Regular physical activity (PA) enhances motor skills and functional capabilities, significantly contributing to overall independence and physical well-being (1). Research indicates that incorporating daily PA and/or exercise into one's lifestyle can reduce the risk of chronic diseases and mortality, serving as a means of primary disease prevention (2). The World Health Organization (WHO) recommends that adults aged 18 to 64 engage in 150 to 300 min of moderate-intensity aerobic PA or, 75 to 150 min of vigorous-intensity PA each week, or an equivalent combination of both (3). Individuals who follow the WHO's PA recommendations have a significantly lower risk of all-cause mortality compared to those who do not (4). Implementing daily PA and exercise interventions can lead to an 80% reduction in the risk of cardiovascular disease (CVD) (5), a 90% reduction in the risk of type 2 diabetes (6), and a 33% reduction in the risk of cancer (7).

In addition to its well-documented benefits for physical health, regular PA is associated with enhanced cognitive functions, such as learning and memory retention, as well as improved mental health (8). Regarding the effect of PA on anxiety, systematic reviews and meta-analyses have shown that regular PA significantly reduces both the severity of anxiety across various populations and age groups. For instance, a meta-analysis by Wipfli et al. (9) found that exercise interventions led to a moderate reduction in anxiety symptoms, comparable to established psychotherapeutic treatments (9). More recently, a comprehensive umbrella review by Singh et al. (10) concluded that PA interventions effectively reduce symptoms of both depression and anxiety, primarily when conducted at moderate intensity (10). These findings provide strong empirical support for incorporating PA into preventive and therapeutic strategies for anxiety disorders.

While the positive effects of PA on mental health are well established, the dose-response relationship—indicating how changes in mental health outcomes are associated with the quantity of PA—has shown nonlinear associations and inconsistent findings. Studies conducted among Canadian adults and adolescents from Europe and North America have shown curvilinear relationships between PA and mental health (11, 12). For instance, Bernard et al. (11) found that moderate levels of objectively measured PA were associated with better self-reported mental health in Canadian adults, while higher levels did not provide additional benefits (11). Similarly, Khan et al. (12) concluded that for adolescents in Europe and North America, PA was positively associated with mental well-being only up to a certain threshold, beyond which the benefits plateaued (12).

Further research from the United States and China has revealed more pronounced nonlinear associations (13, 14). Chekroud et al. (13) identified an inverted U-shaped curve among U.S. adults, suggesting that mental health was optimized at around 45 min of PA, three to five times per week (13). Poorer outcomes were observed with insufficient or excessive levels of exercise, particularly for those engaged in high-intensity or high-frequency training (13). Likewise, Zhou et al. (14) reported a reverse J-shaped relationship in a large Chinese sample, indicating that both low and excessively high levels of PA were linked to worse mental health outcomes (14).

These findings imply that higher levels of PA may lead to either a plateau or a decline in mental health benefits. Both physiological factors, such as overtraining and elevated cortisol levels, as well as psychological factors like maladaptive coping and exercise dependence, have been proposed as explanations for this phenomenon (13). However, the point of diminishing returns regarding the protective effects of PA on anxiety varies across the studies mentioned. Possible explanations for these discrepancies include inconsistencies in measuring mental health outcomes and the neglect of other health behaviors that may also influence mental health. Relying solely on a single-item question about general mental health can introduce randomness in data collection and obscure the associations between PA and common mental health disorders (15). Additionally, various sociodemographic and lifestyle factors have been shown to influence the severity of symptoms of mental health disorders, making it essential to consider these factors when examining the relationship between different levels of PA and mental health (15).

Young adults have historically been at a higher risk for various mental health issues compared to other age groups (16). Students in particular represent a distinct group of young adults who encounter various stressors throughout their educational journeys. These stressors can include moving to a new city, achieving financial independence, forming romantic relationships, and assuming new responsibilities within their communities. All of this occurs while they strive to balance work and personal life within a demanding schedule (17). Adopting healthy lifestyle habits, such as engaging in regular PA, can significantly help these individuals manage stress during this critical stage of life (18). Moreover, young adults who adopt healthy behaviors are more likely to sustain these habits in the future (19).

