Ultra-processed foods and overeating through the trial evidence

Ultra-processed diets can increase food intake, but results depend on the meals. Read the trials, exceptions and limits before changing your shopping.

日本語で読む（Japanese version）

The meals offered had matching calories and nutrients, yet the amount people actually ate differed by 508 kcal a day. That was the finding in a controlled feeding trial comparing an ultra-processed diet with an unprocessed diet.[1] A nutrition label can help you choose what to put on the table. It cannot tell you how much you will eat from that table, or whether you will reach for more afterward.

The participants stayed in a research facility and could eat as much or as little as they wanted. Researchers arranged the meals to make the comparison more controlled, then measured what people consumed.[1] The gap between the food provided and the food eaten is the important starting point. It directs attention toward the whole meal, including its texture and how easily it can be eaten, alongside its nutritional composition.

But the statement that eating ultra-processed foods inevitably makes you gain weight goes further than the evidence. In another trial, participants lost weight even on an ultra-processed diet when meals were arranged around dietary guidelines.[5] For anyone fitting dinner around work, school or a journey home through Tokyo, the useful question is what these studies actually compared. Understanding the meals, the participants and the limits gives us a firmer basis for changing everyday food choices than a blanket verdict on a category.

Matching the meals offered did not match the amount eaten

Hall and colleagues studied 20 adults whose weight had been stable. Their average age was 31.2 years and their average body mass index, or BMI, was 27. BMI describes body weight in relation to height. Participants were admitted to a research facility and randomly assigned to begin with either an ultra-processed or an unprocessed diet. They ate that diet for 2 weeks, then immediately switched to the other diet for another 2 weeks.[1]

This is a crossover design: the same person experiences both conditions. Rather than comparing people who normally shop and eat differently, the researchers could compare each participant’s response to the meals provided. That makes the arrangement useful for testing whether changing the diet changes intake under controlled conditions. It also means the result belongs to these particular meals and periods, rather than to every possible product carrying the same processing classification.[1]

Here, ultra-processed foods refers to a category used in research on the extent of food processing. The NOVA classification distinguishes processing categories, but foods within a category can differ in how quickly they are eaten and how much energy they deliver.[9] Whether something comes in a package or was prepared at home is therefore an inadequate shortcut for interpreting the studies. A category can identify a pattern worth investigating without describing every feature of an individual meal.

The diets in Hall’s trial were designed to match the calories presented, energy density, macronutrients, sugar, sodium and fiber. Macronutrients are the main nutrient groups supplying energy; energy density describes the amount of energy relative to the quantity of food. Participants were then told to eat as much or as little as they wanted.[1] The study did not require them to finish a fixed portion. It measured the amount they chose to consume from the food available.

On the ultra-processed diet, average intake was 508 ± 106 kcal per day higher. This included an extra 280 ± 54 kcal from carbohydrate and 230 ± 53 kcal from fat. The difference in energy from protein was 2 ± 12 kcal and was not statistically significant.[1] The extra intake therefore came from carbohydrate and fat, while the protein result did not show a significant difference. The difference in consumption was not shared equally across the nutrients.

The distinction between offering and eating makes the result easier to understand. A meal plan specifies the food made available. What is left behind and what is eaten determines actual intake. In this trial, matching the planned nutritional characteristics did not prevent a difference in the energy consumed.[1] For a household, that is a reason to look beyond the shopping basket and include the meal itself and any additional food afterward when thinking about eating patterns.

Participants gained 0.9 ± 0.3 kg during the ultra-processed period and lost 0.9 ± 0.3 kg during the unprocessed period. Weight changes were strongly correlated with energy intake.[1] These were changes in body weight during each 2-week diet period. They should not all be described as changes in body fat. A number on the scale and a measurement of body fat are different outcomes, even when a study reports a clear difference in weight.

The controlled setting is both a strength and a limit. It makes it easier to investigate the response to the diets than a questionnaire about usual eating would. However, the trial was small and took place in an inpatient facility.[1] It did not reproduce a person’s freedom to choose meals around overtime, family preferences, food costs or the effort of cooking. The finding gives a strong reason to examine meal design; it does not provide an exact forecast for a particular family’s dinner routine.

A Japanese trial found a difference in a small, selected group

Hamano and colleagues compared ultra-processed foods with non-ultra-processed foods in Japanese men with overweight or obesity. Participants were randomly assigned to the order of the diets. Each diet lasted 1 week, with a 2-week washout period between them. Nine men took part and all completed the trial. The meals were designed to match total energy and macronutrients.[4] This was an open-label crossover study, meaning participants could know which diet they were receiving.

The difference in weight change between the diet periods was 1.1 kg, with a 95% confidence interval of 0.2–2.0 kg, in the direction of greater weight gain during the ultra-processed period. Daily energy intake was 813.5 kcal higher in that period, with a 95% confidence interval of 342.4–1284.7 kcal.[4] In these nine men, the estimated intake difference was 813.5 kcal per day, but the interval of 342.4–1284.7 kcal shows how imprecisely its size was estimated.[4]

The direction agrees with the earlier inpatient trial, but the studies had different participants, meals and durations.[1][4] It would be misleading to take the larger daily intake difference here as the amount someone in Japan should expect from changing their own diet. The result describes this comparison in this group. The shared direction is informative, while the differences in study design still matter when deciding how far to carry either number into ordinary life.

