Introduction

It is estimated that depression occurs in about 5% of adults [1]. Young adults are particularly vulnerable to experiencing depressive symptoms [2, 3] due to the peculiarities of the developmental period of emerging adulthood, characterized by a sense of instability, self-focus and exploration of one’s own identity [4]. The need to adapt to new conditions is a challenge and an additional burden that arises for those making decision to continue education at university [5]. The years spent at university are an increased risk period for depression onset–particularly for the occurrence of first episodes [6]. In addition to anxiety symptoms, depression is a frequently diagnosed problem among university students [7]. A nationwide survey conducted in 2019 with 4500 university students found that 30% had symptoms of depression, including 6.5% with moderate to severe depression [8].

According to an annual survey by the Association for University and College Counseling Center Directors (AUCCCD), an international organization of university representatives from the United States, Canada, Europe and Asia, in 2019 nearly 90% of counseling center directors reported an increase in the number of students seeking support [9]. Along with stress and anxiety, depression is the most diagnosed problem among students [7, 9–13]. In addition, female gender is a risk factor for experiencing depressive symptoms. This tendency occurs in the general population [14–16], as well as in university students [2, 17–19], among whom the mentioned trend was also present during pandemics [20–22].

Findings from studies conducted during the COVID-19 pandemic showed that young adults are also more susceptible to experiencing depressive symptoms compared to older adults. The study, which was carried out in 2020 during lockdown on a representative sample of Polish adults, confirms that high levels of depressive symptoms were manifested primarily by 18–29-year-olds [23]. Reports from other studies indicate that undergraduate students are more afraid of COVID-19 than graduates [24]. In addition to experiencing the same anxiety and stress related to the epidemiological situation as others, and sometimes being infected themselves or facing disease or the loss of a loved one, they experienced the stress of the enforced isolation more strongly [22].

Studies on university students mental health report that there is a strong relationship between the severity of depressive symptoms and phenomena such as suicidal ideation, suicidal feelings/ actions and problematic substance use [25, 26]. The mental health of students is also related to the course of their study. According to Eisenberg et al. [27], a higher probability of dropping out of college and also obtaining a lower grade point average (GPA) are predicted by depression.

From this perspective, it is a particularly important and urgent challenge to take action against depression, with a special focus on young adults. Untreated, it can persist for a long time and interfere with daily functioning, and in severe cases can cause substance abuse or even lead to suicidal behavior. Recognizing the warning signs and making an early diagnosis is therefore crucial to treating the symptoms of depression and preventing its recurrence.

In this context, systematic screening carried out on university campuses promotes early identification of mental health problems, their nature and, consequently, better adaptation of the offer of assistance, as well as of preventive actions [2, 28]. The use of tools with proven psychometric properties can be helpful in this process and one of the suggested tools is the Center for Epidemiologic Studies Depression Scale–Revised (CESD-R).

The CESD-R is a revised version of the CESD, created by Lenore Radloff [29] and developed by William Eaton and colleagues [30]. The updated version covers the symptoms of depression included in the DSM-IV-TR [31] and the DSM-5 classification [32]. It consists of 20 statements regarding mood and behavior over the past two weeks (in the previous version of the tool, it was 7 days) and includes nine depressive symptoms: sadness/dysphoria (2, 4, 6), loss of interest/anhedonia (8, 10), appetite (1, 18), sleep (5, 11, 19), thinking/concentration (3, 20), guilt/worthlessness (9, 17), fatigue/fatigue (7, 16), movement/agitation (12, 13) and suicidal ideation (14, 15) (https://cesd-r.com). The maximum possible score on the CESD-R scale is 80 and the minimum score is 0. The original version set the cutoff point at 16. A study by Van Dam and Earleywine [33] compared two types of classification, including a cutoff point of 16 points. This solution resulted in a significant number of false positives compared to the algorithm-based classification proposed by Eaton et al. [30].