Emerging evidence indicates that the connection between PA and mental health encompasses both physiological and psychological aspects (20, 21). Hamidi et al. (20) discovered that psychological well-being serves as a mediator between health-promoting behaviors and death anxiety among older adults who have experienced COVID-19 (20). Although this study did not focus specifically on PA or student populations, its proposed mechanism may provide insights into how psychological well-being could mediate the relationship between PA and anxiety among students. Additionally, Dehkordi and Chtourou (21) emphasized that structured PA can positively influence anxiety by improving emotional regulation, increasing self-efficacy, and enhancing coping strategies (21). Together, these findings underscore the importance of considering both physiological and psychological pathways when examining how PA and other health-promoting behaviors can serve as modifiable factors in the prevention and management of anxiety, especially in high-stress environments like universities.

The aim of this study was to investigate differences in mean scores on the Zung Anxiety Scale among groups of students with different levels of PA and to determine whether there is a dose-response relationship between PA and anxiety symptoms, taking into account various sociodemographic and lifestyle factors and study year. Our research aims to provide new insights into whether increasing levels of PA continue to offer protective effects against anxiety symptoms or if they reach a point at which these benefits diminish. Additionally, we seek to contribute to the literature by clarifying the point at which the positive effects of PA on anxiety plateau.

The mean and median Zung anxiety scale scores were the lowest in the group with an energy expenditure of 4,801–6,000 MET-minutes per week. There were no significant differences between the scores on the Zung anxiety scale between the group with the lowest energy expenditure and the groups with 6,001–7,200 MET-minutes/week and with >7,200 MET-minutes/week. The mean and median scores for each energy expenditure category are presented in Table 1.

After merging the categories that significantly differed from the category with the lowest energy expenditure, and then the categories with higher energy expenditure that showed no differences in score on zung anxiety scale compared to the lowest energy expenditure category, total of 136 students were in the category of 0–600 MET-minutes/week (10.7%), 1,038 were in the 601–6,000 MET-minutes per week (82.0%) and total of 92 students (7.3%) were in the category >6,000 MET-minutes/week.

The mean Zung anxiety scale score for the group with 601–6,000 MET-minutes/week was 34.65 ± 8.24, while the mean Zung anxiety scale score for the group >6,000 MET-minutes/week was 36.61 ± 9.59. The mean scores in the examined groups are presented in Figure 1.

The prevalence of energy expenditure of >6,000 MET-minutes/week was higher in 2023 (8.9% vs.5.1%), among students in Belgrade University (9.3%), among male participants (11.3% vs. 5.0%), among participants with poor self-rated financial status (17.5%), and poor self-rated health (16.2%). The prevalence of energy expenditure of >6,000 MET minutes/week was higher among participants who reported alcohol use (7.4% vs. 6.6%), binge drinking (11.3% vs. 5.3%), and cannabis use (13.3% vs. 6.2%). The participants across different PA categories exhibited significant differences in various socio-demographic, lifestyle, and mental health characteristics, including age, GPA, BMI, time spent playing video games, number of meals consumed, Social Support Scale scores, and Zung Anxiety Scale scores. The characteristics of participants in all three groups are presented in Table 2.

Multivariate logistic regression analysis with the energy expenditure of 601–1,200 MET-minutes/week compared to 0–600 MET-minutes/ week showed the association with studying in Novi Sad (OR: 0.32, 95% CI: 0.14–0.76), average (OR: 0.09, 95% CI: 0.01–0.76), or good self-rated financial status (OR: 0.10, 95% CI: 0.01–0.87), alcohol use (OR: 2.79, 95% CI: 1.38–5.65), and score on Zung anxiety scale (OR: 0.96, 95% CI: 0.93–0.99).

