Mental health among university students: The associations of effort-reward imbalance and overcommitment with psychological distress

Background: Mental health problems are highly prevalent among university students, but little is known about their underlying determinants. This study explores mental health among university students, the association between “ effort-reward imbalance ” (ERI), overcommitment and mental health, and to what extent ERI and overcommitment explain gender differences in mental health. Methods: Cross-sectional data were analyzed from 4760 Italian university students. The Kessler Psychological Distress Scale-10 was used to measure self-reported psychological distress, as an indicator of mental health, and the ERI – Student Questionnaire to measure effort, reward and overcommitment. The associations between ERI and overcommitment with psychological distress were estimated with multinomial logistic regression analyses. Results: 78.5% of the respondents experienced psychological distress, with 21.3%, 21.1%, and 36.1% reporting respectively mild, moderate and severe psychological distress. Female students were more likely to report moderate and severe psychological distress. ERI and overcommitment were strongly associated with severe psychological distress with ORs respectively up to 19.9 ( 95% CI: 12.2 – 32.5) and 22.2 ( 95% CI: 16.1 – 30.7). ERI and overcommitment explained part of the higher odds of severe psychological distress among female students comparing to males, attenuating the ORs from 2.3 (95% CI: 1.9 – 2.7) to 1.4 (95% CI: 1.2 – 1.7). Limitations: This cross-sectional study was performed on a large, but convenient sample. Discussion: More than one out of three students reported severe psychological distress. Decreasing ERI and overcommitment may be beneficial in the prevention of psychological distress among university students and may reduce gender differences in psychological distress. Longitudinal studies are needed to further investigate these associations.


Background
The prevalence of mental health problems among university students exponentially rose during the past decade (Benton et al., 2003;Duffy et al., 2019). Recent meta-analyses estimated a prevalence of 33.8% for anxiety and 27.2% for depressive symptoms among university students worldwide (Quek et al., 2019;Rotenstein et al., 2016). Psychological distress was identified as the most prevalent mental health problem for university students (Benton et al., 2003;Gibbons et al., 2019). The prevalence of mental health problems was found to be higher among female students than among the male ones (Benton et al., 2003;Duffy et al., 2019). Many studies estimated the prevalence of mental health problems among university students, but less attention was given to their underlying determinants. More knowledge is needed on the determinants of mental health problems among university students to develop effective interventions to promote mental health in the academic environment.
In working populations, the Effort-Reward Imbalance model has been successfully used to study the determinants of mental health problems (Hinsch et al., 2019;van Vegchel et al., 2005). There is empirical support that high effort in combination with low reward, so-called effort-reward imbalance (ERI), and overcommitment increase the risk of health problems such as depression (Hinsch et al., 2019;van Vegchel et al., 2005). Recently, the validity of the model was broadened to the student setting, where effort was defined as high study load, and reward as esteem, feeling respected in the academic environment and future work prospective (Portoghese et al., 2019;Wege et al., 2017). To date, only a few studies demonstrated that high effort, low reward, ERI and high overcommitment were associated with burnout, depressive and anxiety symptoms among university students (Hilger-Kolb et al., 2018;Hodge et al., 2019). Furthermore, no studies explored whether ERI and overcommitment could partly explain the higher prevalence of mental health problems among female compared to male university students.
The aims of this study are to 1) explore mental health among Italian university students; 2) the association of ERI and overcommitment with mental health; and 3) to what extent ERI and overcommitment explain gender differences in mental health.

