Agomelatine

Validation of the 21-item Depression, Anxiety, and Stress Scales (DASS-21) in individuals with autism spectrum disorder

Abstract

The purpose of the study was to examine the internal consistency and validity of the 21-item Depression Anxiety Stress Scale (DASS-21) in individuals with Autism Spectrum Disorder (ASD) and without intellectual disability (IQ >= 70). Participants (NN = 123) were consecutively recruited from the Brain and Mind Centre in New South Wales, Australia. Internal consistency was determined using Cronbach’s alpha. Item-total correlations were evaluated by Pearson’s product-moment correlation coefficient. The convergent validity of the DASS-21 was examined by measuring its associations with quality of life and other measures of depression and anxiety. Factorial validity was assessed using confirmatory factor analysis. The DASS-21 demonstrated good internal consistency, adequate convergent validity, and all items exhibited satisfactory item-total correlations. Considering fit indices and factor loadings, the confirmatory factor analysis results provided support for the original 3-factor oblique model consisting of depression, anxiety, and stress factors. The model fit could be further improved with some modifications. Overall, the results indicate that the DASS-21 is a viable self-report screening measure for depression, anxiety, and stress in individuals with ASD and without intellectual disability.

1. Introduction

Autism spectrum disorder (ASD) is the collective term for neuro- developmental disorders characterized by impaired social interaction and communication, as well as a restricted range of activities and in- terests (American Psychiatric Association, 2013). Co-occurring depres- sion and anxiety are common in this population (Eaves and Ho, 2008). The reported prevalence rates for depression disorders in adults with ASD range from 38% to 70%, while the reported prevalence rates for anxiety disorders range from 50% to 65% (Hofvander et al., 2009; Lugnegard et al., 2011). Their presence has been found to exacerbate the core features of ASD (Rai et al., 2018). Despite the high prevalence, diagnosis, and treatment of these co-occurring conditions in ASD, they are often under-recognized and under-treated (Mazzone et al., 2012).

Co-occurring depression is associated with impaired social com- munication, maladaptive behaviours, lower adaptive functioning and a higher risk of suicide in individuals with ASD (Cassidy et al., 2014; Stewart et al., 2006). Cassidy et al. (2014) found that 66% of in- dividuals with ASD reported suicidal ideations, and 35% reported past planned or attempted suicides. There is also considerable overlap between the symptoms of ASD and depression, such as social with- drawal, appetite, and sleep disturbance, which can also result from difficulties related to autism (Stewart et al., 2006). Consequently, it has been suggested that clinicians and researchers may overestimate the severity of depressive symptoms in individuals with ASD when using traditional depression measures developed for those without ASD (Cassidy et al., 2018).In people with ASD, higher levels of anxiety are associated with increased functional impairment (McKnight et al., 2016), theory of mind impairment (Ozsivadjian and Knott, 2011), re- petitive behaviours (Lidstone et al., 2014), and insomnia symptoms (Richdale et al., 2014). The severity of anxiety symptoms in individuals with ASD may also be challenging to diagnose due to the overlapping symptoms (e.g., social avoidance and repetitive behaviour) of anxiety and ASD (Hollocks et al., 2019; Kerns and Kendall, 2012). As a result, the reliability and validity of traditional anxiety measures in individuals with ASD have been questioned (Vasa et al., 2018).

Given that co-occurring depression and anxiety can profoundly affect individuals with ASD, there is a crucial need to identify reliable and valid measures of these constructs in this cohort. However, there has been limited research identifying reliable and valid self-report measures (Cassidy et al., 2018; Rodgers et al., 2016). Many researchers have highlighted the need to validate existing tools used in the general po- pulation before developing specific instruments for ASD, partly to en- sure people have access to mainstream assessment and treatment ser- vices (Cassidy et al., 2018; Rodgers et al., 2016).

As a self-report questionnaire for measuring depression, anxiety, and stress, the 21-item Depression, Anxiety, and Stress Scale (DASS-21; Lovibond and Lovibond, 1995) has been used extensively in a wide range of research and clinical settings, in part due to its ability to adequately differentiate between specific anxiety and depression clus- ters of symptoms (Antony et al., 1998; Henry and Crawford, 2005; Lovibond and Lovibond, 1995). Unlike the structured or semi-struc- tured interviews that can take 1 to 4 h of administration, the DASS can be administered quickly by clinicians and as a self-report instrument. The DASS is designed to measure the dimensions of depression, anxiety, and stress that vary along a continuum of severity. Therefore, the DASS cannot be used on its own to determine the presence or absence of depression or anxiety and should not replace a clinical interview. However, high scores on the DASS alerts clinicians to identify patients who are at high risk of being affected by depression or anxiety, in- dicating the need for further diagnostic assessment (Lovibond and Lovibond, 1995). In addition to the use of DASS as a screening tool, the DASS can also be used to assess the patient’s response to treatment (Ronk et al., 2013).

