In the present study, we assessed the intake of foods according to the NOVA groups among Norwegian adults, and the contribution of each NOVA group to nutritional quality and climate impact. UPF contributed to almost half of the energy intake, while foods from NOVA 1 contributed to one third of the energy. When we re-classified bread from UPF to NOVA 3, the energy contribution from UPF was changed with 5% percentage points (from 48 to 43%). The NOVA 1 foods consumed had a significantly higher nutrient density for protein, fiber, and all micronutrients, except for sodium, compared to foods consumed within the UPF category. Conversely, the nutrient density of added sugar, total fat, SFA, MUFA, PUFA and sodium, was significantly lower in NOVA 1 foods compared to UPF. For climate impact, NOVA 1 contributed with a higher proportion of the impact from GWP in both the main analysis and the sensitivity analysis, compared to the other NOVA groups.

UPF contributed to 48% of the total energy intake, which is slightly lower than found in adult populations in the US and UK, where UPF was estimated to contribute to more than 50% of the total energy intake [33]. Two other Norwegian studies based on sales data [10] and on pregnant women [11] found similar results as the present study. NOVA 3 contributed to 19% of the total energy intake in the present study, which is in line with the 22% reported in the study of Norwegian pregnant women [11].

Of the food groups in the present study, bread contributed the most to total energy intake from UPF, followed by meat and meat products. These results contrast with findings from the rest of Europe, where fine bakery wares and soft drinks were found to be the main contributors to energy intake from UPF [34]. It is worth noting that all UPF, including energy-poor items such as sugar-free soft drinks, were included in our analyses. Although the consumption of sugar-free soft drinks is relatively high in Norway, these products contribute little to total energy intake, and their impact on the overall energy contribution from UPF is therefore minimal. Bread is a traditional staple food consumed in several meals in Norway, and especially whole grain bread has strong traditions [35, 36]. This is reflected in our results, as bread with more than 25% wholegrains contributed more to total energy intake from both NOVA 3 and UPF, compared to bread with less than 25% wholegrains.

Foods in the UPF group had higher nutrient densities for added sugar, total fat, and sodium compared to foods in NOVA 1. Foods in the NOVA 1 group had higher nutrient densities for protein, fiber, and all micronutrients except sodium. These findings are consistent with previous findings from other countries, including UK, Belgium, Australia, Mexico, Brazil, US, and Canada [4, 6, 9]. In the sensitivity analysis (Table 4), where more bread intakes were reclassified as NOVA 3, the nutrient density of the NOVA group 3 increased for fiber and iron, and decreased for sugar, fat, and sodium, giving NOVA 3 as a group an improved nutrient profile. The nutrient density for vitamin A, C, and D, as well as calcium, iodine, and selenium decreased as well, which can be explained by bread generally being low in these nutrients. UPF obtained a reduced nutrient profile in the sensitivity analysis, except for the micronutrients that bread is low on, which improved relatively for UPF (vitamin A, C, calcium, iodine and selenium). These results show that the nutrient density associated with a specific NOVA group will vary significantly depending on how the classification of foods are done.

UPF was estimated to contribute to 32% of the total GWP from the diet (median percentage). This was lower than the contribution from NOVA 1 (38%), and higher than from NOVA 3 (20%). UPF also contributed to a relatively higher percentage of energy intake, compared to the percentage of GWP. The opposite was observed for NOVA 1 (Fig. 1). These findings align with a study from the UK, which found that UPF had a lower GHGE per 100 cal compared to NOVA 1 foods [20]. However, this contradicts other research, such as a 2-year longitudinal study involving nearly 6,000 participants from Southern Europe [37]. Our observations could be due to the primary production of produce being the largest contributing part of the food production to climate impact, while post-harvest processing, transport and additional processing have a lower contributing parts of the climate impact of food [38]. This makes the type of ingredients in UPF more important than the level of processing itself. However, GWP is just one indicator of environmental impact. To fully understand how UPF contributes to the environmental footprint, other measures, such as water footprint, land use, and eutrophication must be considered. Moreover, as mentioned earlier, UPF often relies on cheap ingredients from monoculture crops, which can negatively affect environment through high pesticide use, and reduced biodiversity [18]. UPF has also been found to possibly facilitate overconsumption of energy, leading to overweight and obesity [8, 39]. This can negatively affect the overall climate impact of the diet, as overconsumption can lead to unnecessarily high food production. Additionally, a higher prevalence of overweight and obesity could increase the pressure on healthcare systems, thereby increase emissions and the use of natural resources [18, 40]. In addition, NOVA 1 foods, which are often less energy-dense than UPF, are recommended to help prevent energy overconsumption. This is important for maintaining good health and avoiding excessive weight gain, which could lead to non-communicable diseases [41, 42]. Therefore, UPF may not necessarily be a sustainable food group despite low GWP per MJ in a Norwegian setting.