The CESD-R scale is free, publicly available and can be used by non-mental health professionals. An additional advantage is that it can be completed in no more than 4 minutes. All of this is encouraging for widespread use of the tool in studies involving representatives of various populations (primarily general populations) in different parts of the world. Currently, 15 language versions of the scale are available (https://cesd-r.com), including a Polish version developed by Koziara [34]. The authors of the revision suggest interpreting the test score, without isolating separate components [30]. However, research using this scale results in different solutions for the number of factors extracted using principal component analysis (PCA) and exploratory factor analysis (EFA) and confirmed by confirmatory factor analysis (CFA).

The factor analysis carried out in the Polish study by Koziara made it possible to distinguish three factors: cognitive-affective (14 items), physical (3 items) and self-destructive (2 items). Factor loadings did not allow item 11 to be assigned to a specific factor [34]. To our knowledge, the structure obtained in this way–a three-factor structure–has not been subjected to CFA verification on the Polish population. In a Nigerian study conducted among students [35], the solution obtained by the Polish researcher was tested, among others, and turned out to have the best fit to the data. Furthermore, a model with two factors [33] has been analyzed. A single-factor solution including all 20 items was tested using EFA by Japanese researchers [36]. In Brazil [37] and Jordan [38], a single-factor model was verified using CFA. In the aforementioned studies, a measurement invariance analysis was performed for gender, but it should be mentioned that in the Jordanian version this analysis was conducted for the 12-item version of the CESD-R [38].

Data on the measurement of depressive symptoms using the CESD-R present inconclusive results relative to the scores obtained by men and women. In a study by Kokou-Kpolou et al. [35], the authors noted that there were no differences in resolution with a single global factor, similar to a Japanese study [36]. Information presented by Koziara [34], on the other hand, indicates a higher average score among men. In contrast, a study conducted among Iraqi students identified more depressive symptoms among women [39], but data on average scores in both groups are not available.

The above information highlights an urgent need for research on the phenomenon of depression, including use of the CESD-R, among young adults as a high-risk group and provides justification for conducting our own research. The purpose of our study was to empirically verify the psychometric properties of the CESD-R. However, due to the lack of data on the Polish population in terms of confirming the structure of the tool using CFA and measurement invariance for gender, it was decided to carry out studies that would allow such analyses.

Materials and methods

Results

Discussion

This study tested the psychometric properties of the Polish version of the CESD-R among students. Based on the results presented in other studies, both three- and one-factor solutions, as well as versions with a reduced number of items, were verified. Based on the analyses, a single-factor solution with nine questions was adopted, which referred to all the criteria for depression presented in the DSM-5 classification. This version was used to verify the measurement invariance within gender and the result of this analysis confirmed that it was present at all levels.

The CESD-R, an update of a tool that has been in use for years in depression research in the general population, has been successively revised in various countries, as evidenced by its involvement in a growing number of studies. The authors of the original version [30] did not presume to extract factors but analyses conducted on the basis of the collected research data indicate that it is possible; however, in some cases the exploratory extracted multi-factor structure is confirmed, as in Kokou-Kpolou et al. [35] and in others it is not e.g., the four-factor solution from the study of Rababah et al. [38]. In a study of the Polish version of the scale, Koziara [34] extracted a three-factor solution based on EFA but without confirmation for this solution within CFA. A research team conducting a study among Nigerian students [35] using only CFA showed that the three-factor solution obtained by Koziara [34] had the best fit to the data and that in the gender invariance analysis full invariance was obtained at the configural and metric levels but only partial confirmation at the scalar and strict levels. In our own study, three factors also emerged and the model itself had good parameters; however, due to the very high intercorrelations of the factors it was decided not to subject this version to further analysis, including measurement invariance, in favour of verifying the univariate solution, which is also supported by other studies [37, 38].

