Ultra-processed food consumption among university students: application of front-of-package-based nutrient profiling scores and associations with physical activity and body composition

Abstract

Background

Recent modernization of food processes and Westernized lifestyles have raised concerns about the increased consumption of Ultra-Processed Foods (UPF). Evidence links UPF consumption to the prevalence of several chronic diseases. Higher UPF consumption correlates with higher fat mass (FM) and lower fat-free mass (FFM), alongside with higher levels of sedentary behavior and lower levels of physical activity (PA) in adults. However, the evidence in university students is still scarce. This study aimed to characterize UPF consumption in university students from the School of Medicine, University of Lisbon and test its association with body composition (BC) and PA. Additionally, these associations were also investigated in regard to UPF nutritional content.

Methods

This is a cross-sectional observational study with a sample of 163 Portuguese university students. Dietary intake was assessed through two non-consecutive 24-hour dietary recalls. UPF were classified using the NOVA system and front-of-package-based nutrient profiling systems (Nutri-Score and Traffic Light). BC was assessed through anthropometry and bioelectrical impedance, and PA and sedentary time (ST) through accelerometry. Independent-sample t-tests/Mann-Whitney tests and Chi-square tests assessed differences between groups. Associations were examined using Spearman’s correlations.

Results

UPF contributed to 24.8% of total energy intake (TEI). There were no associations between UPF and BC or PA. FSAm-NPS DI-NS and FSAm-NPS DI-TL were negatively associated with FFM (rho=-0.232, p = 0.003; rho=-0.237, p = 0.003; respectively) and positively associated with ST (rho = 0.238, p = 0.002; rho = 0.245, p = 0.002; respectively).

Conclusions

One-quarter of students’ TEI comes from UPF, aligning with national data. Our findings suggest that the consumption of UPF with worse nutrient profiles is related to lower FFM and higher ST. Enhancing nutrition literacy among university students may foster healthier food choices, potentially improving BC and ST.

Background

Recent modernization of food processes and Westernized lifestyles have raised concerns about the increased consumption of Ultra-Processed Foods (UPF) [1–5]. UPF are formulations of ingredients, mostly for exclusive industrial use, that result from a series of industrial processes involving physical, chemical, and biological modifications to enhance flavour, texture, and shelf life [4].

Cumulative evidence links UPF consumption to the prevalence of several chronic diseases [6, 7]. Evidence also suggests that UPF consumption is associated with an increased risk of developing metabolic syndrome (79%) and a concurrent increase in waist circumference (WC; 39%), indicating not only an increase in body fat but also a higher android-gynoid ratio [7, 8]. Additionally, some studies showed that UPF consumption is inversely associated with moderate-to-vigorous physical activity (MVPA) [9] and positively associated with sedentary time (ST) [10].

The NOVA system is one of the most widely used and recognized food classification systems [5], which categorizes foods into 4 groups based on their level of processing and the presence of specific ingredients (e.g., food additives) [1]. UPF fit into group 4, representing the category most distant from natural foods [1]. In most previous studies, UPF have been classified exclusively using this system, which does not account for the nutritional composition of foods [1]. Consequently, some of the negative health effects attributed to UPF consumption may be partially explained by their nutritional characteristics rather than processing alone [11, 12].

Although the degree of food processing is a rising concern among general consumers, there is still somewhat of a gap between the actual concept of UPF and individual’s perceptions of these products [13]. In this context, front-of-package nutrition labelling systems provide at-a-glance information on nutrient profile of foods and beverages, guiding consumers toward healthier food choices [14, 15]. Rather than classifying foods based on their processing level, these labelling systems provide nutritional information based on the food´s nutritional profile [14]. Different front-of-package labelling formats have been developed [16], being the Nutri-Score and the Traffic-Light systems the most recognized. The Nutri-Score employs a color-coded scale combined with letter grades (ranging from A to E) to classify the overall nutrient profile of food products. In contrast, the Traffic-Light system uses color indicators - red, yellow, and green - to reflect high, medium or low amounts of total fat, saturated fat, sugars and salt in the food products. These systems aim to simplify nutritional information, helping consumers - particularly those in Westernized, fast-paced food environments (e.g., college students) - make healthier choices more easily [13].