A Japanese sample makes the research particularly relevant to readers living here, but it does not make the sample representative of everyone in Japan. Participants were men with overweight or obesity. People regularly attending hospital for diabetes, hypertension or other medical conditions were excluded.[4] The study therefore cannot by itself establish the response in women, people of different body sizes or people receiving regular care for those conditions. Familiarity of location should not replace attention to eligibility criteria.

Researchers also measured chewing, expressed as the number of chews per calorie consumed. This was lower during the ultra-processed period.[4] Alongside the differences in intake and weight, there was a difference in how food was processed in the mouth. That provides another feature of the meals to investigate. It does not establish that fewer chews alone caused the entire increase in intake; the food conditions changed together, and the study measured several responses to that change.

At home, this suggests a more useful question than simply asking whether dinner included packaged food. What textures were present, and how did the meal unfold? Was there something that required chewing, or was most of the meal easy to finish quickly? These questions broaden attention from the names of foods to the experience of eating the whole plate. They are ways to observe a meal, rather than a claim that a particular chewing target has been tested here.

The trial changed the foods provided and measured intake and chewing. It did not instruct participants to reach a prescribed number of chews.[4] That distinction protects against turning a measurement into a rule. A finding about how people ate two different diets cannot automatically tell us how many times to chew each mouthful. It can, however, justify examining texture and the ease of eating when choosing a main dish and the foods served with it.

Food texture adds information that a processing label misses

Forde and colleagues examined processing categories alongside energy intake rate. This measure combines eating rate with energy density to describe how much energy enters the mouth per unit of time. Being able to eat food quickly and having a lot of energy in a small amount of food are separate characteristics. Considering them together helps explain why two meals with similar labels may deliver energy differently while they are being eaten.[9]

The researchers pooled data on 327 foods from 5 published studies. Average energy intake rates were 35.5 kcal per minute for unprocessed foods, 53.7 for processed foods and 69.4 for ultra-processed foods. There was substantial variation within every category.[9] The category averages point toward a pattern, but the variation prevents us from treating all foods in a group as equally easy to consume. The food’s texture and energy density remain relevant features.

This was an analysis of food measurement data, not a trial that used those average rates to predict daily weight change.[9] A reader should not multiply a category average by the length of dinner and treat the answer as a personal calorie calculation. The analysis shows differences between categories and diversity within them. Its practical value is to make the processing label less of an endpoint and more of a starting point for looking at the food itself.

For a purchased dinner, that can mean looking across the main dish, the staple and the sides. Which parts are easy to finish, and which remain in the mouth long enough to require chewing? Observing these differences may help connect the meal with the experience of wanting more afterward. The pooled analysis supports looking for differences within processing categories.[9]

The wider question of deliberately changing eating speed is covered in the related feature on chewing, eating rate and the evidence on meal size and weight. Here the issue is whether processing alone explains a comparison. Replacing one diet with another can change texture and the way energy is consumed alongside nutritional composition.[9] Looking at these characteristics together gives a clearer account of what may differ between the meals.

Another trial tested texture while keeping both diets ultra-processed. Forde and colleagues studied 41 participants who experienced each diet for 14 days. The foods were selected for textures associated with slower or faster eating. The diets were matched for palatability, portion size served, total energy served, energy density excluding beverages, and meal variety. Participants’ average age was 27, average BMI was 23.4, and 21 were men.[15]

Daily energy intake was 369 kcal lower with the slower-eating textures, with a 95% confidence interval of 221–517 kcal. The effect on intake did not change significantly with the number of days on the diet and was not explained by meal liking or macronutrient intake.[15] In this comparison, a difference in the way food was eaten accompanied a sustained difference in consumption, even though the processing category was the same in both conditions.

The weight findings need equal attention. Body weight did not change from before to after the diets, and there was no difference in weight between the two conditions. Body fat mass decreased by 0.43 kg on the slower-eating diet.[15] Lower energy intake, body weight and body fat mass are distinct outcomes. Reporting only the intake reduction would hide the neutral weight finding; reporting only the scale result would miss the difference in intake and the reported body-fat result.

The intervention changed food textures rather than telling people to consciously increase their chewing count.[15] That offers a different way to think about a hurried evening meal: examine the combination of foods, as well as the intention to slow down. Because both diets remained ultra-processed, the trial also shows why a blanket rule about processing can miss features worth changing on days when purchased foods are part of dinner. It does not guarantee the same intake difference in a household.

Numbers: ultra-processed diet comparisons

20 adults experienced each inpatient diet for 2 weeks in a crossover trial.[1]

508 ± 106 kcal per day was the extra intake on the ultra-processed diet.[1]

Weight rose by 0.9 kg on one diet and fell by 0.9 kg on the other in that trial.[1]

9 Japanese men had an intake difference of 813.5 kcal per day, with a 95% confidence interval of 342.4–1284.7.[4]

327 foods were analyzed, with variation in energy intake rate within every processing category.[9]