Psychometric properties of the Persian version of Depression Anxiety Stress Scale-21 Items (DASS-21) in a sample of health professionals: a cross-sectional study

In this nationwide cross-sectional study, a convenience sampling method was applied. During the COVID-19 pandemic, nurses who worked in public hospitals between September 2020 and July 2021 were recruited in this study. A structured online questionnaire was distributed to gather information from the target population. This study is part of a larger study examining the relationship between burnout and depression, anxiety, and stress, based on self-reports of patient care quality and adverse events among nurses in Iran. One companion paper from this study assessed burnout and its relationship to self-reported quality of patient care and adverse events among nurses [31].

The inclusion criteria were as follows: nurses working full-time in public hospitals; nurses working in medical wards; and nurses with more than a year of working experience in the hospital. On the other hand, newly recruited nurses and part-time nurses were excluded.

The sample size was calculated based on an item-to-respondent ratio of 1:11 [32]. The minimum accepted sample size was 231 participants. However, 1135 nurses participated in the current study during the data collection period. To measure the test-retest reliability coefficient, a previous study showed that 52 participants are required [33]. Seventy participants were randomly selected to complete the questionnaires; also, 57 participants completed the questionnaire again after 3 weeks.

In the current study, DASS-21 with 21 items and three subscales (7 items for each subscale) was used [34]. This instrument measures the prevalence of depression, anxiety, and stress signs and symptoms during the past weeks. The items of the scale are rated on a four-point Likert scale, with the score for each item ranging from 0 (“does not apply to me at all”) to 3 (“applies to me most of the time”). The subscale scores were calculated by summing the scores of the individual items, and the maximum sum for each subscale is 21; higher scores represent higher psychological distress. The original study reported the high reliability of DASS-21, Cronbach’s Alpha coefficients for depression, anxiety, and stress were reported as 0.91, 0.84, and 0.90, respectively [19]. The Persian version of DASS-21 has been previously translated and revised in an Iranian non-clinical setting [30]. This questionnaire includes seven items to assess the sociodemographic characteristics of individuals (i.e., gender, age, marital status, educational level, nursing experience, work unit, shift work, and work hours per week).

The study questionnaire was distributed online via Avalform (https://​avalform.​com), a viable survey tool widely used in Iran. At the beginning of the survey, a brief explanation was given regarding the purpose of the study and how to complete the questionnaire. After reading the explanations, if the participant responded positively to the question about willingness to participate in the study, he/she was allowed to enter the questionnaire; otherwise, he/she will not be able to access the survey questions. The answers to all questions were fixed, and the respondents could not proceed with the questionnaire unless all questions were completed; therefore, no data is missing. One answer to each question can be submitted via phone or computer account. The survey link was sent to the nurses via social media such as Telegram, WhatsApp and LinkedIn. Participating nurses were asked to send the link to other nurses in their departments and hospitals. To achieve a high response rate, the survey was sent out multiple times, and the survey link was available for about 12 weeks.

We used SPSS V.20.0 and AMOS V.21.0 software of Windows for statistical analysis. Descriptive statistics were measured to describe the participants’ characteristics, each DASS-21 subscale, and individual items. Cronbach’s alpha coefficient was also determined to evaluate the internal consistency of DASS-21; a coefficient above 0.70 was reported, which is considered acceptable [35, 36]. Besides, corrected item-total correlations of the three subscales were calculated, which represent the correlation of each item with its subscale when the item is removed. The test-retest reliability method, based on the intraclass correlation coefficient (ICC) with a two-way random model, was used to evaluate reliability. An ICC above 0.7 indicates an acceptable test-retest reliability [37]. Moreover, the construct validity of DASS-21 was examined using a confirmatory factor analysis (CFA) to determine if the Persian version supports the three-factor construct of the original DASS-21. We did not use exploratory factor analysis (EFA) because as far as we knew, DASS-21 was developed using EFA [19]. In general, CFA is a robust method of statistical analysis for examining a predetermined factor structure or hypothetical theory; it describes how well each item evaluates the scale dimensionality. In many studies, CFA has been considered an important index for culturally adapted scales [38].

The goodness of fit was established based on the following fit indices: The Chi-square (χ²) goodness of fit should have a P-value > 0.05; the comparative fit index (CFI) should be ≥0.90 [39]; the Tucker–Lewis index (TLI) should be > 0.90 [40]; the root mean square error of approximation (RMSEA) should be ≥0.08 [40]; the goodness of fit index (GFI) and the adjusted goodness-of-fit index (AGFI) should be ≥0.85 [40], and the standardized root mean square residual (SRMR) should be ≤0.08 [40]. The correlations between the dimensions of DASS-21 were evaluated by measuring Pearson’s correlation coefficients.

The characteristics of the study respondents are presented in Table 1. A total of 1245 nurses completed this survey; however, 110 questionnaires were excluded, based on the inclusion and exclusion criteria. Finally, 1135 nurses were included in the analysis, with a mean age of 33.8 ± 7.8 years. The majority of the participants were female (75.7%), had shift work (77.4%), and had a bachelor’s degree (91.4%). The mean experience of the participants in nursing was 9.6 ± 7.3 years.

In Table 2 the results of the corrected item-total correlations and Cronbach’s alpha coefficients for the index subscales, in addition to the mean scores and standard deviations (SDs) for the items and the three subscales (depression, anxiety, and stress) was presented. The corrected item-total correlations for the DASS 21 were ranged from .59-.77 which could be considered acceptable, from the point of view that more than half of the retained elements have a score in the range of 0.30-0.70 [41]. The corrected item-total correlations were above 0.59 for the items of all subscales. The Cronbach’s alpha coefficient was 0.93 for the Persian version of DASS-21; it was 0.88 for depression, 0.88 for anxiety, and 0.89 for stress, indicating the acceptable internal consistency of the scale. In addition, the test-retest reliability of the Persian Version DASS-21 was acceptable, and the ICC of all domains was adequate which ranged from 0.75 to 0.86. Table 3 displays the test-retest reliability.

The analysis yielded a 21-item three-factor model that fit the data of the Persian of DAAS-21 (Fig. 1). The CFA revealed the acceptable fit indexes of the three-factor model (Table 4). The goodness-of-fit indices in the CFA showed RMSEA of 0.078, TLI of 0.91, CFI of 0.92, NFI of 0.91, SMRS of 0.047, and GFI of 0.88. Except for the AGFI, the fit indices were lower than the cutoff points in our study. Also, χ²/df was higher than the cutoff point (χ²/df = 7.83) [25, 38, 40, 42, 43]. The factor loadings ranged from 0.61 to 0.86 for all items, and the correlation coefficients between the subscales were between 0.86 and 0.94. Accordingly, the good fit of the three-factor model was established, suggesting the stability of this instrument.

Table 5 presents the correlations (r) between the subscales of the Persian DASS-21. The subscales of the Persian DASS-21 were found to be significantly correlated. The coefficients for the correlations of subscales were as follows: 0.77 for depression and anxiety; 0.83 for depression and stress; and 0.82 for anxiety and stress.

There are some limitations to the present study, Due to coronavirus concerns, data was collected online and using a convenient sampling method, which could result in selection bias. Also, only nurses with access to the Internet could complete the questionnaire. Besides, other healthcare professionals, such as physicians, laboratory technicians, and paramedics, were not included in this study. Nevertheless, the present study established the validity and reliability of the Persian version of DASS-21, which can be applied in future research to measure depression, anxiety, and stress in different healthcare professionals in the Iranian healthcare system.