1 Introduction

Depression, anxiety, and stress are prevalent mental health concerns on a global scale, posing substantial challenges to persons' overall functioning [1]. According to available data, it has been estimated that around 322 million people have been affected by depression, while over 264 million individuals suffer from anxiety. These figures correspond to approximately 5% and 4% of the world population, respectively [2]. Depressive and anxiety symptoms are becoming more common, ranging from 9 to 27% [3,4,5]. Several recent research undertaken in Bangladesh examined healthcare workers' mental health during the COVID-19 epidemic. These investigations have shown that over 50% of healthcare professionals had symptoms of anxiety and depression [6,7,8]. Depression and anxiety exhibited a correlation with a range of social and physical consequences, such as strained familial connections, elevated rates of suicide (exceeding 800,000 annually), and diminished occupational productivity [1,2,3,4,5].

The Depression, Anxiety, and Stress Scale (DASS-21) is a widely utilized measure globally among both clinical and non-clinical populations [9]. The first iteration of the DASS consisted of 42 questions and was designed as a self-report measure to evaluate three psychological constructs: depression, anxiety, and stress [10]. The 21-item DASS scale is the brief version of the 42-item DASS [11]. The DASS-21 is more often used and favored than its extended counterpart because of its ease of research deployment and reduced time requirements. Furthermore, it is worth noting that the shorter version has the capability to evaluate all three aspects of mental health issues by using a single scale and minimizing the potential overlap between symptoms of depression and anxiety [12, 13].

Despite the absence of adequate validation in the specific cultural and linguistic context of Bangladesh, the DASS-21 has been used by researchers in several studies conducted among different groups of population in Bangladesh [6, 12, 14]. The DASS-21 has been translated and validated in several languages around the globe, including Asian, South Asian, and Southeast Asian regions. This validation has occurred in multiple countries, such as India, Turkey, and Nepal [15,16,17]. The DASS-21 has shown consistent outcomes and exhibited robust concurrent validity with several scales used to measure mental health symptoms in validation studies conducted across many nations and languages [10, 11, 13].

It is essential to ensure that the research tools used are evidenced within the particular setting and environment prior to commencing the investigation. The need for this validation process arises from variations in language, culture, country-specific context, emotional nuances, modes of expression, linguistic dialects, and social characteristics [18]. Although the DASS-21 questionnaire has been translated into Bengali [19], its validation among healthcare professionals has not been conducted. The absence of empirical support pertaining to the authenticity and consistency of the initial measurement instrument has the potential to engender ambiguity for scholars, those with vested interests, and consumers of scholarly literature. Given the circumstances above, the primary objective of this research was to assess the validity, reliability, and factorial composition of the Bengali-translated version of the DASS-21 questionnaire among Bangladeshi healthcare professionals. The outcomes of this study will not only make a valuable contribution to the comprehension of the psychometric characteristics of the scale but also offer valuable insights for future researchers intending to employ the DASS-21 among healthcare professionals or in other healthcare-related contexts in Bangladesh.

2 Materials and methods

2.1 Study design and participants

The study employed a cross-sectional design and utilized convenience sampling to recruit 258 respondents, including 62 doctors and 196 nurses, between November 2020 and February 2021. These respondents were drawn conveniently from 28 health centers in Bangladesh, comprising 20 private hospitals and 8 government hospitals. The selected health centers covered 18 out of the 64 districts in Bangladesh. In order to be included in the study, respondents had to meet certain criteria: (a) they had to be registered health professionals in Bangladesh as recognized by the national institutions, i.e., Bangladesh Medical and Dental Council [20] and Bangladesh Nursing and Midwifery Council [21], (b) they had to be native Bengali speakers, and (c) they had to willingly participate in the study without expecting any benefits. Initially, 312 health professionals were interested in the data collection process. All of them were provided with the data collection tool, but ultimately, only 258 respondents completed the questionnaire, while the remaining participants were excluded from the study due to incomplete information.

2.2 Measures

This study utilized a semi-structured data collection instrument, which included questions to gather respondents’ socioeconomic information such as age, gender, marital status, and professional details. Additionally, the instrument incorporated relevant scales aligned with the study objectives.

