This study analysed depressive symptoms in student samples from three universities in three European countries and their association with perceived burdens related to being a student. We found a relatively high level of depressive symptoms as compared to a representative sample of the German population, where the 85th
percentile of the M-BDI for 20 to 30 year olds is 35 for the total sample, with a five point difference between genders [19
]. Similarly, previous reports have shown high levels of depressive symptoms among college or university students [1
]. Our bivariate finding of higher depression scores among students in Bulgaria and Poland as compared to those from Germany is consistent with two previous studies that reported substantially higher prevalences of depressive symptoms among student samples in Eastern Europe than in Western Europe [27
]. However, the difference in M-BDI scores between Poland and Germany disappeared after adjusting for perceived stress, which was particularly high in Poland.
All four subscales of perceived burden were associated with depressive symptoms and, apart from burdens related to course work and exams (subscale "Course work"), the associations persisted after adjusting for perceived stress. Since perceived stress explained the association between the "Course work" burden subscale and depressive symptoms, it seems that stress fully mediated the effects of these burdens. On the contrary, perceived stress only partially acted as a mediator for the "Isolation" and "Relationships" subscales, and did not act as a mediator at all for the "Future" subscale (compare partial eta-square values from models 1 and 2 in Table ). These findings partly support the distinction between total and mediated effects specified in our aims. While perceived burdens have some conceptual similarity with perceived stress, they have also independent association with depressive symptoms. This may be a result of the perceived stress scale measuring only selected aspects of stress or may be a result of a true conceptual distinction between perceived burdens and stress. Our findings on the extent to which the subscales' associations were mediated by perceived stress ("Course work" the most and "Future" the least) support the suggestion of a true conceptual distinction. Course work is most directly a stressor in students' lives, and the concept of perceived stress is mostly related to the present time, not to the future outlook. This supports the importance of stress management interventions for student health, but also indicates that interventions addressing perceived burdens would be useful.
Only few studies have examined perceived burdens related to being a student, such as grades and competition, career and future success, too many demands and deadlines, financial and health-related burdens [28
], and their impact on mental health [29
]. Our findings suggest that more students felt burdened by their course work and exams and an uncertain outlook for the future than by problems with relationships or being isolated, with little differences between countries. When present, however, the feeling of being isolated, had the strongest impact on depressive symptoms; a fact which highlights the importance of social support and the sense of belonging to a group while being a student.
Similarly, economic situation (measured by sufficiency of income) mainly had an indirect effect on depressive symptoms (mediated over perceived stress). After controlling for perceived stress, sufficiency of income was no longer significantly associated with depressive symptoms. These findings underline findings from previous studies that report a strong association between stress and depressive symptoms [13
Low perceived academic performance at the university was associated with higher depression scores. Depression scores were lower in the group rating their academic performance highest than in the group with lowest performance, but the lowest depression scores were found among students considering themselves average. This suggests a stronger disposition or tendency towards depression or more pressures in the group with the best academic performance. Low performance may be both a cause and a consequence of depressive symptoms. Other studies have found that students with higher depressive symptoms engage more often in social comparison processes and react more negatively to upward comparisons [30
Perceived stress had the strongest association with the depression score in both genders. But whereas females had higher perceived stress scores on average, the impact of perceived stress on depressive symptoms was stronger for males. Gender differences in depressive symptoms increase from preadolescence to young adulthood, evolving around the age of 14 years [31
] and reaching a maximum t about 16–18 years of age [32
]. During this same age span, stress also increases in adolescent girls [33
]. Avison and McAlpine [34
] showed that the higher level of depression in female high school students was explained by higher levels of stress. In our analysis, the female gender continued to be associated with higher M-BDI scores in the bivariate analysis and after adjusting for other variables in the multivariate analysis, until perceived stress was included in the model (which would agree with the above finding). When perceived stress was included in the model only as the main effect, the gender difference disappeared. However, the gender difference reappeared with the interaction between gender and stress score included in the model. Whereas male students had lower depressive symptoms for low levels of stress as compared to females, their depressive symptoms increased more strongly with stress. This would suggest an additional gender difference with respect to coping with stress and its impact on depressive symptoms, which was not examined previously.
In samples that include older adults, higher depression scores are usually associated with a lack of an intimate or marital relationship [35
]. We did find significantly lower depression scores among students who had an intimate partner as compared to those who did not have a partner in the bivariate analysis. However, this association was weak and disappeared after adjusting for perceived stress. Due to the likely rich network of social interactions among students and in university life, we would not expect having a partner to make a strong impact on depression scores Though, since perceived stress does (albeit weakly) mediate the effect of having a partner on depression scores, there is an indication that some level of stress could be associated with not having a partner, even in our relatively young sample.
The sample employed in this study was restricted to first-year students from only one university per country. Thus caution should be exercised in generalizing these findings to all students. The differences we found between the countries might only reflect differences between the three universities that participated in the study. However, we are not aware of any reason to believe that students at these three universities are different from students at other universities in these countries. At each survey site, we achieved a high response rate, preventing selection bias within the investigated populations. We used a relatively recent modification of BDI with the one drawback of limited information (restricted to German samples) on the validity and reliability of the M-BDI [16
]. The study only measured depressive symptoms during the past few days. No information was collected on events within or outside the university setting, which could have an affect on depressive symptoms. At all three universities, the survey was conducted during the second half of the summer term, when end of the term exams were not too distant. However, since the timing of the survey was similar at all three universities, this should not have biased our comparisons. The instrument used to measure perceived burdens was not established prior to this study and, apart of the analysis of internal reliability, no further investigations were performed. Given the cross-sectional character of our study, all investigated relationships were associations and no inference about causal association can be made. In particular, depressive symptoms could have also manifested in higher scores on perceived burdens and stress scales.