In comparison to the DASS, most traditional self-report scales (e.g., Hamilton Rating Scales for Anxiety and Depression; Hamilton, 1959, 1960) may not distinguish well between anxiety and depression, as these scales predominantly measure the common factor of negative af- fectivity (Moras et al., 1992; Watson and Clark, 1984). Several studies have evaluated the factor structure of the DASS-21 using confirmatory factor analysis in both clinical (e.g., Clara et al., 2001) and nonclinical samples (e.g., Henry and Crawford, 2005). Previous studies have reported adequate internal consistency of the DASS-21 (α = 0.88–0.92 for depression, 0.79–0.83 for anxiety, 0.84–0.86 for stress and 0.93 for total) in adults with ASD (Cage et al., 2018; Maddox and White, 2015; Nah et al., 2018). There was, however, no report of its convergent and factorial validity, which is required to determine its utility for this population (Cassidy et al., 2018). In Australia, the 21-item version of the DASS (DASS-21) is also used nationally for reporting on publicly funded mental-health care plans for psychological therapy, owing to its excellent psychometric properties and free public use (Pirkis et al., 2011). Given that the scale has not been previously validated in ASD populations, this has led to limited use and subsequent barriers for accessing subsidized mental health care for the ASD population. It is, therefore, critical that its psychometric properties are evaluated in neurodiverse populations to encourage its use and appropriate access, assessment, and monitoring for publicly funded mental health pro- grams.

The present study aimed to evaluate the psychometric properties of the DASS-21 in individuals with ASD and without intellectual disability. More specifically, we assessed the instrument’s internal consistency, item-total correlations, convergent validity, and factorial validity. It was hypothesized that in adults diagnosed with ASD, all three scales of the DASS-21 would show an adequate level of internal consistency and correlate significantly with other instruments of depression and an- xiety, such as Hamilton Rating Scale for Depression (HAM-D; Williams, 1988) and Liebowitz Social Anxiety Scale Self-Report (LSAS- SR; Fresco et al., 2001; Liebowitz, 1987). Further, it is expected that all three DASS-21 scales would correlate negatively with quality of life (QoL) measured by the 12-Item Short Form health survey (SF-12; Ware et al., 1996). Finally, we predicted that all items on the DASS-21 would be a relevant measure of their intended constructs.