2.2.1 Depression, anxiety, and stress scale—21 (DASS-21)

Lovibond et al. developed DASS in 1995, and the DASS-21 is the brief version of the original DASS [10]. The tool was designed to examine mental health symptoms, such as depression, anxiety, and stress, among the adult groups of the population. The manual of DASS-21 suggests that, at first, the respondents were instructed to remember their past seven days’ life experiences. Then, they were given the DASS-21 and asked to rate each reasonable statement. Each statement of the DASS-21 has four responses—from “0” to “3”. The “0” indicates—“Did not apply to me at all/Never,” and “3” indicates “Applied to me very much or Most of the time—Almost always”. The scores are calculated on three subscales, i.e., Depression, Anxiety, and Stress. Each subscale comprises 7 items, and their score extends from 0 to 21. The DASS-21 has been translated into Bengali among medical students by Alim et al. [19]. Although Alim et al. have tried to validate the DASS-21, the authors of this study have noticed some errors in their tool validation methodology and outcomes. Therefore, we utilized only the Bengali-translated questionnaire of that study [19]. The English and Bengali versions of DASS-21 can be found in Supplementary File 1 and Supplementary File 2, respectively.

2.2.2 Patient health questionnaire—9 (PHQ-9)

The patient health questionnaire-9 consists of nine items, and this instrument is used to evaluate the level of depression [22]. As per the manual of the PHQ-9, respondents were queried about their past two weeks of mental health symptoms. Each statement of the PHQ-9 contains four responses, and the rating scale ranges from “0—Not at all” to “3—Nearly every day.” The minimum score of this tool would be 0, and the maximum would be 27. The PHQ-9 tool identifies five types of depression severity, i.e., “None”, “Mild Depression”, “Moderate Depression”, “Moderately Severe”, and “Severe”. This study employed the Bengali version PHQ-9 scale [23]. The Bengali PHQ-9 tool has been validated in some diverse samples of Bangladesh [23, 24].

2.2.3 Generalized anxiety disorder (GAD-7)

The generalized anxiety disorder (GAD-7) scale is a seven-item data collection instrument used to screen rapidly and rapidly assess the severity of the generalized anxiety disorder [25]. The GAD-7 is a four-point rating tool that ranges from “0—Not at all” to “3—Nearly every day”. The minimum summed number of this scale would be 0, and the highest calculated number would be 21. According to the manual, the calculated total score of the GAD-7 could be categorized into four groups—from “Minimum Anxiety” to “Severe Anxiety” [26].

2.3 Data collection procedure

Prior to the survey, all respondents were notified in advance and asked to provide their availability for scheduling. The data collection tool ensured respondent anonymity as it did not include any identifying questions. On average, the interviews took approximately 32 min to complete. After completing the questionnaire, all questionnaires were securely stored in sealed boxes. After the survey, the data collectors returned the sealed boxes containing the questionnaires to the respective researchers.

2.4 Statistical analysis

The final data were thoroughly checked for accuracy, consistency, completeness, and missing values before the final analyses. Descriptive statistics were used to present the participants’ socioeconomic information. The categorical variables were presented by frequency and percentages. Mean and standard deviations were calculated for the continuous variables. The Shapiro–Wilk’s test was applied to assess the data normality. The internal consistency and the homogeneity of the DASS-21 were measured using Cronbach’s alpha coefficient and the test–retest reliability test. The KMO (Kaiser–Meyer–Olkin) test and Bartlett’s test of sphericity were used to measure the data eligibility to perform the factor analysis. Confirmatory factor analyses were done to test various models of DASS-21. This study assessed three models with one-factor, two-factor, and three-factor structures. All the essential fit statistics, such as likelihood ratio chi-square statistics (χ2), Akaike's information criterion (AIC), Root mean square error of approximation (RMSEA), Comparative fit index (CFI), Bayesian information criterion (BIC), and Standardized root mean squared residual (SRMR); Tucker-Lewis index (TLI) for each of the models were estimated. For the best-fitting model, the CFI value would be ≥ 0.90; the p-value ≤ 0.05; RMSEA ≤ 0.60; SRMR ≤ 0.08; the model with a lower AIC value would be the best-fit model. The Pearson correlation measured the concurrent validity with the PHQ-9 and GAD-7. The R statistical computing software version 4.1.0 was used to analyze the data [27]. The following R packages were used to analyze the data—“GPARotation”, “Laavan”, “PerformanceAnalytics”, and “Hmisc”.

3 Results

The socioeconomic characteristics of the respondents are shown in Table 1. This study enrolled 258 respondents whose mean (± SD) age was 27.15 (± 4.7). More than half of the respondents were female (60.9%), and 46.5% were married. Around one-third of the respondents (37.2%) worked in tertiary hospitals. Most of the respondents (82.9%) lived in urban areas. About 1 out of 10 respondents in this study had at least one comorbidity. The descriptive statistics of each of the items of the DASS-21 scale are outlined in Table 2. All the items depict relatively normal distribution except one item. Nevertheless, all the items of the DASS-21 were considered for the final analysis to determine the influence of that item on the reliability of the DASS-21 scale.

