Introduction
The rising prevalence of pediatric mental health issues underscores the importance of validated, scalable screening tools [
1,
2]. Specifically, there is a need for structured psychiatric assessments that both can be administered by lay interviewers and are validated by established measures of psychopathology. Considering the high rates of mood and anxiety symptoms (prevalence of 25.1% for anxiety and 11.5% for depressive mood) in youth [
1,
2], this need applies particularly to the domain of emotional problems. The current study evaluates the relationship between parent-, child-, and clinician-rated scales of internalizing symptoms and the Development and Well-Being Assessment (DAWBA), a widely used structured psychiatric interview [
3].
Prior attempts to relate lay interviews to clinical assessments generate relatively poor psychometrics for emotional problems. One set of studies utilized the Diagnostic Interview Schedule for Children (DISC) [
4], which exhibits acceptable reliability and validity combining youth and parent reports for externalizing disorders (test-retest kappa = 0.48–0.66, validity with retest kappa = 0.49–0.70, concurrent validity kappa = 0.65–0.80) [
5]. However, relatively poor psychometrics exist for the DISC’s internalizing modules (test-retest kappa = 0.35–0.52; validity with retest kappa = 0.32–0.53; concurrent validity kappa = 0.37–0.57) [
5] in which the DISC demonstrates poor to fair test-retest reliability and yields moderate concurrent validity across both anxiety and depression. Additional studies similarly report weak associations between the DISC and established self-report anxiety measures [
6]. Other interviews and more recent versions of the DISC may generate superior psychometrics [
7]; however, few studies have evaluated concordance between rating scales and structured psychiatric interviews in samples where the need may be greatest – that is, in youth seeking treatment for emotional problems.
The Screen for Child Anxiety Related Disorders (SCARED) is one widely used self- and parent-reported measure of anxiety spanning multiple diagnostic domains [
8]. Likewise, the Mood and Feelings Questionnaire (short version; MFQ) is a self-report questionnaire assessing depressive symptoms [
9]. The Pediatric Anxiety Rating Scale (PARS) is an established clinician-rated measure of anxiety severity and impairment [
10]. While these assessments are useful for measuring internalizing psychopathology, the DAWBA offers a brief, widely distributable cross-disorder evaluation of youth psychopathology, an adaptive question structure, and more comprehensive integration of child and parent response data [
3,
7]. The DAWBA has also been validated across several psychiatric disorders and generates independent ratings of mood and anxiety symptoms that may relate to other scales of internalizing psychopathology [
3,
11‐
13]. Given the relevance and challenge of integrating reports from multiple sources [
6], in the current study, we evaluate relations between these measures and band scores generated by the DAWBA that integrate response data from both caregivers and youth [
14].
The DAWBA evaluates the probability of 19 psychiatric illnesses via an adaptive online questionnaire in which respondents are presented a unique series of questions for each disorder dependent upon previous responses [
3]. The potential value of the DAWBA lies in both its scalability as a short computerized assessment and its integration of data across multiple respondents to predict risk for diagnosis [
7,
11]. If the DAWBA’s psychometrics translate to populations of youth seeking treatment for emotional problems, the interview could enhance accessibility of psychiatric screening in multiple settings where resources are limited [
15,
16]. In addition to screening, we assessed the DAWBA’s ability to predict treatment response. Previous clinical trials have demonstrated the sensitivity of the SCARED, MFQ, and PARS to symptom improvement in efficacious treatments [
17‐
19]. Moreover, a recent review reported that baseline severity constitutes one of the strongest predictors of clinical trajectories [
20]. In a study that combined measures of internalizing symptoms at baseline, findings indicated that low anxiety severity predicted better treatment outcomes after 12 weeks, as indexed by the PARS [
21]. The SCARED has also been used to predict treatment response and remission across both parent and child reports [
22]. Therefore, we aimed to determine whether the band scores generated by the DAWBA at baseline predict therapeutic response, as indexed by these established measures across treatment. If predictive of symptom trajectories in treatment-seeking youth, the DAWBA band scores might reasonably offer clinical utility beyond a baseline diagnostic assessment.
The current study tests the hypothesis that, first, DAWBA band scores are associated with self- and parent-reported measures of anxiety and depression and clinician-reported measures of anxiety. Second, we hypothesize that the DAWBA anxiety band scores collected at baseline will predict the change in SCARED and PARS scores over the course of treatment. Due to the lack of availability of MFQ data across treatment in the depression sample, we did not examine the DAWBA MDD band score in relation to depression symptom trajectories.
Discussion
This study examined associations between the DAWBA and established measures of anxiety and depression at baseline and across treatment. Three main findings arose from the study. First, the DAWBA band scores significantly predicted both self- and parent-reported measures (SCARED-CP, MFQ-CP) in the anxiety and depression samples. Second, the DAWBA anxiety band scores did not predict the change in SCARED-CP across treatment. Third, the DAWBA band scores significantly predicted the PARS, a clinician-rated measure of anxiety; however, we did not observe any association between the DAWBA band scores and the change in PARS score across treatment.