The academic transition period can lead to significant changes in dietary habits, with UPF consumption reaching up to 40% of the total energy intake (TEI) in university students [17]. Regarding the Portuguese population, UPF consumption seems to contribute for 24.0% of the TEI [18] in adults and 13,4% in young adults [19]. Additionally, university students seem not to meet physical activity (PA) recommendations for health [20], with only 36% of young Portuguese individuals aged 15–21 being physically active [21]. As the evidence on this matter is still scarce in this population, it is important to understand the reality regarding UPF consumption in these students and how UPF consumption is associated with their body composition (BC) and PA levels. Therefore, this study aimed to characterize UPF consumption among university students, by incorporating front-of-package-based scoring approaches that consider nutritional quality, and assess its association with PA and BC.

Methods

Study design, setting and participants

This cross-sectional observational study is based on self-reported and objectively collected data from one hundred sixty-three participants from a Portuguese University. Reporting followed the STROBE-nut checklist [22].

The sample was convenience-based. Potential participants (2374 students) were recruited via institutional emails, which included a digital brochure with information about the study and a link to an online form, between October 2023 and October 2024. Interested participants were invited to complete this eligibility form, providing details regarding their age, reproductive status (if pregnant or breastfeeding), impaired physical condition, intention regarding accelerometer use, availability for assessments, and a declaration of data consent. To be eligible, participants had to be enrolled in one of the two degree programs at the School of Medicine, University of Lisbon (Integrated Master’s in Medicine, a 6-year program, or Bachelor’s in Nutrition Sciences, a 4-year program), and be aged ≥ 18 years. Exclusion criteria included pregnant or breastfeeding women and individuals with any physical deficits that would interfere with engaging in activities of daily living. After completing the form, eligible individuals received the informed consent, along with a document with detailed information about the study via email, and were invited to visit the Advanced Nutrition Centre of the School of Medicine, University of Lisbon for assessments. Of those, 188 students responded to contact and booked the first assessment, of which 163 completed dietary recalls, BC and PA assessments. Figure 1 represents participants’ flowchart. This study was approved by the Ethics Committee of the Lisbon Academic Medical Centre of the School of Medicine, University of Lisbon (number 163/23) and all participants gave their written consent prior to participation.

Fig. 1.

Participants’ flowchart

Assessments

Assessments occurred at the Advanced Nutrition Centre of the School of Medicine, University of Lisbon at two-time points, spaced one week apart, following methodologies from a similar Portuguese study [18] and the National Food and Physical Activity Survey (IAN-AF) [21].

Sociodemographic information

In the first assessment moment, participants were asked to provide brief sociodemographic information, including gender, age, degree program, year, and residency status (resident vs. displaced).

Dietary intake

At both assessment moments, dietary intake was evaluated via two non-consecutive 24-hour dietary recalls - one corresponding to a typical weekday and another to a typical weekend day, to capture variability -, following the European Food Safety Authority’s guidelines [23]. The 24-hour dietary recalls were conducted by trained interviewers using the USDA’s multiple-pass method, which comprises five distinct steps designed to enhance accuracy and completeness of dietary data: (1) Quick List, in which participants provide an uninterrupted list of all foods and beverages consumed during the previous day (from waking up to bedtime); (2) Forgotten Foods, where interviewers probe for commonly omitted items such as snacks, beverages, or condiments; (3) Time and Occasion, which records the time and eating occasion for each item reported (e.g., breakfast, lunch, snack); (4) Detail Cycle, where detailed descriptions, portion sizes, preparation methods, and additions (e.g., oils, sauces) are collected; and (5) Final Probe, a final review to verify that no foods or drinks were missed [23]. A structured interview guide was used to standardize data collection, including standardized probing questions and portion size estimation aids (photographs) [24], in order to reduce prompting bias and minimize potential misreporting. TEI was estimated using the nutritional information from the Portuguese Food Composition Table [25].

Foods were classified using the NOVA system [1]. According to the NOVA classification system, there are four groups of foods based on their level of industrial processing: unprocessed or minimally processed (group 1), processed culinary ingredients (group 2), processed foods (group 3), and UPF (group 4). Foods in group 4 may include sugars such as fructose, inverted sugar, dextrose, or lactose; fats modified by hydrogenation or esterification; protein sources such as hydrolysed or isolated protein, gluten, casein, whey protein, and “mechanically separated meat”. They may also incorporate food additives, including flavours, emulsifiers, sweeteners, thickeners, gelling agents, among others [1].