2. Methods

2.1. Participants

This study was approved by the University of Sydney’s Human Research Ethics Committee (2012/1631; 2013/352) and informed written consent was obtained directly from each participant prior to inclusion in the study. In total, 153 individuals primarily diagnosed with ASD were consecutively recruited from the headspace and collo- cated Autism Clinic for Translational Research (ACTr) at the Brain and Mind Centre, the University of Sydney between January 2012 and May 2019. Participants presented to the headspace or ACTr for treatment and/or a range of concerns (e.g., assessment of ASD, social skills de- velopment and mental health concerns) and were self-referred or re- ferred by mental health professionals. Clinical diagnoses were con- firmed by research qualified clinicians using the Autism Diagnosis Observation Schedule – 2nd edition (ADOS-2; Lord et al., 2012). For those diagnosed with ASD, participants met clinical cut-off on the ADOS-2, and a clinical interview assessing DSM-V criteria. The Wechsler Test of Adult Reading (WTAR; Wechsler, 2001) was used as an estimate of the pre-morbid intellectual quotient (IQ) for screening the presence of intellectual disability in participants. All participants were excluded from the study if they had an intellectual disability (IQ < 70). Thirty participants did not complete the DASS-21 questionnaire and were excluded. A total of 123 participants met inclusion criteria for this study (Age: M = 23.38, SD = 6.95, range = 16–46). As a part of the standard assessment procedure, clinicians asked their participants to complete the self-report questionnaires without any assistance. Participants could complete the questionnaires at home and send the questionnaires back to clinicians, prior to their first CBT session. 2.2. Measures 2.2.1. Demographic information Demographic information was obtained from the self-report ques- tionnaires, including gender, age (in years), education (in years), and occupational status. 2.2.2. The 21-item depression, anxiety, and stress scales The DASS-21 (Appendix A) is comprised of three self-report scales assessing anxiety (7 items), depression (7 items), and stress (7 items). Each item is scored from 0 (did not apply to me at all) to 3 (applied to me very much or most of the time). The scale score is calculated by summing the scores for the relevant items, ranging from 0 to 21. The summary score is also computed by summing all the item scores, ran- ging from 0 to 63. As the DASS-21 is a short form version of the DASS (42 items), the final score of each scale were multiplied by two, so that they can be compared with the normal DASS scores (Antony et al., 1998; Lovibond and Lovibond, 1995). In this study, the recommended cutoffs provided in the DASS manual were used for defining mild, moderate, severe and extremely severe scores for each DASS-21 scale (Lovibond and Lovibond, 1995). The DASS-21 questionnaire can be downloaded from the DASS website (http://www.psy.unsw.edu.au/ dass/). 2.2.3. Hamilton rating scale for depression The 17-item version of the clinician-administered HAM-D was used to measure the severity of depressive symptoms and administered in accordance with the Structured Interview Guide for the HAM-D. Eight of the HAM-D items are rated on a 5-point scale (0 = absent to 4 = very severe), while the remaining nine items are rated on a 3-point scale (0 = absent to 2 = definite), yielding a total score ranging from 0 (least severe) to 52 (most severe). 2.2.4. Liebowitz social anxiety scale self-report The LSAS-SR is one of the most widely used instruments for the assessment of social fear and social avoidance in 24 potentially anxiety- provoking social situations (13 concerning performance anxiety and 11 concerning social interaction anxiety). The scale contains 24 items that produce separate scores for fear and avoidance of social interaction and performance situations. Each item is rated separately for fear (0 = none to 3 = severe) and avoidance behaviour (0 = never to 3 = usually). The scale score is calculated by adding the scores obtained from each item, with a maximum score of 72 for the social fear and 72 for the social avoidance. These scale scores are combined to provide the overall scores with a maximum of 144 points. Higher LSAS-SR scores indicate greater social anxiety. 2.2.5. The 12-item short form health survey The SF-12 is a 12-item measure of health and well-being, which yields physical component (PCS) and mental component summary (MCS) scales. These scales assume a mean of 50 and a standard de- viation of 10, with low scores indicating poor QoL. The SF-12 was found to be a valid and reliable measurement of QoL in adults with ASD (Khanna et al., 2015). In this study, the PCS and MCS scores were calculated based on the Australian specific scoring algorithm empiri- cally derived from the study conducted by Tucker et al. (2010). 2.3. Statistical analyses All Statistical analyses were performed in SPSS (version 24) except the confirmatory factor analysis (CFA), which was conducted using AMOS (version 24). The internal consistency of the DASS-21 was de- termined using Cronbach's alpha. A value of Cronbach's alpha > 0.7 was considered as adequate and a value > 0.9 as excellent internal consistency (Nunnally, 1978). Item-total correlations were evaluated by Pearson’s product-moment correlation coefficient. Items with very low item-total correlations (< 0.3) were endorsed to be deleted (Nunnally, 1978). Floor and ceiling effects were considered present if more than 15% of the participants reported the lowest or highest pos- sible DASS-21 scores (Terwee et al., 2007). Convergent validity of the DASS-21 was examined based on Spearman's correlations between DASS-21 and instruments similar constructs such as HAM-D and LSAS- SR. Correlation values < 0.20 were considered as very weak, 0.20 to 0.39 as weak, 0.40 to 0.59 as moderate, and 0.6 and above as strong (Swinscow, 1996). Spearman correlations were conducted with Bon- ferroni corrections to limit family-wise error. The significance of the difference between the two correlation coefficients was examined using Fisher's r-to-z transformation. Finally, confirmatory factor analysis was performed to assess several competing models of the DASS-21 and identify the best-fitting model for the DASS-21 in individuals with ASD. The model fits were assessed using the following criteria: 1) Acceptable Fit: standardized root mean square residual (SRMR) < 0.08 and root mean square error of ap- proximation (RMSEA) < 0.08 (Awang, 2012; Hu and Bentler, 1999) and comparative fit index (CFI) > 0.9 (Bentler, 1990); 2) Good fit: SRMR and RMSEA < 0.05, and CFI > 0.95 (Hair et al., 2014; Hu and
