Exploring an energy-density adjustment to the Nutri-Score algorithm for general foods.

PURPOSE: The Nutri-Score categorizes foods by nutritional value per 100 g or 100 ml, which may lead to poor discriminatory ability for products with low energy density and large portion sizes, like ready-to-eat meals. We hypothesized that adapting the Nutri-Score algorithm to account for variation in energy density could better capture nutritional quality, particularly in lower energy-dense foods like ready meals, without substantially affecting other food scores. The aim of this study was to perform a proof‑of‑concept evaluation of adapting the Nutri-Score algorithm for general foods to different levels of energy density. METHODS: This study utilized the Norwegian food databases Tradesolution and Unil (N = 25,813) to compare the energy-adjusted algorithm with the Nutri-Score 2023. RESULTS: The energy-adjusted Nutri-Score shifted food categorizations for 22% of products. Foods with low energy density, such as ready meals, showed increased scoring variation, indicating a more nuanced nutritional quality assessment, as intended. In these foods, we also saw a decline in favourable scores and an increase in unfavourable ones. However, high-energy-density food categories, like cakes and pastries, more often received more favourable scores. Also, the discriminatory ability between low-fat and full-fat products was reduced. CONCLUSION: While the energy-adjusted Nutri-Score improved scoring differentiation for certain low-energy-density foods, it showed unfavourable side-effects, indicating that overall, it is not superior to the Nutri-Score 2023. Nonetheless, adapting the algorithm to energy density offers potential for improved nutrient profiling methodologies in specific contexts. Further refinement of the weighting system could address observed weaknesses, optimizing this approach for broader application.