The percentage of TEI from UPF was calculated [17]. Supplements such as protein powder, carbohydrate gels and others were considered in this calculation. UPF nutrient profile was accessed using the Nutrient Profiling System of the British Food Standards Agency for Dietary Index, linked to Nutri-Score (FSAm-NPS DI-NS) [26]. According to the Nutri-Score algorithm, this index ranges from − 15 to 40 points, with lower values indicating a more favourable nutrient profile [26].

Nutri-Score and the Traffic Light system are front-of-package labelling schemes based on distinct conceptual frameworks. Nutri-Score relies on a nutrient-profiling algorithm that integrates both unfavourable (e.g., energy, sugars, saturated fat, sodium) and favourable components (e.g., fibre, protein), whereas the Traffic Light system provides a categorical evaluation of selected risk nutrients (total fat, saturated fat, sugars, and salt) using colour codes (green, yellow, and red) [27]. Despite these conceptual differences, both systems aim to summarize nutritional quality and guide healthier choices. Because the Traffic Light system is the most widely used front-of-package labelling scheme in Portugal, we developed an exploratory adaptation of the FSAm-NPS DI for the Traffic Light system (FSAm-NPS DI-TL). This adaptation seeks to enable a standardized, quantitative comparison of UPF nutrient profiles derived from Traffic Light information and to examine how results may differ depending on the front-of-package framework applied. This adaptation was developed as followed:

1. First, since Nutri-Score assigns higher scores to foods with worse nutrient profiles [26], we attributed progressively higher point values to the Traffic Light colour categories: green = 0 points, yellow = 1 point, and red = 2 points for each nutrient component [27];

2. For each UPF, points were assigned for total fat, saturated fat, sugars and salt according to their Traffic Light classification [27]. A total score was calculated by summing these values, yielding a minimum score of 0 (all four nutrients classified as green in the Traffic Light system) and a maximum score of 8 (all four nutrients classified as red in the Traffic Light system) [27];

3. To facilitate comparability with the Nutri-Score scale, this 8-point scale was linearly transformed to the 55-point Nutri-Score range (-15 to 40) [26], applying the rule of 3.

For example, a UPF containing 2,5 g/100 g of total fat, 2 g/100 g of saturated fat, 25 g /100 g of sugars, and 1 g /100 g of salt would be classified as green for total fat, yellow for saturated fat, red for sugars, and yellow for salt according to the Traffic Light system [27]. This corresponds to a total of 4 points (0 for total fat, 1 for saturated fat, 2 for sugars, and 1 for salt). Applying the rule of 3:

To align with the Nutri-Score scale, which begins at -15 points, 15 points were subtracted, resulting in a final FSAm-NPS DI-TL score of 13 points. This transformation allows direct numerical comparison between the Traffic Light-derived index and the Nutri-Score index while maintaining transparency regarding the methodological assumptions. As such, besides facilitating a more nuanced understanding of how different front-of-package systems may influence the perceived nutritional quality of UPF, this approach allows exploration of how variations in UPF nutrient profiles, as captured by each system, are associated with BC and PA parameters.

Physical activity

In the first assessment moment, participants were provided with a wGT3X-BT accelerometer to assess their PA for a period of five complete and consecutive days [28], which was returned in the second assessment moment. Participants wore an accelerometer on their waist except during water activities and sleep and were instructed to continue their usual routines to minimize behaviour change. The device recorded data at 100 Hz in 10-second epochs. To be included in the analysis, participants needed at least three valid days of wear (one weekend day included) with a minimum of 600 min per day. Data processing was done using ActiLife Software (version 6.13.4). The PA intensity was categorized using Troiano et al. (2008) cut-off points [27]: sedentary (0–99 counts/min), light (100–2019 counts/min), moderate (2020–5998 counts/min), and vigorous (≥ 5999 counts/min). PA data were converted into minutes per valid day and ST to hours per day. The recommendations of PA of the World Health Organization (WHO) were used to classify whether participants met the guidelines [20].