Bentler, 1999). Because the chi-square (χ2) is highly influenced by sample size (Bentler, 2007), the χ2 was reported for completeness in
this study. Furthermore, previous research has shown that the RMSEA also tends to overreject models with small sizes (n < 250; Hu and Bentler, 1999). For CFA, multiple imputation by chained equations (MICE; Buuren and Groothuis-Oudshoorn, 2010) was conducted to impute missing values. The models were further evaluated in terms of factor loadings, residuals, and modification indices. According to Hair et al. (2014), items with low standardized factor loadings (< 0.50) on their respective domains were considered for removal. 3. Results 3.1. Summary statistics As shown in Table 1, the mean age of ASD participants was 23.38 (SD = 6.95, range 16–46). Their mean years of education was 12.22 (SD = 1.94, range = 8–17), and the majority of these participants (69.1%) were male. For occupation, 34.7% were employed, and 35.6% were unemployed. The remaining 29.7% were students. The reported means on the DASS-21 scales and total scores were significantly higher than what has been reported in the general population (ps < 0.001; Crawford et al., 2011; Henry and Crawford, 2005). In our ASD cohort, the mean Anxiety score (N = 123, Mean = 15.63 and SD = 9.98 was in the severe range (15 - 19), whereas the mean Depression (N = 123, Mean = 20.61 and SD = 12.66) and Stress (N = 123, Mean = 21.70 and SD = 10.61) scores were in the moderate range (14 - 20 and 19–25, respectively). In comparison to the mean DASS-21 Depression score, on average, the clinician-rated HAM-D score (N = 55, Mean = 7.89 and SD = 6.03) was in the mild range (8 - 16). A previous study investigating social anxiety in adult males with ASD (Spain et al., 2016) reported similar mean LSAS-SR score (N = 50, Mean = 67.0 and SD = 28.5) compared to the mean LSAS-SR score in this study (Mean = 71.79 and SD = 29.1). The mean SF-12 MCS score in this study (Mean = 36.15 and SD = 10.97) was slightly below that reported (N = 291, Mean = 40.77 and SD = 10.36) by Khanna et al. (2015). On the other hand, the reported mean SF-12 PCS score in this study (Mean = 33.96 and SD = 9.47) was noticeably lower than that reported in their study (N = 291, Mean = 49.44 and SD = 10.36). 3.2. Internal consistency, item-total correlations, missingness, ceiling, and floor effects As shown in Table 2, Cronbach's alpha value (α) for the DASS-21 total score was 0.94. Consistent with psychometric theory, scale reli- abilities were slightly lower for DASS-D (α = 0.93), DASS-A (α = 0.84) and DASS-S (α = 0.88). In addition to this, all items exhibited sa- tisfactory item-scale correlations (r = 0.32–0.82) and item-total cor- relations (r = 0.40–0.77). For the DASS-21, Most participants (96.7%) answered all items, 2.4% left one question, and 0.8% left four items. Item 18 “Touchy” had the highest level of missing data (n = 2, 1.6%). 3.4. Confirmatory factor analysis For DASS-21, the following structural models were tested (see Table 4): a) a single-factor model in which all 21 items loading on general distress (Lovibond and Lovibond, 1995); b) a 2-factor model with depression in one factor and anxiety/stress combined in a second factor (Brown et al., 1997); c) Lovibond and Lovibond (1995)’s original 3- factor oblique model, as shown in Fig. 1; d) Henry and Crawford (2005)’s bi-factor (quadripartite) model with a general distress (negative affectivity) factor and depression, anxiety and stress as spe- cific factors; and e) Tully et al. (2009)’s tripartite model in which all items loading on a general negative affect factor and anxiety and de- pression as specific factors. The fit statistics for the DASS-21 showed that the single-factor and 2-factor model had unacceptable fit values. The original 3-factor ob- lique (CFI = 0.90, SRMR = 0.07 and RMSEA = 0.08), bi-factor (CFI = 0.94, SRMR = 0.05 and RMSEA = 0.07) and tripartite models (CFI = 0.93, SRMR = 0.05 and RMSEA = 0.07) showed acceptable fit. As shown in Fig. 1, the original 3-factor oblique model showed accep- table factor loadings from 0.50 to 0.87, except the item 2 “Dryness of mouth” (0.38). The item 2 “Dryness of mouth” was therefore con- sidered for removal. The final modified model also permitted correlated error between the items (with high modification indices > 10) from the same scales (see Fig. 2). The modified DASS-20 model presented im- proved fit for the data (CFI = 0.92; SRMR = 0.06; RMSEA = 0.08).
On the other hand, the factor loadings of both the bi-factor and tripartite model were unsatisfactory (Table 5). In fact, a substantial number of factor loadings were from negative to < 0.4. Therefore, the original 3-factor oblique model was the best-fitting model for the DASS- 21, and the model fit could be further improved with some modifica- tions. 4. Discussion The present study is first to examine the psychometric properties of the DASS-21 in individuals with ASD, where there is a crucial need to identify reliable and valid measures for assessing depression and an- xiety in this cohort. Overall, our results suggest that the DASS-21 may be a useful self-report measure for assessing the core symptoms of common and overlapping mental health constructs of depression, an- xiety, and stress in individuals with ASD and without intellectual dis- ability. However, the “Dryness of mouth” item from the DASS-21 Anxiety scale showed weak performance in our ASD cohort, suggesting further research is warranted in larger populations to determine the utility of the item. A further finding of particular relevance in this study was a discrepancy between clinician and self-report ratings of depres- sion. Clinicians underrated symptoms of depression in comparison to self-report on the DASS-21. The findings of this study are particularly important given some mental health systems rely on the DASS-21 as part of their reporting requirements and, overall, our study supports its use. In our ASD cohort, more than half the participants reported mod- erate to extremely severe levels of (63.4%; DASS-21 Depression > 13), anxiety (65.0%; DASS-21 Anxiety > 9) and stress (53.3%; DASS-21 Stress > 18). These percentages were similar to the prevalence rates of depression and anxiety in individuals with ASD, based on time-con- suming structured or semi-structured interviews (e.g., Hofvander et al., 2009; Lugnegard et al., 2011). This finding highlights the potential utility of the DASS-21 for screening anxiety and depression in this po- pulation.

