News · Nutrition & Diet
Plant-based only helped if it wasn't ultra-processed
An EClinicalMedicine study of 17,374 UK Biobank adults found plant foods tracked with less weight gain, but plant-sourced ultra-processed foods tracked with more.
Based on a peer-reviewed cohort study in EClinicalMedicine
- Researchers followed 17,374 UK Biobank participants aged 40-69 for a mean 5.3 years, publishing in EClinicalMedicine.
- Diet came from at least two and up to five repeated 24-hour recalls, which is better than the single food-frequency questionnaire most cohorts rely on.
- Foods were sorted twice over: by origin (plant or animal) and by processing level using the Nova system.
- A 10% higher share of energy from plant-sourced non-ultra-processed food tracked with 11% lower hazard of gaining 5% or more body mass index.
- A 10% higher share from plant-sourced ultra-processed food tracked with 10% higher hazard of the same outcome.
- Waist circumference followed the same split, at 10% lower and 9% higher respectively.
- Swapping ultra-processed food of any origin for whole plant food was consistently linked to lower risk.
- This is observational, and people eating more whole plant food differ from people eating more packaged plant food in many other ways.
The plant-based label has done an enormous amount of commercial work on the assumption that plant origin is the thing that matters. A study in EClinicalMedicine sorted food twice over, by origin and by how heavily it had been processed, and found the second axis pointing the opposite way from the first.
A 10% higher energy intake from plant-sourced non-ultra-processed food was associated with 11% lower hazard of a 5% or greater increase in body mass index. The same share from plant-sourced ultra-processed food went the other way.
The concern the study was built around
The premise being tested is a commercial one. Although plant-based diets are often promoted for health and sustainability, the rising inclusion of ultra-processed foods within these diets raises concerns.
Until now those two properties had mostly been studied apart: research on plant-based eating, and separate research on processing. This study assessed the associations of food origin and processing level with weight and waist gain in a single analysis, which is what lets them be compared.
Two axes, five years, one cohort
A total of 17,374 participants aged 40-69 years from the UK Biobank cohort study were included.
The diet measurement is stronger than the norm. Intake was assessed using a minimum of two and a maximum of five repeated 24-h dietary recalls, rather than the single food-frequency questionnaire most cohorts settle for. Foods were then classified by plant or animal origin and processing level using the Nova system.
Outcomes were deliberately blunt: increases of 5% or more, and 10% or more, in body mass index and waist size over a mean follow-up of 5.3 years.
Where the two axes disagree
Whole plant food behaved as expected. A 10% higher share of energy from plant-sourced non-ultra-processed food tracked with 11% lower hazard of a 5% body mass index increase and 10% lower hazard of a 5% waist increase.
Processed plant food did the reverse. A 10% higher intake of plant-sourced ultra-processed food was linked to increased risk of both, by roughly a tenth in each case.
Same origin. Opposite direction. That is the entire finding, and it is not what the label on a plant-based ready meal implies.
What the swap analysis adds
The more useful analysis asks what happens when one food displaces another. The substitution analysis indicated that a higher contribution of plant-sourced non-ultra-processed food, at the expense of ultra-processed food of either origin, was consistently associated with a lower risk.
That framing matters because nobody eats in percentages. The practical version is replacing the packaged thing with the whole thing, and the direction held whether the food being displaced was plant or animal.
What the UK Biobank data cannot show
Nobody was assigned a diet. People who eat more whole plant food are wealthier, more active and less likely to smoke, and adjustment reaches only as far as what was recorded.
Five years is also short for weight change, and the outcomes are thresholds rather than continuous measures. Someone gaining 4.9% counts as unchanged.
The authors keep the claim to the comparison they made: plant-sourced non-ultra-processed food may confer protective effects, whereas plant-sourced ultra-processed food contribute to adverse weight outcomes.
Which lands close to advice that predates the plant-based aisle entirely: the less a food has been rebuilt before it reaches you, the better it tends to do.
People also ask
What counts as a plant-sourced ultra-processed food?
Plant-based sausages and burgers, packaged vegan desserts, sweetened plant milks, most meat-free ready meals, many protein bars and soft drinks. The Nova system classifies by degree of industrial processing rather than by ingredient origin, so a food can be entirely plant-derived and still land in the most processed category.
Does this mean plant-based diets do not work?
No, and the study makes the opposite point. Plant-sourced non-ultra-processed food was linked to lower risk of weight and waist gain. What it argues against is treating plant origin as sufficient on its own. Origin and processing turned out to be separate axes, and the study found both mattered.
How good is the diet measurement here?
Better than usual. Most large cohorts use one food-frequency questionnaire, which asks people to recall a year of eating in a single sitting. This used a minimum of two and a maximum of five repeated 24-hour dietary recalls, which captures actual days rather than an impression of a year. It still relies on self-report.
Could it just be that healthier people eat whole plant foods?
That is the standing problem with nutrition cohorts and it is not solved here. People eating more whole plant food tend to be wealthier, more active and less likely to smoke, and adjustment can only account for what was measured. The substitution analysis helps a little, because it compares people at similar overall intake, but nobody was assigned a diet.
What is the practical upshot?
The study points at swapping ultra-processed food, whatever its origin, for whole plant food rather than at plant-based products as a category. That is closer to conventional dietary advice than to the marketing on the packaging. It is general information rather than a prescription, and it describes population averages, not individuals.