Although both the SCARED-CP and MFQ-CP were significantly associated with the DAWBA band scores, the MFQ-CP exhibited a strong association with the DAWBA MDD band score, while the association between the SCARED-CP and the DAWBA anxiety band scores was weak to moderate. This could reflect the high degree of heterogeneity in the anxiety sample, a transdiagnostic grouping of three distinct anxiety disorders, which were not equally represented in our sample. Evaluating anxiety symptoms within specific diagnostic categories likely enables more robust prediction than clustering across disorders [
29]. Furthermore, previous studies find that incorporating both self-report and clinician interviews better capture the heterogeneity of internalizing psychopathology and provide more accurate symptom assessment and prediction [
30,
31].
As with the SCARED-CP, all DAWBA anxiety band scores significantly predicted the PARS. Relative to other band scores, the relationship between the PARS and the Sep AD band score appeared notably weaker. This is potentially reflective of the low prevalence of Sep AD within our sample. Other studies have also found discrepancies between established self-rated measures and clinician interviews [
32,
33]. One study examining the self-reported and clinician-rated versions of the same instrument across novel interventions for depression suggested that each rater contributes distinct and important information for predicting treatment outcomes [
31].
In our exploratory analysis, we tested whether the DAWBA band scores, collected at baseline, predicted anxious participants’ response to treatment, as measured by the difference score between participants’ pre- and post-treatment PARS and SCARED-CP. For both measures, none of the three DAWBA band scores (GAD, Sep AD, or Social AD) significantly predicted the change in anxiety across treatment. It is possible that the small sample size in this subset of data constrained our power to detect a predictive effect of the DAWBA band scores on treatment progression. Alternatively, given the DAWBA’s design as a scalable, computer-based diagnostic screening tool administrable by non-expert interviewers, the DAWBA might not possess the granular sensitivity to reliably detect small changes in symptom presentation over relatively short time scales. As established clinical measures collected at baseline are often predictive of symptom progression across treatment when assessed via repeated measurement [
17‐
19], we tested whether PARS and SCARED-CP at baseline predicted the PARS and SCARED-CP difference scores, respectively. As with the DAWBA, neither baseline measurement predicted the change in symptoms across treatment, suggesting overarching sample size constraints and perhaps an inherent complexity in anxiety psychopathology, which may limit accurate prediction over extended timescales.
Although the DAWBA takes advantage of multiple reporters – both parents and children – to make predictions about psychopathology severity via computer algorithms, it remains unclear whether the optimal weighting of parent and child data differs between concurrent prediction of diagnostic risk and prospective prediction of treatment response. Differential weighting of parent and child data may be necessary, especially given the discrepant nature by which parents and children often perceive treatment progression and efficacy [
34‐
36], thereby potentially reducing the DAWBA’s sensitivity to temporal changes in symptomology. Future studies would benefit from exploring the parent- and child-specific DAWBA band scores (rather than the combined, as used in the current study) in conjunction with disorder-specific subscales of established measures to independently predict longitudinal symptom progression across treatment.
To our knowledge, this study is one of the first to examine the DAWBA in relation to established internalizing measures in a sample of treatment-seeking youth both at baseline and across CBT. Importantly, clinicians in the anxiety cohort were masked to the DAWBA when assessing symptom severity. Our results suggest that when collected at baseline, DAWBA band scores are associated with the SCARED-CP, MFQ-CP, and PARS; however, this baseline measurement was not predictive of symptomatology following treatment within the anxiety cohort. Therefore, our findings indicate that DAWBA band scores are relatively predictive of current symptom presentation as per established measures of youth depression and anxiety. However, several limitations are worth noting. First, as our sample comprised only treatment-seeking youth who met diagnostic criteria for a DSM-5 disorder, the associations observed may be reflective of individuals who present with more severe psychopathology and may be less generalizable to subclinical populations. Second, we did not include a clinician-rated measure of depression given sample size constraints of such data across treatment. Third, across all analyses in the anxiety cohort, subjects were aggregated into a transdiagnostic sample spanning three distinct DSM-5 anxiety disorders.
Future studies would benefit from exploring disorder-specific associations between the DAWBA’s three anxiety band scores and subscales of established measures within diagnostically homogenous groups. More precise approaches such as these would allow for more rigorous evaluation of the DAWBA’s anxiety band scores specific to particular symptom clusters. Additionally, assessing the DAWBA’s parent- and child-generated band scores individually in relation to established clinical measures remains important for refining the way in which semi-structured lay interviews differentially weight parents’ and children’s response data to optimize predictions. Finally, future studies should explicitly evaluate the DAWBA band scores with repeated administrations across treatment to more thoroughly assess sensitivity to treatment response.
In conclusion, this is one of the first studies to examine the DAWBA in relation to validated and widely used measures of internalizing psychopathology in a sample of treatment-seeking youth. Principally, our findings suggest that the DAWBA may be an effective tool for screening youth anxiety and depression at relatively transient timescales in relation to established clinical and self-reported measures. However, the DAWBA appeared to be notably limited in predicting anxious participants’ symptom progression across treatment. Despite this, our results suggest some potential for clinical utility in identifying internalizing symptomology among treatment-seeking youth; future studies should further evaluate the DAWBA’s validity and reliability as a scalable tool for mental health screening and assessment.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.