Body composition

In the second assessment moment, anthropometric data (weight, height, WC, and hip circumference) were collected using a digital scale, stadiometer, and anthropometric tape (SECA) by trained researchers, according to standard procedures [29]. Additionally, participant’s BC, including fat-mass (FM) and fat-free mass (FFM), was assessed using a bioelectrical impedance analyser (Bodystat Quadscan 4000). Body mass index (BMI) was calculated from weight and height (kg/m2) and interpreted according to the WHO guidelines [30]. WC and waist-to-hip (WHR) ratio were also interpreted according to the WHO guidelines [31] to assess metabolic risk. FM and FFM percentages were interpreted based on NHANES III cutoffs [32].

Statistical analysis

Statistical analyses were conducted with IBM SPSS version 28.0, setting significance at p < 0.05. Normality was assessed using the Shapiro-Wilk test. Descriptive statistics were used to characterize the sample. For qualitative variables, simple and relative frequencies were calculated, while for quantitative variables, means, standard deviations, minimums, and maximums were computed. Independent-sample t-tests or Mann-Whitney tests for continuous variables and Chi-square or Fisher exact tests for categorical variables were used to compare differences between groups (i.e., sex, degree program, year os studies, and residency status). Spearman correlations were performed to examine associations between UPF consumption/nutrient profile score and BC and PA. Quintiles of both FSAm-NPS dietary indexes were created based on the sample distribution. Values were ranked in ascending order and divided into five groups of equal size (each representing 20% of the sample) using the Rank Cases function. Participants were therefore classified into quintiles ranging from Q1 (lowest dietary index scores, indicating better nutritional quality) to Q5 (highest scores, indicating poorer nutritional quality). Quintile-based comparisons were conducted using non-parametric tests due to non-normality of the variables. Differences between quintiles were assessed using the Kruskal-Wallis test for continuous variables and the Chi-square test for categorical variables. Post-hoc pairwise comparisons were conducted with the Mann-Whitney U test, with Bonferroni correction applied where appropriate.

Results

Sample characteristics and descriptive results

Participant characteristics are described in Table 1.

Table 1.

Participants characteristics

Variable	Mean (n = 163)	SD	Min	Max
Sex (%)
Women	74.8
Men	25.2
Demographic
Age (years)	21.30	2.65	18	33
Degree program (%)
Bachelor’s in Nutrition Sciences	25.8
Integrated Master’s in Medicine	74.2
Year of studies	3.36	1.88	1	6
Academic transition (%)
Freshmen	24.5
Non-freshmen	75.5
Residence (%)
Resident	62.0
Displaced	38.0
Dietary intake
Total energy intake (kcal [kJ]/day)	1901 [7954]	475 [2000]	348 [3131]	4399 [18405]
% of total energy from UPF	24.82	15.72	0	62.89
FSAm-NPS DI (linked to Nutri-score)	11.78	6.16	-1.22	32.00
FSAm-NPS DI (linked to Traffic Light)	9.83	8.12	10.56	29.50
Body composition
Body mass index (kg/m2)	22.82	3.10	17.10	39.80
Body mass index class (%)
Underweight	3.7
Normal weight	77.3
Pre-obese	17.2
Obese	1.8
Waist circumference (cm)	71.12	8.13	56.00	95.10
Hip circumference (cm)	97.77	7.12	83.00	127.00
Waist-to-hip ratio	0.73	0.06	0.60	0.90
Fat mass (%)	21.83	6.62	6.60	46.70
Fat-free mass (%)	69.20	23.08	6.20	93.40
Physical activity
Sedentary time (hours/day)	11.30	1.79	2.42	14.93
Sedentary breaks (times/day)	160	36	93	322
Moderate-to-vigorous physical activity (minutes/day)	54.23	21.78	4.78	134.61
Physical activity compliance (%)
Fulfills WHO additional recommendations	66.9
Fulfills WHO recommendations	30.1
Does not fulfill WHO recommendations	3.1

Differences between groups were accessed by non-parametric statistical tests

UPF Ultra-processed food, FSAm-NPS DI Nutrient Profiling System of the British Food Standards Agency for Dietary Index, WHO World Health Organization

The majority of participants were female (74.8%) and 74.2% were enrolled in Medicine. Age ranged from 18 to 33 years, with a mean age of 21 years. Only 24.5% were freshmen, and most lived with their families (62.0%). Students consumed on average 1901 kcal (7953.784 kJ) daily, with 24.8% (475 kcal; 1999.920 kJ) coming from UPF, and only 3.1% reported not consuming UPF. FSAm-NPS DI-NS varied between − 1.22 and 32, with a mean of 12 points and FSAm-NPS DI-TL varied between − 10.56 and 29.50, with a mean of 10.

