Back

Is energy density an implicit part of the purpose behind ultra-processed food? A comparison of relative energy density across ultra-processed and minimally processed food.

Dicken, S. J.; Brown, A.

2025-10-22 nutrition
10.1101/2025.10.21.25338444 medRxiv
Show abstract

IntroductionDiets high in ultra-processed food (UPF) are linked with greater energy intake and less favourable weight change than diets high in minimally processed food (MPF), as defined by the Nova classification. Energy density is an important factor influencing palatability and energy intake and suggested as a key mechanism of UPF. Whether there are consistent differences in energy density between MPF and UPF across different types of food and drink is unknown. MethodsThis study compared the energy density of like-for-like MPF and UPF within food and drink groups from the nationally representative UK National Diet and Nutrition Survey (NDNS) Intake24 food and drink database, as well as within healthy items only (containing no red front-of-package label traffic lights). ResultsOverall, UPF food had a significantly higher energy density than MPF food (1.03 kcal/g vs. 2.07 kcal/g; p < 0.001). Most UPF (breakfast cereals, puddings, yogurt, fromage frais and other dairy desserts, egg end egg dishes, meat, fish and meat/fish dishes, fruit and vegetables, and potato and potato products) had a significantly higher energy density than MPF from the same food group. Overall, UPF drinks had a significantly lower energy density than MPF drinks (0.20 kcal/g vs. 0.39 kcal/g; p = 0.003). Findings were similar when assessing healthy food and drink only. ConclusionThese results suggest that energy density may be an implicit part of UPF and have implications for understanding the role of energy density in the context of the purpose of food processing. These findings call for caution in over adjusting for energy density in studies of UPF.

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.