Study design, study sample and recruitment
Cross-sectional data were collected among Italian university students (Ministero dell'Istruzione dell'Università e della Ricerca 2019). Data were collected with an online questionnaire implemented in Lime-Survey®, available from the 11th to the 23rd of December 2018. Respondents were recruited through web platforms commonly used by students for academic-related announcements, and through a public invitation on social media. Participants were informed about the study and provided digital informed consent. Given to the explorative aim of this observational study and the absence of follow up measurements, the questionnaire was anonymous and data were collected online in order to guarantee privacy to respondents. A total of 7773 individuals filled out the questionnaire. Individuals were included when (a) they gave information on mental health (n = 6150) and ERI and overcommitment (n = 4883); (b) were aged between 18 and 35 (n = 4852); (c) were enrolled in a bachelor's, master's or a combined degree (n = 4783) and (d) they did report to be male or female (n = 4760). Totally, 4760 (61.2%) individuals were included in the study.
No ethical approval is required in Italy for observational studies as they are not defined as medical/clinical research, referring to the Italian law 211/2003. This study complied with the Declaration of Helsinki and with the Italian privacy law.

Measures
Mental health. Psychological distress assessed with the Kessler Psychological Distress Scale (K10) was used as indicator of mental health . A wide range of mental disorders is typically characterized by a high level of psychological distress (Andrews and Slade, 2001). Measuring the level of generic psychological distress, compared with the use of a specific diagnosis, allows to investigate mental health with a broader approach and taking into account the severity of the condition (Kessler et al., 2010). Moreover, psychological distress is the most common mental health problem among university students (Benton et al., 2003;Gibbons et al., 2019). The K10 includes ten items (Cronbach's α = 0.91), investigating how often the person has been experiencing a specific feeling (such as tiredness, hopelessness, worthlessness), over the past 30 days, with answers ranging from (1)"none of the time" to (5)"all of the time". The scores ranges from 10 to 50, with higher scores indicating higher levels of psychological distress. Respondents were divided into four groups based on their sum score: individuals experiencing no psychological distress (10-19), and individuals experiencing mild (20-24), moderate (25-29) or severe (30-50) levels of psychological distress (Andrews and Slade, 2001). Respondents experiencing a low level of psychological distress were used as reference in the analysis.
ERI model. Effort, reward, and overcommitment were measured with the Effort-Reward Imbalance -Student Questionnaire (ERI-SQ), a version of the ERI questionnaire adapted to the student setting (Wege et al., 2017). The Italian version consists of 12 items and three subscales: effort (two items, Spearman's coefficient = 0.56) investigating the feelings toward the study workload (e.g. feeling being constantly under pressure due to the study workload), reward (five items, Cronbach's α = 0.68) investigating the perception of being treated fairly by peers and university staff, receiving proper credits, and job prospective, and overcommitment (five items, Cronbach's α = 0.77) investigating the constant thinking of academic duties and the ability to disconnect from studying (Portoghese et al., 2019). All items are scored on a 4-point Likert-scale ranging from 1 (strongly disagree) to 4 (strongly agree). A lower score is more favourable for effort and overcommitment, while a higher score is more favourable for reward. ERI was estimated using the effort-reward ratio calculated with the algorithm "(effort score)/[(reward score)*(correction factor)]", with a correction factor (0.4) accounting for the different number of items investigating effort and reward (Siegrist et al., 2004). A higher ERI indicated an imbalance with higher effort than reward. For each of the four considered dimensions, individuals were divided into three groups (low, middle, high) according to the 33rd and 67th percentiles. The groups with low effort, high reward, low ERI and low overcommitment were used as references.
Sociodemographics. Information concerning gender, age, work, living-studying location, type of degree and discipline were collected. With regards to gender, respondents could indicate whether they would define themselves as female, male, other, or prefer to not declare. Since only six respondents answered gender with 'other', only female and male individuals could be included in the study. Students with paid work were those who reported having paid employment next to study. The variable "living-studying location" classified students into three categories: (i) students who study in the city they lived in with their family before enrolling university ("studying in hometown"), (ii) students who travel on a daily basis from another town to reach the university ("commuting for studying"), and (iii) students who moved to a new city to study ("moved for studying"). Italian higher education includes three different types of degree: (i) Bachelor's degree of the duration of three years, (ii) Master's degree of the duration of 2 years, and (iii) "combined degree" for specific disciplines only and of the duration of 5 (e.g. law) or 6 (e.g. medicine, dentistry) years combining the bachelor's and master's degree.