Figure 2 shows the main UPF consumed, according to the Nutri-Score classification [26]. Classification D includes some of the food groups listed in both C and E, due to differences in their composition, hence classification, between brands.

Fig. 2.

Main UPF consumed, according to the Nutri-Score classification [26]

Figure 3 shows the main UPF consumed, according to the Traffic Light classification [27], with foods in green (all four components within the green interval), yellow (no components within the red interval), and red (at least one component within the red interval). Major differences between the two classification systems were identified for hummus, fast food restaurant meals (burger, pizza, kebab, wrap), low-fat sauces, chocolate powder, poultry meat, chips, vegan burgers and breaded hake and chicken; all of these foods belong to NOVA group 4 as a result of the presence of additives and/or industrial processing methods. All of these were classified as red in the Traffic Light - meaning that these products were high in at least one of the four nutrients - and varied between A and D in the Nutri-Score classification. These differences are reflected in the FSAm-NPS DI scores and can be explained by a key distinction between the two systems: Nutri-Score values favourable components such as protein, fibre and proportion of fruits and vegetables, whereas the Traffic Light system evaluates only selected risk nutrients. Consequently, an UPF high in total fat, saturated fat, sugar or salt receives a higher (less favourable) FSAm-NPS DI-TL score even if it contains substantial amounts of protein, fibre and/or fruits/vegetables, as exemplified by the products described above [26, 27]. Nevertheless, in general, these front-of-package labelling systems were positively and strongly statistically correlated with each other (rho = 0.864, p < 0.01).

Fig. 3.

Main UPF consumed, according to the Traffic Light classification [27]

The mean BMI was 22.8 kg/m², which resonates with 77.3% of the students being in the normal weight range, and the mean HRW was 0.73 (Table 1). The mean FM was 21.8% (24.5% for women – within the normal range [32], and 13.9% for men − 1.7% points lower than the average range [32]). Mean FFM was 69.2% (67.6% for women, 73.3% for men). Participants spent an average of 54 min/day in MVPA, with 95% meeting the WHO recommendations [20]. Nevertheless, they spent an average of 11 h/day in ST, with a mean of 160 daily breaks.

As an exploratory analysis, we examined UPF consumption and nutrient profile score, BC and PA by sex, degree program, academic transition, and residency status. Medical students had higher FSAm-NPS DI-NS and FSAm-NPS DI-TL than Nutrition students (12.42 ± 6.34 vs. 10.01 ± 5.29, p = 0.033; 10.72 ± 8.01 vs. 7.33 ± 7.99, p = 0.020; respectively), as well as higher WC (72.0 cm ± 8.4 vs. 68.6 cm ± 6.8, p = 0.016) and WHR (0.73 ± 0.06 vs. 0.71 ± 0.05, p = 0.040). No differences were found between men and women, freshmen and non-freshmen students and resident and displaced students (p > 0.05).

Associations between UPF consumption/nutrient profile score and body composition and physical activity

Correlations of UPF consumption and nutrient profile score with BC and PA are depicted in Table 2.

Table 2.

Correlations of UPF consumption and quality with body composition and physical activity

	Age (years)	Degree program (year)	TEI (kJ/day)	BMI (kg/m2)	WC (cm)	HC (cm)	WHR	FM (%)	FFM (%)	SB (times/day)	ST (hours/day)	MVPA (min)
UPF Variables
UPF (%)	0,007	-0,033	0,029*	-0,046	-0,068	-0,045	-0,048	0,094	-0,054	0,001	0,069	-0,017
FSAm-NPS DI (NS)	-0,038	-0,038	0,021	-0,015	-0,035	-0,021	0,013	0,131	-0,232**	-0,030	0,238**	-0,001
FSAm-NPS DI (TL)	-0,090	-0,106	-0,021	-0,077	-0,046	-0,044	-0,002	0,113	-0,237**	-0,020	0,245**	-0,040