Multivariate logistic regression analysis with the energy expenditure of more than 6,000 MET-minutes/week compared to 0–600 MET-minutes/ week showed the association with 2023/2024 study year (OR: 1.69, 95% CI: 1.16–2.45), female sex (OR: 0.13, 95% CI: 0.02–0.72), and average self-rated financial status (OR: 0.04, 95% CI: 0.002–0.68).

Multivariate logistic regression analysis with the energy expenditure of more than 6,000 MET-minutes/week compared to with the energy expenditure of 601–1,200 MET-minutes/week showed the association with female sex (OR: 0.35, 95% CI: 0.14–0.89), and score on Zung anxiety scale (OR: 1.06, 95% CI: 1.01–1.11). The results of the multivariate analyses are presented in Table 3.

To our knowledge, this is the first study from Serbia to investigate differences in mean scores on the Zung Anxiety Scale among groups of students with different levels of PA and to determine whether there is a dose-response relationship between PA and anxiety symptoms, taking into account various sociodemographic and lifestyle factors and study year. Among the participants, 10.7% were in the category of 0 to 600 MET-minutes per week of PA, while 82.0% were in the range of 601 to 6,000 MET-minutes per week of PA, and 7.3% exceeded 6,000 MET-minutes per week of PA. Overall, 89.3% of participants met the WHO recommendations for PA. Our findings align with a systematic review that assessed the physical activity and fitness levels of university students, concluding that they generally exhibit satisfactory levels of PA and fitness (26).

The multivariate logistic regression analysis revealed that individuals engaging in 601 to 6,000 MET-minutes of physical activity (PA) per week were 4% less likely to experience anxiety symptoms compared to those in the 0 to 600 MET-minute group, and 6% less likely than those exceeding 6,000 MET-minutes per week. No significant difference in the likelihood of experiencing anxiety symptoms was observed between the group with 0 to 600 MET minutes and the group exceeding 6,000 MET-minutes per week. This pattern illustrates that while PA offers protective effects against anxiety symptoms, these effects tend to diminish after a certain volume of PA.

Regular PA is known to promote the release of neurotransmitters such as endorphins and dopamine, as well as hormones like serotonin, all of which can help reduce anxiety symptoms (27). Furthermore, various psychological mechanisms—such as increased self-efficacy, improved sleep quality, and a lower physiological stress response—may help explain how PA protects against anxiety (28). These mechanisms have been described in models like the stress-buffering model and self-determination theory, which could inform future interventions (29).

Our results indicated that the protective effects of PA against anxiety plateau after 6,000 MET-minutes per week. Several interpretations could explain this plateau in the relationship between PA and mental health. One possibility is that the mental health benefits of PA may reach a saturation point, suggesting that beyond a certain level of PA, additional exercise does not provide significant advantages. Another explanation could be reverse causality; for example, individuals experiencing higher anxiety may engage in increased PA as a coping mechanism, resulting in higher PA levels without a corresponding decrease in anxiety symptoms. To fully understand these phenomena, future longitudinal studies are needed.Our findings align with a large population-based study that used UK Biobank data, which also indicated a similar threshold effect in the relationship between PA and anxiety (30). Additionally, a nationally representative cross-sectional study demonstrated that while moderate levels of PA are beneficial for mental health, engaging in excessive PA does not provide additional benefits (31). Furthermore, another study involving adolescents from North America and Europe confirmed the non-linear relationship between PA and mental health (12).

This body of evidence emphasizes the importance of integrating PA as a modifiable factor in promoting mental health, especially in the post-pandemic era, where concerns about student mental health are increasing. However, it is essential not to recommend excessive PA as part of these mental health strategies.

Regarding study year and university, participants from 2023 study group were more likely to be in >6,000 MET-minutes per week of PA than 0–600 MET-minutes per week of PA and studying in Novi Sad was negatively associated with 601–6,000 MET-minutes per week of PA when compared to less than 600 MET-minutes per week of PA. To our knowledge, there is insufficient evidence to compare our findings on disparities in PA levels between different universities in Serbia. Conversely, the observation that participants from the 2023 group had a higher likelihood of engaging in over 6,000 MET-minutes per week of PA compared to those with 0 to 600 MET MET-minutes per week of PA contradicts evidence suggesting that, in recent decades, physical activity levels among the general population, including young adults, have steadily declined (32).

