Trials link ultra-processed food to weight gain — but maybe via energy density

Ultra-processed food has become the most-blamed category in modern diets, but most of the case against it rests on observational studies that cannot separate the processing from everything else about the way people who eat it live. Randomised trials — where the same people are fed different diets and their intake is measured — are the harder evidence, and there have not been many. A new pre-registered systematic review pooled the ones that exist, and its conclusion is more careful than the usual framing [s1].

What the review did

The analysis, from a group at the University of Liverpool, gathered randomised controlled trials comparing ultra-processed food (UPF) with less-processed food (LPF) on energy intake and body weight in humans [s1]. Ten eligible studies were included [s1]. Beyond the headline comparison, the authors ran secondary analyses — meta-regression and subgroups — on appetite sensations, eating rate, palatability, and, crucially, on whether the nutrient profile of the diets explained any effect [s1].

That last point is the crux. If ultra-processed diets in trials happen to be more energy-dense, saltier or more sugary than the comparison diets, then a difference in intake could be about those properties rather than about "processing" as such.

The numbers

Ultra-processed trial arms tended to have higher energy intake, with standardised mean differences ranging from 0.18 to 0.44 — but the statistical significance varied depending on which analytic model was used [s1]. Two outcomes were more clear-cut: weight gain was significantly greater on the ultra-processed diets (standardised mean difference 0.65), as was eating rate (standardised mean difference 0.96) [s1]. There were no significant differences in palatability, appetite sensations, or energy intake later in the day [s1].

Then the caveat that reframes the whole picture: the diets used in these trials were not matched for nutrient profile [s1]. When the authors split trials by energy density, the effect on energy intake was large where the ultra-processed diet was more energy-dense than the comparison (standardised mean difference 0.71) and essentially absent where the two were similar in energy density (standardised mean difference 0.02) [s1].

The authors' bottom line is hedged accordingly: ultra-processed foods may increase energy intake and body weight, but the results may be explained by the energy density of the foods used in the trials to date, and the overall certainty of the evidence is low [s1].

Why the energy-density point matters

The dominant story about ultra-processed food is that something about the processing itself — texture, additives, hyper-palatability, speed of eating — drives overconsumption. This review does find one mechanistic signal consistent with that: people ate faster on the ultra-processed diets [s1]. But it also shows that the intake effect largely rode on energy density, a property that is not unique to ultra-processing and can be present in home-cooked food too. If the calories-per-gram are matched, most of the intake difference in these trials goes away [s1].

That does not exonerate ultra-processed food. Weight gain and faster eating were still greater on the ultra-processed arms, and energy density and ultra-processing are entangled in the real food supply for a reason [s1]. But it does mean the trials done so far cannot cleanly separate "processing" from "energy density," and that anyone citing this literature as proof that processing per se drives overeating is going beyond what the randomised evidence supports.

The low certainty grade is worth dwelling on too. It reflects not just the small number of eligible trials — ten — but their short duration, their modest sizes and the fact that they were not built to isolate processing from nutrient content in the first place [s1]. A pooled estimate can look precise while resting on studies that individually cannot answer the question, and the authors are careful not to oversell it. The absence of any significant difference in palatability, appetite sensations or later-in-the-day energy intake also complicates the tidy "hyper-palatable food overrides fullness" narrative, at least as those constructs were measured across these trials [s1].

The wider picture

The theme recurs across 2026's nutrition trials. A separate randomised study in patients with fatty-liver disease found that improvement in liver fat tracked weight loss and reductions in ultra-processed food intake rather than the specific diet pattern assigned — again pointing at what a diet does to weight and quality rather than at a single villain nutrient or process [s2].

What to watch

The review's own recommendation is the honest one: trials that match diets on nutrient profile, so that level of processing is the only thing that differs, are what is needed to say whether ultra-processing harms health independent of energy density [s1]. Until those exist, the evidence supports eating less ultra-processed food as sensible, without claiming the mechanism is settled.

This article is informational and does not constitute medical advice.

Sources

• [s1] A systematic review and meta-analysis of randomized controlled trials examining the effect of ultra-processed food on energy intake and weight gain. Critical Reviews in Food Science and Nutrition, 16 September 2026. https://doi.org/10.1080/10408398.2026.2731838
• [s2] Impact of weight loss and reduction of ultra-processed foods on liver fat content in MASLD: A randomized controlled trial. JHEP Reports, 22 June 2026. https://doi.org/10.1016/j.jhepr.2026.101929