food came from — plant or animal — tracked weight change at least as closely as how processed it was.

The lead author is Hana Kahleova, director of clinical research at the Physicians Committee for Responsible Medicine.

What was reanalysed

The underlying study was a randomised crossover trial in 62 adults with overweight. Each participant followed a Mediterranean diet and a low-fat vegan diet for 16 weeks apiece, separated by a four-week washout, and kept three-day food records throughout. For this analysis every food in those records was classified twice: by NOVA processing level, and by whether it came from a plant or an animal.

A crossover design is the strength here. Because each person did both diets, each acts as their own control, which removes the between-person variation that makes nutrition studies so noisy.

The diets shifted intake substantially in both dimensions. Compared with the Mediterranean diet, the vegan diet reduced minimally processed animal foods by 135 grams a day more, and increased minimally processed plant foods by 202 grams a day more.

The one association that held

Pooling across both diets, higher intake of minimally processed animal foods was associated with weight gain. This was the strongest association in the analysis and the only one that remained statistically significant after correction for multiple comparisons.

That last clause is the most important sentence in the study, and it cuts both ways. Testing several food categories against weight change means some will reach significance by chance, and correcting for that is what separates a finding from an artefact. One association surviving is a real result. It also means the others did not.

Those others pointed in intuitive directions without clearing the bar. Processed and ultra-processed animal foods tended toward weight gain, and minimally processed plant foods toward weight loss, both weakly. Ultra-processed plant foods showed no association with weight change at all.

That last null is the most provocative number on the page, since it is the category that current dietary advice treats as straightforwardly bad.

The authors’ framing

“Ultra-processed foods get most of the blame for weight gain,” Kahleova said, “but our results suggest we also need to ask whether a food comes from a plant or an animal.”

“A chicken breast is minimally processed. So is an apple,” she said. “In our study, they did not move the scale in the same direction.”

The mechanistic explanation offered is energy density. Minimally processed plant foods tend to be high in fibre and water and low in calories per gram, so a given volume of food delivers fewer calories — which is a well-supported account of why plant-heavy diets are associated with lower intake without deliberate restriction.

In the original 2022 trial, participants lost an average of 6 kilograms, about 13 pounds, on the vegan diet and showed no change on the Mediterranean diet.

“If you want to lose weight, fill your plate with beans, vegetables, fruits, and whole grains, and leave the animal products off,” Kahleova said. “The degree of processing still matters, but it is not the whole story.”

What a secondary analysis can and cannot show

Several limits apply, and the release does not raise them.

The trial was designed to compare two diets, not to compare food categories. The associations reported here are observational relationships within trial data — participants were randomised to diets, not to eating particular amounts of minimally processed animal food, so the comparison between food categories carries none of the causal weight that the diet comparison does.

The two variables are also entangled by design. The vegan diet both removed animal foods and produced the weight loss, which means a within-person association between animal food intake and weight gain partly reflects which diet the person was on. Disentangling processing level from biological origin is hard when the intervention changed both at once.

Sixty-two people is a small sample for subdividing food into four categories, and three-day food records are self-reported. Under-reporting in dietary recall is well documented and is not random: people systematically under-record foods they regard as indulgent.

Nor does weight change over 16 weeks establish anything about health. Weight is one outcome among many, and the analysis reports no cardiometabolic measures.

Who did the work

One piece of context belongs with this study, and the release does not provide it. The Physicians Committee for Responsible Medicine is an advocacy organisation that promotes plant-based diets, and it has done so since its founding in 1985. That is a matter of public record rather than an accusation, and the trial itself was randomised and peer-reviewed.

It does mean a reader should notice that a finding favourable to plant-based eating comes from a group whose purpose is to advance plant-based eating, and should weigh the result accordingly — particularly the framing choices, such as which comparison is highlighted and which caveats are omitted. The release states neither funding sources nor conflicts of interest, which in this case is the disclosure that would have mattered most.

The underlying observation remains worth having. Processing level and biological origin are different properties of food, nutrition research has largely collapsed them into one, and a dataset in which they can be separated is a reasonable place to point that out. Nothing here is a reason for any individual to change what they eat, and questions about diet and body weight are better taken to a clinician or dietitian who knows the person asking.

The analysis was published on 22 October 2026.

---

Notes

1. Kahleova, H., et al. “Animal vs plant-based processed foods have different effects on body weight: a secondary analysis of a randomized, cross-over trial.” Nutrition Research, 2026. DOI: 10.1016/j.nutres.2026.09.010.
2. Physicians Committee for Responsible Medicine. “For weight loss, where food comes from may matter as much as how it is processed, finds new analysis.” EurekAlert!, 22 October 2026. https://www.eurekalert.org/news-releases/1148924.

Photo by Pixabay on Pexels.com