Statistical analyses
Descriptive statistics stratified by gender were used to present the characteristics of the study population. The presence of gender differences was investigated using T-tests (psychological distress, effort, reward, ERI, overcommitment, age) and Chi-squared tests (paid work, living-studying location, type of degree).
Analysis of Variance (ANOVA) with Post Hoc test according to Tukey was used to analyze the differences in psychological distress across sociodemographic subgroups. A clustered boxplot was drawn to explore the distribution of the population in terms of psychological distress across the type of degree. Multiline charts showing means with one standard deviation were drawn to explore trends in effort, reward and overcommitment across age.
Spearman's coefficients were estimated as measures of the correlation between effort, reward, ERI, and overcommitment to verify multicollinearity between independent variables.
Adjusted odd ratios (ORs) with corresponding 95% confidence intervals (95% CI) were estimated to study the associations between effort, reward, ERI, overcommitment, and psychological distress. The multinomial logistic regression allowed comparisons of individuals with no psychological distress, used as a reference, with individuals experiencing mild, moderate, and severe psychological distress. The multinomial logistic regression was stratified by gender, and adjusted for sociodemographic factors., An interaction analysis for performed to investigate the presence of interaction between gender with effort, reward, ERI and overcommitment in their association with psychological distress.
A multistep multinomial logistic regression was used to investigate if and to what extent ERI and overcommitment explained the association between gender and psychological distress. In the first model, the unadjusted ORs of gender for psychological distress were estimated. The second model took into account sociodemographic factors, and the third model also included ERI and overcommitment.
All analyses were performed using IBM Statistics SPSS 25 (IBM, 2019).
In total, 78.5% of the respondents experienced psychological distress, of which 21.3% mild, 21.1% moderate, and 36.1% severe levels. Female students (39.1%) were more likely to report severe psychological distress compared to male students (26.8%). Psychological distress among female (M = 27.4, SD= 8.3) was significantly higher (F = 97.885, p = 0.001) than among male students (M = 24.6, SD= 8.1). Among female students, a statistically significant difference in psychological distress was found for the living-studying location (F = 5.318, p = 0.005). The post-hoc test indicated that commuters reported a significantly higher level of psychological distress (M = 28.0, SD= 8.3) compared to those studying in their hometown (p = 0.006) and those who moved for studying (p = 0.029). Among male students, no significant differences were found across living-studying location (F = 0.920, p = 0.399). Significant differences in psychological distress were found across type of degree among females (F = 6.811, p = 0.001) and males (F = 6.153, p = 0.002). Among female students, the post-hoc test showed that psychological distress was significantly higher among those enrolled in a bachelor's degree (M = 27.9, SD= 8.4) compared to those enrolled in a master's degree (p = 0.001) or in a combined degree (p = 0.001). Among male students, the post-hoc test indicated that those enrolled in a master's degree (M = 22.6, SD= 7.6) reported significantly lower level of psychological distress compared to those enrolled in a bachelor's degree (p = 0.018) or in a combined degree (p = 0.001) (Fig. 1). More information can be found in Table 2.
Concerning the ERI model factors, with older age, reward decreased while effort and overcommitment increased both among female and male students (Fig. 2). The changes of effort, reward and overcommitment with age were statistically significant both in men and women (Supplementary file, Table A).

The association of ERI and overcommitment with mental health
Spearman's coefficients (r s ) showed that overcommitment was moderately correlated with effort (r s = 0.56), ERI (r s = 0.50) (Supplementary file, Table B).
Among male and female students, effort, reward, ERI and overcommitment were statistically significantly associated with psychological distress ( Table 3). The strength of the associations increased with the severity of psychological distress with ORs up to 22.2 (95% CI= 16.1-30.7) for the highest level of psychological distress. The strength of the associations between the ERI model factors and psychological distress increased with a higher level of effort, ERI and overcommitment, and with a lower level of reward. The only exception to this trend concerned the association of effort and ERI with mild psychological distress among male students.