Analyzing in detail the data for the items that make up the CESD-R, we observed that item 11 ("I slept much more than usual") and item 18 ("I lost a lot of weight without trying to") obtained the lowest factor loadings each time. In the Polish study by Koziara [34], item 11 was also problematic because the factor loadings did not allow it to be specifically assigned to any of the three extracted factors. Furthermore, in the Faro et al. [37] study, item 11 had the lowest factor loadings in the EFA procedure and one of the lowest loadings in the CFA. Item 18 on weight loss has not performed unequivocally poorly in other studies. In Koziara [34], item 18 had a very high factor loading in EFA (λ = .86), being part of the "physical" factor including three items. In Faro et al. [37], item 18 had a factor loading of λ = .50 in EFA and λ = .40 in CFA, both of which can be considered relatively low values. In the Rababah et al. [38] study, item 18 remained in the CESD-R after being shortened but the CFA factor loading value was the second lowest. Furthermore, in a publication by Van Dam and Earleywine [33], when testing a single-factor solution, items 11 and 18 received the lowest factor loadings of all items: .62 and .61, respectively.

The presented regularities of the indicated items are worth interpreting, taking into account the period of implementation of our own research, which to some extent can explain their low factor loadings. The research took place during the period of lockdown and distance learning at universities in Poland. This involved students staying at home or in residential facilities, without the possibility of moving to university buildings. These circumstances may have influenced the widespread increase in time spent sleeping, resulting from not having to get up early and prepare to spend the day away from home. For this reason, the increase in sleep time may not have been considered as a different condition than before. Decrease in physical activity due to restrictions caused by the COVID-19 pandemic [44] and dietary choices and eating habits [45] might have promoted weight gain. The CESD-R item on weight loss (item 18) achieved the lowest mean score (M = .41) in the entire sample analyzed, even lower than item 15 on the desire to harm oneself (M = .45). It seems, therefore, that items 11 and 18 may have worked differently in the pandemic context than in other social circumstances, although, as indicated earlier, these items, or more precisely the values of their factor loadings, are outliers in other studies.

In our own study, invariance of reduced 9-item version of CESD-R was confirmed at four levels of measurement, indicating that the results obtained with this scale can be compared across genders. Other studies using the scale, whether for the full [35] or the shortened version [38], have also considered and demonstrated gender measurement invariance analysis. Our results for the abbreviated version of the scale therefore do not differ from data collected in other countries or on age-matched and socially similar populations.

Conclusions and future directions

As presented in the Discussion, the CESD-R is successively undergoing analyses aimed at determining its basic psychometric properties and verifying its structure. Studies conducted in different countries and among different age and social groups confirm the single-factor structure of the tool (e.g. Faro et al. [37]) and also indicate the possibility of extracting more factors [34, 35] or reducing the number of items forming the scale (e.g. Rababah et al. [38]). In our presented study conducted among students during the lockdown period associated with the COVID-19 pandemic and the time of remote learning, we verified various solutions present in the literature for the CESD-R, starting with a three-factor solution and ending with a single-factor version using a reduced number of items (nine items). The selected solution was subjected to a gender-invariance analysis and this measurement invariance was confirmed, allowing mutual comparisons between the men and women tested with this version of the scale.

An important next step, taking into account the shortened 9-item version of the CESD-R, would be to verify the criterion validity of the tool. It would be desirable to test convergent validity (using other existing tools to measure depression), divergent validity (using tools to measure other variables that should not correlate with depression). A valuable action in relation to the usefulness of the shortened version of the scale would also be to collaborate with a psychiatrist who, in a follow-up, could make the diagnosis of depressive disorder. By treating, for instance, the diagnosis made by the psychiatrist as the gold standard, it would also be possible to check the sensitivity and specificity of the scale, as well as to determine, using the Receiver Operating Characteristic (ROC) curve, what number of points obtained by the person completing the tool represents the best cut-off point to be used as an indication for an individual to consult a health professional. Using a tool with proven psychometric properties can facilitate the organization of surveys among students, through which the need among them for support will be recognized. This is particularly important in view of the increasing incidence of depression in this group. In Poland, psychological support is not an obligatory part of the functioning of higher education institutions, although many universities are making changes in this regard [46]. The results of the research described in this paper were used to adjust the assistance offered in order to meet the demand of students in the target university.

Acknowledgments

The authors would like to thank all students for their voluntary participation in this study.