Published in Cell Metabolism in 2019, the Kevin Hall NIH trial is often cited as proof that ultra-processed food causes obesity. However, the study''s own data reveals an important nuance: weight changes were highly correlated with energy intake (r = 0.8, p < 0.0001).

This means that the weight gain observed on the ultra-processed diet was almost entirely explained by the fact that participants ate more calories. The study found no evidence of a metabolic effect from processing itself - participants did not gain more weight per calorie consumed on the ultra-processed diet. If calorie intake had been held constant (rather than allowing people to eat as much as they wanted), the processing level would likely have made no difference to weight.

This distinction matters because it suggests the solution to UPF-related weight gain is portion control and calorie awareness, not necessarily avoiding all processed food. The processing may make overeating easier, but calories remain the direct mechanism of weight gain.

Published in the American Journal of Clinical Nutrition in 2022, nutrition scientist Arne Astrup argued that the concept of "ultra-processed food" does not add useful information beyond what we already know about diet quality from conventional nutritional analysis.

Astrup pointed out that the NOVA classification system relies on poorly defined food processing categories and the presence of food additives from a chemically diverse group. He noted that observational studies linking ultra-processed food to obesity have not demonstrated independent associations after controlling for likely confounders such as energy density (calories per gram), nutrient composition, and food structure.

He concluded that the NOVA system should be considered "largely redundant" outside of research settings, because traditional dietary guidance focused on nutrients, calories, and food groups already captures the relevant information without needing to classify foods by their manufacturing process.

Published in Current Developments in Nutrition in 2019, Mike Gibney analyzed published data from national dietary surveys in the US, UK, France, Brazil, and Canada to assess whether ultra-processed food intake actually predicts body weight.

Two national surveys in the UK and France failed to show any association between BMI and consumption of ultra-processed foods. Across quintiles of UPF intake (from lowest to highest consumers), nutritionally meaningful changes were seen only for sugars and fiber, not for total fat, saturated fat, or sodium.

Gibney also highlighted that the definition of ultra-processed food has varied considerably since its introduction, with little consistency across studies. This inconsistency makes it difficult to draw firm conclusions, because different researchers may classify the same food differently, and the category lumps together very different products like fortified breakfast cereals and sugary sodas.

Published in the American Journal of Medicine in 2014, Ladabaum and colleagues analyzed over 20 years of data from the National Health and Nutrition Examination Survey (NHANES) to understand what drove the rise in US obesity rates.

Physical inactivity among US women surged from 19.1% in 1994 to 51.7% in 2010. Among men, it rose from 11.4% to 43.5%. During the same period, obesity climbed from 25% to 35% in women. However, total daily calorie intake did not change significantly over this period.

The authors concluded that the rise in obesity is more strongly associated with declining physical activity than with changes in caloric intake. This challenges the idea that food processing is the main driver, since physical activity levels changed dramatically during the obesity epidemic while total food consumption remained relatively stable.

Published in the New England Journal of Medicine in 1990, Stunkard and colleagues used data from the Swedish Adoption/Twin Study of Aging to separate genetic from environmental influences on body weight. They compared BMI in identical twins who had been raised in different families.

Identical twins reared apart had BMIs as similar as twins raised together, indicating that genetics, not the shared family food environment, drives body weight. Genetic factors accounted for as much as 70% of the variance in BMI. The childhood family environment - including what food was available at home - had little or no measurable influence on adult body weight.

A related 1986 adoption study of 540 Danish adoptees found a strong relationship between the adoptee''s weight and their biological parents'' BMI (statistically significant for mothers) but no relationship with their adoptive parents'' BMI. This suggests that individual genetic susceptibility, not dietary environment, is the dominant factor in determining who becomes obese.

Published in The Journal of Nutrition in 2023, scientists from the USDA Agricultural Research Service created meal plans to test whether a nutritionally adequate diet could be built primarily from ultra-processed foods.

They demonstrated that a diet deriving 91% of its calories from ultra-processed foods could meet the 2020-2025 Dietary Guidelines for Americans, scoring 86 out of 100 on the Healthy Eating Index-2015. The only shortfalls were sodium (slightly exceeded) and whole grains (slightly below target).

The authors concluded that "the nutrient content of a food and its place in a food group are more important than the extent to which a food was processed." This directly challenges the idea that ultra-processed food is inherently obesogenic, since a diet almost entirely composed of UPF can meet nutritional standards. It suggests that specific food choices within the ultra-processed category matter more than the processing label itself.

Published in the European Journal of Clinical Nutrition in 2022, Braesco and colleagues tested how reliably different people can classify foods using the NOVA system - the classification that underpins nearly all ultra-processed food research.

When 159 evaluators classified marketed foods and 177 evaluated generic foods, inter-rater agreement was poor: the Fleiss'' kappa statistic (a measure of agreement where 1.0 means perfect agreement) was only 0.32 for marketed foods and 0.34 for generic foods. Only 3 out of all marketed foods tested were classified identically by all evaluators.

The NOVA system performed worse than two other food classification systems in terms of reliability. This is a fundamental problem because if experts cannot agree on which foods are "ultra-processed," then the research linking UPF to obesity may be measuring different things in different studies. A classification system this inconsistent is a weak foundation for public health policy or dietary recommendations.

Published in The Lancet Regional Health - Europe in 2024, Dicken and colleagues analyzed data from 311,892 individuals across 8 European countries in the EPIC cohort, followed for an average of 10.9 years with 14,236 cases of type 2 diabetes.

The results showed large differences between subgroups of ultra-processed food. Ultra-processed breads and breakfast cereals were actually inversely associated with type 2 diabetes risk (meaning they appeared protective), while sugar-sweetened beverages and animal-based ultra-processed products were positively associated with higher risk.

This directly undermines treating "ultra-processed" as a single harmful category. If some ultra-processed foods are associated with better health outcomes while others are associated with worse outcomes, then the processing level itself is not the meaningful variable. The specific ingredients, nutrient profile, and type of food matter more than whether it was manufactured in a factory.

Published in the American Journal of Public Health in 2002, Young and Nestle documented the dramatic expansion of food portion sizes in the United States and its timing relative to the obesity epidemic.

Marketplace food portions began growing in the 1970s, rose sharply in the 1980s, and continued to increase in parallel with rising body weights. Current portions exceed federal standard serving sizes across virtually all food categories. Between 1977 and 2006, increases in eating occasions and portion sizes accounted for most of the change in daily energy intake. Total daily calorie availability rose 22% from approximately 1,955 kcal/day in 1970 to approximately 2,269 kcal/day by 2010.

This timeline suggests that expanding portion sizes - which affect both processed and unprocessed foods - are a more direct driver of increased calorie consumption and obesity than the specific degree of food processing. Restaurants, fast food chains, and packaged food all increased serving sizes regardless of processing level.