Gender difference in mental health: the role of effort-reward imbalance and overcommitment
No statistically significant interaction effect was found between gender with effort, reward, ERI, overcommitment and psychological distress.
The association between gender and psychological distress did not change significantly after adjustment for sociodemographic factors ( Table 4). The strength of the association between being a female student and psychological distress attenuated after additional adjustment for ERI and overcommitment from OR 2.3 (95% CI = 1.9-2.7) to OR 1.4 (95% CI = 1.2-1.7).

Discussion
In this study, over a third of the students experienced a severe level of psychological distress in the previous month. Female students were more likely to report psychological distress than male students. Students reporting higher effort, lower reward, higher effort-reward imbalance, and higher overcommitment were more likely to experience psychological distress, these associations were stronger with higher levels of psychological distress. The higher prevalence of psychological distress among female students could partly be explained by effort, reward and overcommitment.

Mental health among university students: psychological distress as indicator
Generic psychological distress was measured as an indicator of mental health for several reasons. Firstly, psychological distress is the most prevalent university students' mental health problem (Gibbons et al., 2019;Tsouros et al., 1998). Secondly, we aimed to investigate mental health status among university students with a more comprehensive approach taking into account the severity of the mental health problems. The impact of mental health problems varies across different conditions but also within the same condition depending on the severity. Measuring the level of psychological distress allowed us to take into account the severity of the mental health problem rather than the presence of a specific condition only (Salomon et al., 2015). Thirdly, we wanted to avoid focusing on specific disorders only because in the literature there is a lack concerning (i) mental health among Italian Table 1 Psychological distress, effort, reward, effort-reward imbalance (ERI), overcommitment and sociodemographic characteristics among 3586 female and 1174 male university students in Italy in 2018. university students, (ii) application of ERI model in the university setting and (iii) the use of the ERI model to explain gender difference. A general approach was preferable, leaving more specific investigation into specific conditions for future studies. Nevertheless, a K10 score of 20 or above discriminates well between individuals who meet the Composite International Diagnostic Interview criteria for anxiety and depression and those who do not (Andrews and Slade, 2001). K10 scores between 20 and 24, 25 and 29, and above 30 were associated with a likelihood of having a mild, a moderate and a severe mental disorder (Furukawa et al., 2003; Rural and Regional Health and Aged Care Services 2001; Stallman, 2010). In our population of students, the high prevalence of mild (21.3%), moderate (21.1%) and severe (36.1%) levels of psychological distress suggests a high prevalence of mental disorders (Andrews and Slade, 2001;Furukawa et al., 2003;Kessler et al., 2010Kessler et al., , 2003; Rural and Regional Health and Aged Care Services 2001). According to the Global Burden of Diseases 2017 (GBD2017), the prevalence of mental disorders among Italians was 17.5% in the group aged 20-24 years, and 17.1% among those aged 25-29 years (Institute for Health Metrics and Evaluation (IHME) 2017) being anxiety and depressive disorders the most common conditions. There are different possible explanations for the higher prevalence in our study than the GBD2017 (Institute for Health Metrics and Evaluation (IHME) 2017). Firstly, the GBD2017 took into account only some specific mental conditions, excluding others such as personality disorders which are highly prevalent (Lenzenweger et al., 2007) and characterized by psychological distress. Secondly, the level of psychological distress is higher among university students than among the general population in the same age groups (Stallman, 2010). Females and medical students are at higher risk for anxiety and depressive disorders (Kessler et al., 1994;Smith et al., 2018). In our sample, both represented a more significant share than they actually do at national level (Istituto Nazionale di Statistica (Istat) 2019). Nevertheless, our findings suggest that psychological distress is highly prevalent among Italian university students, and consequently, they might be at high risk for mental disorders (Andrews and Slade, 2001;Furukawa et al., 2003;Kessler et al., 2010; Rural and Regional Health and Aged Care Services 2001).