Table 1 Socio-demographic characteristics of 258 enrolled health professionals
Full size table
Table 2 Descriptive statistics of the Depression Anxiety Stress Scale-21 (N = 258)
Full size table

3.1 Summary statistics of the depression, anxiety, stress—21 scale

Table 2 displays the summary statistics of the Bengali-translated DASS-21 scale for the study's total sample, only among the male respondents of the survey and among the female respondents. The raw scores of the DASS-21 scale and its subscales were multiplied by two to obtain the standardized scores of the DASS scale. The standardized scores are portrayed in Table 3 for comparing the DASS-21 and DASS scales.

Table 3 Summary statistics and reliability of DASS-21 scale and its subscales among 258 health professionals of Bangladesh
Full size table

3.2 Factorial construct validity

The overall score of the Kaiser–Meyer–Olkin test for the Bengali-translated DASS-21 scale was 0.94, i.e., an allowable estimate to perform the factor analysis. Bartlett’s test of sphericity (K-squared = 166.02, p-value < 0.001) was also measured, indicating the data acceptability for factor analysis (Table 4).

Table 4 Standardized scoresa of the DASS-21 scale among 258 health professionals of Bangladesh
Full size table

Table 5 illustrates the DASS-21 scale’s factorial structure and fit indices of the various models. In the beginning, Model A was tested, a one-factor model of the DASS-21 scale. The Model A exhibits a poor fit indices (χ2 = 444.3; df = 189; p < 0.001; RMSEA = 0.072; AIC = 11751.9; BIC = 11901.2; CFI = 0.89; SRMR = 0.054; TLI = 0.88) of DASS-21 to the data. In the following phase, Model B, i.e., the two-factor model suggested by Sinclair et al., was tested [28]; however, no satisfactory fit to the data could be established (χ2 = 393.2; df = 169; p < 0.001; RMSEA = 0.072; AIC = 11277.1; BIC = 11422.8; CFI = 0.90; SRMR = 0.052; TLI = 0.89). Finally, we tested the conventional 3-factor model, i.e., Model C and the fit indices of Model C describe satisfactory fit indices (χ2 = 372.4; df = 186; p < 0.001; RMSEA = 0.062; AIC = 11686, BIC = 11845.9; CFI = 0.92; SRMR = 0.051; TLI = 0.91). In Model C, we can observe that the chi-square value was the lowest out of the three models; similarly, the CFI and TLI values are also low compared to Models A and B, which indicates a good model fit. Although the AIC and BIC values are slightly lower than Model A, considering the overall features of Model C, we identified Model C as the best fit for the scale of interest. 

Table 5 Model fit indices for confirmatory factor analysis of DASS-21 among 258 Bangladeshi health professionals
Full size table

Table 6 demonstrates the standardized factorial loadings and all the loadings of Model C, i.e., the conventional 3-factor model of DASS-21. The loadings of every item of DASS-21 were statistically significant (p < 0.001), and the loading values implied an excellent correlation.

Table 6 Results of confirmatory factor analysis (standardized factor loadings), internal consistency (Cronbach’s alpha), and test–retest reliability (intraclass correlation coefficient) of the DASS-21 in 258 health professionals
Full size table

3.3 Reliability of the Bengali translated DASS-21

The reliability of the Bengali-translated DASS-21 scale was measured by Cronbach’s alpha coefficient and the intraclass correlation coefficient test. Both tests showed good internal consistency, i.e., Cronbach’s Alpha gave a value of 0.93 for the overall DASS-21, and the intraclass correlation coefficient gave an overall value of 0.92. In the same pattern, the three subscales of the DASS-21 had good internal consistency and reliability (Table 6).

3.4 Discriminant validity

The differences between the χ2 value (Δχ2) presented in Table 5 indicate that if the correlation remained constant, the comparison models had a more inadequate fit compared to the first model; thus, it provides evidence of discriminant validity.

3.5 Convergent validity

Figure 1 illustrates the evaluation of the DASS-21’s convergent validity. The Pearson correlation matrix was employed to assess the correlation between the total score of the DASS-21 scale and this study's relevant indicators. All the indicators showed high correlation and significance (p ≤ 0.001 and r > 0.5) with the total score of the DASS-21 scale. The significant positive correlation between stress and anxiety (r = 0.94, p < 0.001) and anxiety and depression (r = 0.80, p < 0.001) was found. Furthermore, the depression and PHQ-9 portrayed a large effect (r = 0.93) with significant correlation (p < 0.001). A positive correlation with high significance was also found between the PHQ-9 and GAD-7 scales (r = 0.92, p < 0.001).