TEITotal Energy Intake (kJ/day), BMIBody mass index (kg/m²), WCWaist circumference (cm), HCHip circumference (cm), WHR Waist-to-hip ratio, FMFat mass (percentage from total body weight), FFMFat-free mass (percentage from total body weight), SBSedentary breaks (times per day), STSedentary time (hours per day), MVPAModerate-to-vigorous physical activity (minutes per week), UPFUltra-processed food (percentage from total energy intake), FSAm-NPS DI (NS)Nutrient Profiling System of the British Food Standards Agency for Dietary Index (linked to Nutri-Score), FSAm-NPS DI (TL)Nutrient Profiling System of the British Food Standards Agency for Dietary Index (linked to Traffic Light)

*p value < 0,05; **p value < 0,01; bold correlations are for significant Spearman’s correlation coefficients

Both FSAm-NPS DI-NS and FSAm-NPS DI-TL were negatively associated with FFM (rho=-0,232, p = 0.003; rho=-0,237, p = 0.003; respectively) and positively associated with total ST (rho = 0,238, p = 0.002; rho = 0,245, p = 0.002; respectively), even when ajusting for TEI. In fact, the fifth quintile of FSAm-NPS DI-NS had significantly less FFM than the first (p = 0.007), second (p = 0,0.004) and third (p = 0.039) quintiles (see Fig. 4).

Fig. 4.

Fat-free mass’ descriptives for each quintile of FSAm-NPS DI (NS)

Similarly, there was a significant difference between the fifth quintile of FSAm-NPS DI-TL and the first (p = 0.038), second (p = 0.001) and third (p = 0.018) quintiles in what concerns ST, with the fifth quintile showing higher time spent in sedentary behaviour (see Fig. 5). No other significant correlations were found.

Fig. 5.

Sedentary time’s descriptives for each quintile of FSAm-NPS DI (TL)

Discussion

This study sought to characterize the consumption of UPF among university students and explore its association with BC and PA. Results showed that UPF accounted for 24.8% of TEI, averaging 475 kcal/day (1999.952 kJ/day). There were no associations between UPF consumption and BC or PA parameters. Nonetheless, the consumption of UPF with a worse nutrient profile score (considering both the Nutri Score and Traffic Light Dietary Indexes) was associated with lower FFM and higher levels of ST. These findings extend the ones from previous studies focusing on the consumption of UPF and reporting positive associations with FM and sedentary behavior [10] and negative associations with FFM [33] and PA [9] in adults, pointing out to the role of UPF nutrient profile in these associations.

Previous studies have reported that UPF consumption among young adults varies widely, typically ranging from 40% to 60% [34, 35]. In fact, a study with a Mediterranean (Greek) sample found a high UPF consumption among university students (around 40%) [17]. In contrast, our study found a lower UPF consumption, although consistent with data from the Portuguese general population [18]. This lower consumption may partially explain the lack of associations between UPF consumption and other parameters, compared to findings from other studies [9, 33, 36]. These disparities may be attributed to cultural and nutritional literacy differences, both of which can influence dietary choices [37, 38]. They may also reflect methodological aspects, including the relative homogeneity of our sample - composed largely of health-conscious students-, which likely reduced the variability required to detect further associations.

Despite the low UPF consumption in our sample, the nutrient profile score of the UPF consumed was poorer (11.78 points) when compared to the NutriNet-Santé Cohort, which reported an average FSAm-NPS DI-NS of 6.59 points [39]. Additionally, we computed an FSAm-NPS DI for the Traffic Light system [26, 27]. To the best of our knowledge, this has not yet been explored in the literature, but it provides a potential method to mirror the two front-of-package labelling systems. In our study, the overall nutrient profile score of UPF was similar between the two systems and they were strongly correlated. Still, differences were observed in the classification of certain food groups, likely due to the inclusion of protein and fibre in the Nutri-Score algorithm [26]. Hummus was the most notable example: as a legume- and seeds-based product rich in fibre and protein, it is more favourably rated by Nutri-Score [26] than by the Traffic Light, which focuses on its fat and salt content [27].