Female participants were more likely to report engaging in less than 600 MET minutes per week of PA compared to those in either the 601 to 6,000 MET-minutes per week of PA or the over 6,000 MET-minutes per week of PA group. These findings align with a recently published repeated cross-sectional study that analyzed data from the Eurobarometer (2005–2022) involving adults from 28 European countries and examined gender differences in meeting PA guidelines. The authors concluded that, overall, men tend to be more physically active than women (33).

The multivariate logistic regression analysis indicated that participants engaging in 601 to 6,000 MET-minutes per week of PA were more likely to report an average or good self-rated financial status compared to those who participated in fewer than 600 MET minutes per week of PA. Similarly, participants who exceeded 6,000 MET-minutes per week of PA were also more likely to report an average self-rated financial status than those in the group with less than 600 MET-minutes per week of PA. Overall, our findings suggest a negative association between PA and self-rated financial status. The relationship between financial status and PA has been extensively studied with mixed results. Some evidence suggests that individuals from lower-income families and those living in under-resourced communities tend to be less physically active (34). Conversely, other research shows that individuals with poorer self-reported financial statuses may engage in equal or even higher levels of PA compared to their wealthier peers, possibly due to the availability of cost-free activities such as unstructured play and outdoor recreation (35).

In comparison to the group with less than 600 MET-minutes per week of PA, those engaging in 601–6,000 MET-minutes per week of PA showed a positive association with alcohol use. Our findings regarding the relationship between alcohol consumption and PA are consistent with evidence indicating that regular PA often coincides with certain risk behaviors (36). One possible explanation for this trend is that students who are more physically active may form social connections and participate in events where alcohol is present, making them more prone to occasional alcohol consumption (36). Additionally, some individuals may engage in PA as a means of counterbalancing the health risks associated with unhealthy habits, influenced by compensatory beliefs and established patterns (36).

This study has several strengths worth noting. First, it uses a large, multi-center sample of medical students from five different universities, which enhances the regional relevance of the findings. Second, it utilizes valid and reliable research instruments for data collection.

Third, the study employs structured MET-based PA thresholds, similar to those used in other studies (30), allowing for a meaningful interpretation of the results. Finally, the analysis was adjusted for various socio-demographic and lifestyle factors to control for potential confounding variables. These features offer valuable insights for designing future interventions targeting young adults in high-stress academic environments. However, there are several limitations to consider. The cross-sectional design limits our ability to infer causation, and we cannot establish the directionality of the relationship between PA and anxiety. Both PA and anxiety symptoms were self-reported, which may introduce recall bias. Although we adjusted for several confounding factors, we could not include variables such as academic pressure, personality traits, or clinical mental health diagnoses, all of which may influence both PA and anxiety. Lastly, the overall response rates indicate the number of respondents compared to the total number of students enrolled in Social Medicine classes each school year. Students absent during the questionnaire distribution week were considered non-respondents, which may have led to an underestimation of the response rates in both study years.

To our knowledge, this is the first study from Serbia to investigate differences in mean scores on the Zung Anxiety Scale among groups of students with different levels of PA and to determine whether there is a dose-response relationship between PA and anxiety symptoms, taking into account various sociodemographic and lifestyle factors and study year. The most significant mental health benefits were observed at a weekly PA level of 601 to 6,000 MET-minutes. No additional mental health advantages were found for PA levels exceeding 6,000 MET-minutes, suggesting that a plateau effect may be present. Various sociodemographic and lifestyle characteristics were associated with different PA levels, including study year, university, sex, Self-rated financial status, and alcohol use. Targeted interventions that promote sustainable PA habits within university settings could be beneficial for enhancing student mental health. Future longitudinal and experimental studies are needed to establish causality and examine the underlying psychological mechanisms in the relationship between PA and mental health.