Fig. 1
figure 1

Correlation matrix between the total score of DASS-21 and relevant indicators. Pearson correlation method was applied as the normality of the variables was obtained. DASS21 (Depression, Anxiety and Stress Scale-21); stress (Stress subscale of DASS-21); anxiety (Anxiety subscale of DASS-21); depression (Depression subscale of DASS-21); depression (Depression subscale of DASS-21); phq9 (Patient health questionnaire-9); GAD7 (General anxiety disorder-7); age (Age of the participants)

Full size image

4 Discussion

The present research included a comprehensive investigation into the factorial structure and convergent validity of the Bengali adaptation of the Depression Anxiety Stress Scale-21 (DASS-21) among healthcare workers in Bangladesh. The findings indicate that the Bengali adaptation of the DASS-21 has a high level of reliability and demonstrates good convergent validity in the cultural setting of Bangladesh, which focuses on health professionals.

The data from health experts in Bangladesh indicates that the typical three-factor model of the DASS-21, which encompasses depression, anxiety, and stress-related components, is the most suitable match. Further investigations have substantiated the results carried out in various nations, including diverse linguistic and cultural settings [27, 28, 30, 31]. Previous research, including the present study, has shown a robust association among the various subscales [31,32,33,34]. One potential approach to address this significant association is to condense the whole structure into smaller factors [34]. As delineated in the findings section, the current investigation demonstrates insufficient congruence for the single and two-factor models. The aforementioned outcome is consistent with other investigations carried out by other academics, who have also documented comparable outcomes [27, 30, 31, 35]. The lack of adequacy in the single-factor model of the DASS-21 scale was not surprising, given that the original scale was designed to assess the multidimensional mental health symptoms in the general population, namely depression, anxiety, and stress [9].

Several additional researches have shown that the two-factor model exhibited superior fit compared to the three-factor model [32, 34, 35]. Nevertheless, it is essential to acknowledge that the studies above may have failed to consider a critical aspect: the general variables’ robust loadings and the specialized components' comparatively feeble loadings. Furthermore, an alternative approach proposed by a separate cohort of researchers involves using the cumulative scores derived from the Depression Anxiety Stress Scales-21 (DASS-21). As stated before, this methodology is not favored because the original scale was specifically developed to evaluate symptoms of multidimensional mental health [9].

The present research provides strong justification for the convergent validity of the scale since it demonstrates a good correlation with other instruments that assess comparable themes [36]. The convergent validity of the DASS-21 scale was also assessed by researchers from various nations [27, 30, 31]. The current investigation assessed the validity of the 9-item public health questionnaire and the 7-item generalized anxiety disorder scale. EH Lee et al. assessed the convergent validity of the DASS-21 in the Korean environment using the techniques above [27].

The internal consistency of the DASS-21 scale and its subscales was found to be good, as shown by Cronbach’s alpha scores ranging from 0.81 to 0.93 and the intraclass correlation coefficient scores ranging from 0.80 to 0.92. The results presented in this study align with previous research undertaken on a global scale, including both clinical and non-clinical samples. These prior investigations have likewise revealed comparable levels of consistency [1, 2, 4, 5, 8].

Therefore, this research confirms that the Bengali adaptation of the Depression Anxiety Stress Scale-21 (DASS-21) has three distinct subscales, namely depression, anxiety, and stress. The good validity, reliability, and factorial structure of the DASS-21 were determined within the specific context of Bangladesh and its healthcare population. Hence, this instrument may be efficiently used in forthcoming studies and initiatives to evaluate mental well-being, including depression, anxiety, and stress, in Bangladesh and other contexts where healthcare professionals communicate in the Bengali language.

5 Conclusion and recommendation

Based on the empirical evidence obtained from our research, we strongly advocate for the utilization of the Bengali-translated Depression, Anxiety, and Stress Scale—21 (DASS-21) as a dependable and valid instrument for evaluating the mental well-being of Bengali-speaking healthcare professionals. This recommendation is applicable not only to Bangladesh but also to specific Bengali-speaking regions in India, such as West Bengal, Tripura, Assam, and others, due to the cultural similarities observed. The study’s rigorous methodology, which encompasses a substantial sample size and the inclusion of participants from diverse healthcare facilities across various districts in Bangladesh, enhances the generalizability of the findings.

6 Strengths and limitations

This study provides the first comprehensive validation of the DASS-21 scale within the distinct context of Bangladesh, particularly among healthcare professionals who speak Bengali. The results of the study provide strong evidence supporting the validity, reliability, and factorial structure of the Bengali version of the DASS-21 instrument, thereby confirming its favorable psychometric properties. One of the notable drawbacks of the study is the relatively high dropout rate of participants (18%), which might be attributed to inadequate questionnaire responses. The significant dropout rate seen in this research may be attributed to the situation surrounding its execution, that was, the COVID-19 pandemic. Given that healthcare personnel were contacted inside hospital settings, it is plausible that their constrained availability hindered them from being able to participate fully in the interview process.