Interestingly, Medical students consumed UPF with worse nutrient profile scores compared to Nutrition students. Differences in educational curricula may potentially contribute to this disparity, as they can influence awareness regarding the importance of a healthy and balanced diet, with evidence suggesting that higher nutrition literacy significantly predicts diet quality among undergraduate college students [38]. In fact, in this School, students enrolled in the Integrated Master’s in Medicine, are typically exposed to nutrition only through optional courses, and only a small proportion have access to these courses, resulting in limited or sometimes non-existent nutrition literacy within their academic training.

Our sample exhibited favourable BC parameters [30–32], with a mean BMI of 22.8 kg/m², WHR of 0.73, and mean FM of 21.8%. This overall positive profile, coupled with lower UPF consumption, may help explain the lack of associations between UPF intake and BMI or BC. Previous research has generally reported positive associations between UPF consumption and obesity-related metrics, including BMI (as a marker of obesity and excess weight), WC (linked to metabolic complications) [7] and body fat [7, 10]. Noteworthy, we observed a negative association between UPF nutrient profile score and FFM and a distinct pattern among Medical students, who exhibited worse BC metrics – specifically, higher WC and WHR - compared to Nutrition students. Again, this may be partially explained by differences in nutrition literacy: at this institution, Nutrition is not mandatory for Medical students and is only available as optional curricular units with limited enrolment and superficial coverage. Even among Medical students with an interest in nutrition, this likely results in lower nutrition literacy when compared to Nutrition students. These differences in BC, alongside potentially lower nutrition literacy, have been linked to lifestyle patterns detrimental to health [1, 6, 7]. Although we didn’t find significant differences between freshmen and non-freshmen students, previous studies have highlighted the academic transition period as a critical phase associated with dietary changes, with these changes being often attributed to reduced support and diminished interest in meal preparation and planning [17]. It’s important to note that not all UPF are nutritionally equivalent. While many are high in sugars and fats [40–43], some can make positive contributions to the diet and, consequently, to BC.

Regarding PA, the average students engaged in a relatively high level of MVPA – roughly 54 min/day-, with 95% meeting the WHO recommendations [20]. However, their average ST was 11 h per day, exceeding the 24-Hour Movement guidelines by three hours, which aligns with recent studies on this matter suggesting no more than 8 to 9 h of ST per day [44, 45]. A 2019 systematic review on dose-response associations between accelerometer-measured PA and sedentary behavior found that 10 or more hours of ST per day increases the mortality risk by 48% [46]. This indicates that despite meeting PA recommendations, these students may face higher mortality risk due to prolonged ST. Notably, sedentary behaviour and PA represent independent health risk factors, particularly for cardiovascular health, meaning that the benefits of PA do not entirely offset the negative effects of excessive sitting [47]. To our knowledge, there are no established guidelines specifying the minimum number of sedentary breaks needed to promote health, as current evidence seems to be insufficient to determine the optimal frequency and/or duration, as highlighted by the WHO guidelines on PA and sedentary behaviour [20]. Nevertheless, considering the total ST, increasing the number of sedentary breaks to more than 160 times/day could be beneficial to students’ health [46–48]. Universities and health practitioners could therefore play an important role in encouraging students to adopt strategies such as active breaks during study time, standing desks, or integrating light-intensity activities throughout the day, as complementary interventions to structured PA.

No associations were found between overall UPF consumption and PA, although UPF nutrient profile score was positively associated with ST. Other studies have reported lower PA among individuals consuming more UPF [10, 39, 49], as well as higher ST among those with higher BMI [50]. Evidence on the relationship between UPF quality, FFM, and ST remains limited, but existing findings suggest that higher UPF consumption and poorer nutritional quality - particularly when combined with low PA and, consequently, greater ST - are linked to less favourable anthropometric outcomes in young adults (e.g., increased FM) [51]. This pattern reflects common behaviors in young adults and college students, who often consume UPF with poorer nutrient profiles while engaging in sedentary activities such as watching TV, gaming, or studying - behaviors that hinder FFM development while reinforcing inadequate dietary habits and FM increase [51].