It is also important to note that the mean DASS-21 Anxiety score of our ASD cohort was in the severe range. The mean DASS-21 Depression
score was also close to the cutoff point for severe, whereas the mean DASS-21 Stress score was close to the cutoff point for mild. The parti- cularly high DASS-21 Anxiety and Depression scores may in part be due to the possible impact of comorbid depression and anxiety. Participants in this study were recruited through a community clinic. They attended for diagnostic, research participation or assistance with mental health concerns. These largely self or other referred attending participants may have more severe and complex comorbid conditions than those recruited from the general population. This finding may, however, also reflect previous population-based studies of ASD that have demon- strated far higher rates of depression and anxiety (Croen et al., 2015; Hudson et al., 2019; Nimmo-Smith et al., 2020). In adults with ASD, depression and anxiety are the most comorbidities with the prevalence rate ranges from 38 to 70% and 50% to 65% (Hofvander et al., 2009; Lugnegard et al., 2011).

Interestingly, clinicians rated depression more frequently as mild, but those with ASD rate it as moderate to severe. The discrepancy be- tween the self-report DASS-21 Depression and clinician-rated HAM-D scores may in part be due to the difference between patient’s and clinician’s perceptions of the depression severity. Given that self-report measures allow patients to describe their own feelings, this measure can provide useful information that is not captured by the clinician-rated measures. It may be that clinicians underrate depression as perceptions of flat affect might overlap with the diagnosis of ASD. Alternatively, people with ASD may over rate emotion intnsity. Further research is required to understand this discrepancy. Therefore, this study suggests self-report and clinician-rated measures of depression should be used together to capture the phenomenon of interest (Kayes and McPherson, 2010). The DASS-21 total and scales scores demonstrated adequate internal consistency. All items also exhibited a satisfactory correlation with their respective domains, indicating that all the DASS- 21 items can discriminate high-scoring individuals from a lower scoring individual. Furthermore, there were no floor and ceiling effects.