At university like at work: effort-reward imbalance and overcommitment matter
The ERI model successfully explains mental health problems among workers (Hinsch et al., 2019;van Vegchel et al., 2005), and this study shows that the ERI model might do the same also in the university setting. Effort, reward, and overcommitment are important modifiable factors and if longitudinal studies will confirm our findings, not only effort and reward separately but also their (im)balance needs to be taken into account by universities' staff to plan effective interventions to   promote mental health among students. ERI had a stronger association with severe levels of psychological distress compared with the effort and reward alone. The imbalance between effort and reward might be tackled by decreasing the effort or by increasing the reward. The ERI model takes into account the perceived effort, which depends on an objective component (e.g. study demands) and a subjective component (e.g. personal resources to accomplish the duties). Hence, universities can address planning and content of educational programmes in order to assure an appropriate study load as well as using teaching methods that enhance students' motivation (Karsenti and Thibert, 1994). Instead, the subjective component is related to the way students deal with the effort. Students reported their willingness to increase their knowledge concerning coping strategies and study-related stress management (Reeve et al., 2013). A few interventions at organizational level such as changes in the evaluation and the grading system (Kerdijk et al., 2013;Reed et al., 2011), and at individual level such as stress-management training, mindfulness and yoga (Akeman et al., 2019;Carpena et al., 2019;O'Driscoll et al., 2019;Regehr et al., 2012;Stillwell et al., 2017), were found to be effective to reduce psychological distress among students. Increasing resilience, defined as the ability to withstand and recover from mental hardship successfully (Herrman et al., 2011), might result in a decrease of the perceived effort. A study found that higher resilience was associated with a lower level of psychological distress among university students (Bacchi and Licinio, 2017).
Among the five items investigating reward, three referred to the feeling of "being fairly treated" at the university, by staff and peers. Consequently, our findings suggest that promoting good relationships with staff and peers might lead to an increase in perceived reward, which is associated with decreased psychological distress. A more supportive academic climate might be achieved with interventions at a group (e.g. mentor programs, extracurricular activities) (Hwang et al., 2017) and individual level (e.g. interpersonal psychotherapy) .

Table 4
Associations between gender and different levels of psychological distress after multistep adjustment for sociodemographics, and effort-reward imbalance (ERI) and overcommitment, among 4760 university students in Italy in 2018. and the ability of the student to disconnect from studying once the study time is finished. University students indicated study-life balance as a major stressor and declared to be interested in learning more about it, and in increasing their knowledge about school-related stress management and about coping strategies (Gibbons et al., 2019;Stallman and Hurst, 2016). More research is needed to understand better the overcommitment among students.