Important limitations of this study includes its cross-sectional design, which precludes us from inferring causality, although it allows for the identification of associations that may serve as a foundation for further investigations; the nature of the sample - a convenience sample composed by students willing to participate, perhaps showing a selection bias towards more motivated individuals, potentially over-representing health-conscious students and making the sample potentially less representative of this particular population; and social desirability as dietary intake assessed through 24-hour dietary recalls relies on self-reported dietary data and may have influenced participants’ responses towards healthier food choices to gain validation from the interviewer, which might underestimate UPF intake and distort associations. Additionally, women tend to underreport more than men with recall-based methods, likely due to stronger beliefs in healthy eating, and our sample is mostly composed by women [52]. Associations should also be interpreted with caution, as the study’s relatively small sample, limited to a single faculty, may have reduced the power to detect small but potentially meaningful effect sizes. Another potential limitation relates to the classification of UPF. The NOVA classification system does not always provide clear criteria for assigning certain foods to group 4, allowing room for subjective interpretation by researchers and introducing variability in UPF consumption data [53]. To minimize this variability, the classification of foods with ambiguous categorization was discussed among the researchers to reach a consensus. Moreover, since the consumption of UPF was quantified based on its contribution to TEI, certain types of UPF - such as low-calorie, zero-calorie, or diet products - may have been underestimated or inaccurately quantified, as they contribute minimally to TEI despite being highly processed. On the other hand, UPF quality was classified through two different metrics: FSAm-NPS DI-NS [26] and FSAm-NPS-DI-TL. The last one constitutes an innovative, exploratory approach developed by the researchers to turn UPF quality through TL measures analytically comparable to NS, a metric that still requires formal validation.

While the study provides valuable insights into UPF consumption, BC and PA among young adults, the extent to which these findings can be extrapolated to other populations requires careful consideration. The sample consisted primarily of university students, who may differ from the broader young adult population in lifestyle, socioeconomic status, and health behaviors. Similarly, the study was conducted within a single geographic region and academic setting, which may limit applicability to populations in other cultural or environmental contexts. Nonetheless, the observed associations between UPF quality, sedentary behavior, and anthropometric outcomes align with previous research, suggesting that the underlying mechanisms may be relevant beyond this sample. Future research involving more diverse and multi-center populations would help to further establish the external validity of these findings.

The present study used objective measures to assess PA and ST, and BC. Additionally, all methods applied were based on validated protocols from related studies, with interviews and assessments conducted according to Standard Operating Procedures outlined by the main investigators, ensuring consistency and data quality across the study. Nevertheless, we acknowledge that accelerometers underestimate some activity types (e.g., swimming) and that reactivity (i.e., participants may increase their activity while wearing the accelerometer) cannot be entirely excluded.

Conclusions

This study found that around one-quarter of the TEI among university students derived from UPF, aligning with patterns observed in the Portuguese general population. Although no associations were observed between overall UPF consumption and BC or PA parameters, students who consumed lower nutritional-quality UPF presented lower FFM and higher ST. Despite the observational nature of the findings, they contribute to the existing literature and may inform public health strategies targeting university students, helping further reduce UPF consumption and promote healthier food choices, potentially leading to improvements in BC and overall health behaviours.

Abbreviations

BC

Body composition

FFM

Fat-free mass

FM

Fat Mass

MVPA

Moderate-to-vigorous physical activity

FSAm-NPS DI-NS

Nutrient profiling system of the British food standards agency for dietary index linked to nutri-score

FSAm-NPS DI-TL

Nutrient profiling system of the British food standards agency for dietary index linked to traffic light

PA

Physical activity

ST

Sedentary time

TEI

Total energy intake

UPF

Ultra-processed foods

WHO

World health organization

WC

Waist circumference

WHR

Waist-to-hip ratio

Authors’ contributions

IS, JM and JC contributed to study design and ethics. JM, JC contributed to data collection. IS, JM, JC and RP analysed and interpreted data. IS, JM, JC, RP and MP prepared and reviewed manuscript for submission. IS finalized the manuscript for submission. All authors read and approved the final manuscript.

Funding

This research received no external funding. JM, JC and IS received funding from the School of Medicine, University of Lisbon, through a research support program (GAPIC) - grant number 20240020.

Data availability

The dataset used and/or analysed during the current study is available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This study was conducted according to the guidelines laid down in the Declaration of Helsinki and all procedures involving research study participants were approved by the Ethics Committee of the Lisbon Academic Medical Centre of the School of Medicine, University of Lisbon (163/23). Written informed consent was obtained from all subjects.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.