Consistent with previous research, the DASS-21 scales were significantly inter-correlated with each other as these constructs share a common factor of general psychological distress or negative affectivity (Henry and Crawford, 2005). Similar to findings for the intercorrela- tions amongst DASS-21 scales, the clinician-rated depression (HAM-D) was strongly correlated with the DASS-21 Anxiety and Stress, and the social anxiety (LSAS-SR) was moderately correlated with the DASS-21 Depression. Convergent validity was demonstrated via significant cor- relations with other measures of depression and anxiety, and significant negative correlations with QoL measures (SF-12). The DASS-21 De- pression scale was strongly correlated with HAM-D, and the DASS-21 Anxiety and Stress scales were moderately correlated with LSAS-SR. As expected, all DASS-21 scales showed significant correlations with physical health- (SF-12 PCS) and mental health-related QoL (SF-12 MCS), although some associations did not remain significant after Bonferroni correction. These findings are similar to what has been re- ported in clinical (Fox et al., 2018) and non-clinical populations (Gloster et al., 2008). Overall, these correlations result indicate ade- quate convergent validity of the DASS-21.

Considering fit indices and factor loadings, the CFA results showed the best fit for the 3-factor oblique model of depression, anxiety and stress as that already validated in other clinical and non-clinical po- pulations (Antony et al., 1998; Clara et al., 2001; Henry and Crawford, 2005; Kyriazos et al., 2018). The 3-factor oblique model produced a better fit than the 1-factor model (in which all items col- lapsed into general distress), suggesting that anxiety and depression have both common and unique features (Watson and Clark, 1984). The 3-factor model also showed a better fit than the 2-factor model, which collapsed Anxiety and Stress factors, indicating that depression, an- xiety, and stress are distinct factors.

The 3-factor oblique model had an acceptable model fit, and the fit of this model was further enhanced through the modifications to the items that do not meet the criteria satisfactorily. The item 2 “Dryness of mouth” showed a substantially poor performance than other items in the best-fitting model. There has been some debate that this item “dryness of mouth” may not represent the anxiety construct (Clara et al., 2001; Le et al., 2017; Parkitny et al., 2012; Shea et al., 2009; Szabo, 2010). In some studies, that this item was loaded pri- marily on the common factor of general distress (Le et al., 2017; Szabo, 2010). Furthermore, the discrimination of mouth dryness be- tween four severity levels might be difficult for respondents (Parkitny et al., 2012). The modified model also permitted correlated error between the items (with high modification indices) from the same scales. In comparison to the factor loadings of the 3-factor oblique model, the bi-factor and tripartite model showed unsatisfactory factor loadings. According to Joshanloo et al. (2017), bi-factor models always tend to support uni-dimensionality, and they criticized higher-order factor structures based only on bi-factor models (Joshanloo and Jovanović, 2017).

Our study has several limitations. First, we acknowledge our results here can only be attributed to those patients with ASD without a sig- nificant intellectual disability as we did not include any participants with an IQ below 70. Second, it is possible that the DASS-21 Anxiety and Stress scales may show a stronger correlation with the measures of general anxiety or tension/irritability than the self-reported measures of social anxiety (LSAS-SR), respectively, warranting further in- vestigation. Third, our study’s sample size for CFA was a relatively small and item level analysis requires a larger sample size before one can be certain of the response of single items. We note, however, the sample size in this study exceeded the minimum required sample size of 105 (5 to 10 subjects per item) based on the recommendations of Hair et al. (2014). They also suggested that the CFA models (which is a part of the structural equation modelling) containing five or fewer constructs, and each construct has more than three items, can be ade- quately estimated with the minimum sample size of 100. A further possible limitation involves the inclusion of participants with relatively poor reading skills (i.e., the participants with estimated borderline IQ scores of 70–79 from the WTAR), which may influence the validity of the questionnaire. Fourth, there has been some debate about whether individuals with ASD may have greater difficulty reporting their own affective states and impairments (Volkmar et al., 2004). While most of our assessments were based on self-report measures, we also note that we applied the HAM-D, which is a clinical interview designed to mea- sure the severity of depression and demonstrated high concurrent va- lidity with depression diagnosis based on DSM-IV (Olden et al., 2009). Lastly, although the DASS-21 Depression scale has shown a strong and positive correlation with the HAM-D, the comparison of the DASS-21 results with the DSM-based clinical diagnosis is needed in future re- search to establish the usefulness of the DASS-21 for the clinical diag- nosis at the item level and scale level.

Despite these limitations, the results of this study indicated that the DASS-21 is a viable self-report screening measure for depression, an- xiety, and stress in individuals with ASD and without intellectual dis- ability. This study also highlights the weak performance of the item “Dryness of mouth” in the ASD cohort and the Agomelatine potential discrepancy in severity between clinician and self-report measures.