Gender differences but the real cause is often elsewhere
Women are more likely to suffer from anxiety and depressive disorders (Gibson et al., 2016; Institute for Health Metrics and Evaluation (IHME) 2017; Kessler et al., 1994;Smith et al., 2018), which are the mental disorders with the strongest association with psychological distress measured with the K10 (Andrews and Slade, 2001). Gender differences in mental health problems were also found among university students, with female students more likely to report a higher level of psychological distress (Auerbach et al., 2018;Backović et al., 2012;Beiter et al., 2015;Chow and Choi, 2019;Fond et al., 2018;Liu et al., 2019;Othieno et al., 2014;Stewart et al., 2019;Wu et al., 2016).
In our sample, females reported more psychological distress than males with increasing differences with higher severity of psychological distress. Our findings confirm that gender matters in mental health. However, gender differences are often caused by other factors rather than gender itself. Some authors suggested the gender difference in mental problems may be due to (i) a reduced likelihood of men's mental problems to be detected (Martin et al., 2013), and (ii) a lower likelihood of studies with no gender difference to be published (Franco et al., 2014). In our study, three main findings may contribute to the literature on this topic. Firstly, we did not find a significant interaction between gender with effort, reward, ERI and overcommitment in their associations with psychological distress, suggesting that effort, reward, ERI and overcommitment may have a similar effect on female and male individuals. Secondly, female students experience high effort, low reward, high ERI and high overcommitment more often and higher level than their male peers and this higher exposures could contributes to the gender difference in psychological distress. Thirdly, the gender difference in psychological distress decreased after adjustments for ERI and overcommitment. In order to truly explain gender differences in mental health problems, we may refer to other factors which might also explain within-gender differences (Pacheco et al., 2019), such as socioeconomic status and educational level, pattern of behaviors to socialize, attitude toward help-seeking and coping strategies (Bildt and Michélsen, 2002;Ennis et al., 2019;Miranda-Mendizabal et al., 2019;Smith et al., 2018;Vermeulen and Mustard, 2000). A meta-analysis on 308 educational programs worldwide reported that female students performed consistently better than male students (Voyer and Voyer, 2014). In Italy, female students graduate more often on time (57.9% vs 52.6%) and with higher grades (103.8/110 vs 102.0/110) then male students (Alma-Laurea, 2020). This could be somehow related with psychosocial aspects such as higher overcommitment, and results in a higher risk of psychological distress and mental health problems.
Future studies should better investigate the determinants and mechanisms behind the gender difference rather than merely testing their presence.

Strengths and limitations
The use of the validated and internationally broadly applied K10 to measure psychological distress due to its clinical relevance is a strength of our study. Contrarily, the use of the Italian version of the ERI-SQ may be a limitation due to its acceptable but limited psychometric properties (Portoghese et al., 2019). Despite it represents only a small share of the total university student population (total target population: 1.720.048, in 2018-2019), the large sample size is a strength of our study. However, our sample does not allow a stratified analysis across all study disciplines. In an additional analysis we noted that the reported associations between effort, reward, and overcommitment with psychological distress did not differ across the three largest studies, given some credence to the generalizability of the findings.
The data collection through a snowball technique came with strengths and weaknesses. The study population was a convenience sample with a possible selection bias which limits the generalizability of our results on the prevalence of mental health problems to the general population. Although the prevalence of mental health problems might be higher compared to the general population, we do not expect that the associations would differ from a broader population. The online snowball is convenient when aiming to reach hidden populations. University students experiencing mental disorders might be less likely to attend lectures and consequently, to fill out a paper-and-pencil questionnaire while being at the university. Moreover, due to the sensitivity of the topic, the online data collection provided more privacy and anonymity to respondents. Nevertheless, the use of an online data collection and online platforms might also have led to a common-method variance.
The cross-sectional nature of our study does not allow to identify a temporal association between effort, reward, ERI, overcommitment, and psychological distress. Further studies are needed in order to investigate this temporality as a mandatory requirement to suggest a causal relationship between effort, reward, ERI overcommitment, and psychological distress.

Conclusion
This study shows a high prevalence of psychological distress among Italian university students, in particular among female students. Our findings suggest that high effort, low reward, ERI and high overcommitment are associated with a higher level of psychological distress among university students. These factors also partly explain the higher prevalence of psychological distress among female students compared to male students. Concerning the high prevalence of psychological distress, there is a need for effective interventions to prevent mental health problems among students. The imbalance between effort and reward as well as overcommitment are potentially modifiable risk factors to target interventions. Longitudinal studies are needed to confirm that ERI and overcommitment are modifiable determinants of university students' mental health.

Availability of the data
Raw data pertaining to analyses performed in this study are available from the authors upon reasonable request. Proposals for research based on these data are welcome and can be sent to unicares@erasmusmc.nl.

Contributors
FP, SR, and AB designed the study. FP, SR, IP and AB contributed to the development of the questionnaire. FP collected and analyzed the data. FP wrote the manuscript. SR, UB, IP, MC and AB helped to draft the manuscript.

Declaration of Competing Interest
None.