A major strength of the present study was the use of data from two 24-h dietary recalls, which provide enough details about the foods to perform the NOVA classification. Many studies on UPF and NOVA classification have been conducted with dietary data derived from methods such as FFQs that do not give the detailed information needed to classify foods correctly. Another strength is the thorough and systematic approach used in the classification process, ensuring an objective classification of food items, with full disclosure of additives, ingredients, and preparation methods for each NOVA group. Moreover, the use of sensitivity analysis for bread strengthens the present study and provides insights into the uncertainties pertaining to bread classification. A potential weakness of the present study is the low sample size and the low participation rate of 45%, which may limit the generalizability of our results. We found that a higher proportion of the participants in the Norkost 4 pilot study had a higher education compared to the general population. Those who chose not to participate may have consumed other foods than what was reported by the participants.

Even though the NOVA classification system today is the most widely used, it has come under criticism in recent years [43]. NOVA has been criticized for being confusing, inconsistent, and controversial [43,44,45]. Lack of categorical specifications and conceptual inconsistencies may be challenging when applied in research and practice [46,47,48]. This was evident also in the present study, where a vast variety of products were classified as UPF by the NOVA classification system, sometimes only due to one ingredient, such as olives in brine containing stabilizer, or granola containing fiber extract. It is questionable if these foods are contributing to the negative health effects associated with UPF, as there is lacking evidence whether these ingredients such as fibers and stabilizers are causing the negative health effects [49]. This raises the question of whether the issue lies more in the specific formulation of these products, rather than the processing itself. Several foods classified as NOVA 1, 2, and 3 have been associated with negative health outcomes as well, such as red meat (classified as NOVA 1) and processed meat (classified as NOVA 3, for example cured meat or ground beef), and high intakes of added sugars (classified as NOVA 2) [50]. One study from UK found the most nutritious and environmentally friendly foods to be distributed throughout all four NOVA groups [20]. Therefore, the usefulness of the NOVA classification and the term UPF is questionable, as it does not necessarily add something unique compared to looking at food based on nutrient profile, food category, or the health outcomes they are associated with [49, 51, 52]. The functionality of the NOVA system to inform dietary guidelines has also been questioned [43, 46], calling for a precise and rigorous standard classification system that differentiates within food groups and considers nutritional quality to limit interpretational uncertainty and risk of misclassification [53]. The inherent limitations of the NOVA classification system necessitate making additional classification criteria when applying the system to food databases. This makes room for subjective interpretations of the classification system between studies and studies have shown large variability in how people with education within nutrition classify foods according to the NOVA system [46]. This variability affects the results and limits direct comparisons between studies and is a limitation to all studies on UPF intake.

If participants did not know whether their food was homemade or industrially produced, the foods were classified according to the manufactured product. This may have led to an overestimation of the contribution from UPF to percentage of energy intake in the main analysis. In the sensitivity analysis, registrations which used generic codes for bread were randomly classified according to the share of bread in the store that were NOVA 3 or UPF. The results showed a significant decrease in the contribution of UPF to total energy intake compared to the main analysis (from 48 to 43%). This emphasizes the importance of having enough details about products consumed to classify foods correctly. It also shows that the chosen method for classification of food, where the NOVA classification system is lacking details, will affect the result.

While NOVA was initially designed to categorize foods based on their level of processing, some argue that its application may not provide additional insights when assessing environmental impact [22]. Although interventions aimed at replacing UPFs with NOVA 1 foods can often improve human health, they do not necessarily result in a reduced environmental impact [22]. Our sensitivity analysis illustrated that the association between UPF and environmental impact is vulnerable to the NOVA classification. This should be considered when interpreting the results and could be identified as a weakness in the approach to linking UPF with environmental impact. In the present study, we have only included data on climate impact. The climate data included in the compiled LCA database used in this study come with several well-known limitations: LCA data availability can vary, and there is limited data on the processing stages of the LCA.

The findings in the present study fit within a broader context of ongoing debates about the utility of food classification systems like NOVA in assessing environmental outcomes. They underscore the complexity of linking food processing levels directly with environmental sustainability metrics. Given these complexities, our study adds to the body of evidence suggesting that a more nuanced approach may be necessary. Since our study only included data on climate impact, further research is needed to provide a comprehensive understanding of how UPF affects environmental sustainability. Future studies should expand environmental metrics beyond climate impact to include, for example, water usage, land use, and biodiversity loss. Additionally, improving LCA databases to cover more detailed data on food processing can reduce the uncertainty of the results. Investigating policy interventions may also be crucial for designing strategies that promote both health and environmental sustainability.