Notwithstanding that the prevalence of psychiatric disorders among detained boys is high (e.g., [1
]), juvenile detention facilities all too often lack budgets, time, and/or qualified personnel to perform in-depth psychiatric assessments. Juvenile detention staff, therefore, may have a special interest in mental health screening tools that can provide concrete directions on how to plan further narrowly-focused psychiatric assessments (e.g., only ascertain the presence of major depression disorder if the youth is high on a scale that screens for depressive feelings). Most screening tools, however, are not designed to map psychiatric disorders as defined by classification systems such as the DSM, and include scales that cover many symptoms, moods, and thoughts that are not specific to any particular disorder. As such, it may come as no surprise that the accuracy of these scales to identify detained youth with a specific disorder is most often moderate at best (e.g., [2
]). An exception may be the DSM-oriented (DSM) scales of the Youth Self-Report (YSR; [3
]). The YSR was designed to be useful in many contexts, including juvenile justice settings [4
], and its DSM scales comprise items identified as being very consistent with specific DSM-IV disorders [4
]. Even though these DSM scales were not intended to be a perfect equivalent of the disorder(s) under consideration, they still may provide concrete directions for further assessment.
1.1. Empirical Studies on the YSR DSM Scales
Very few studies tested the relations between the DSM scales and their corresponding psychiatric disorder(s). The DSM Affective Problems scale seems to be a highly accurate screener for major depression disorder (MDD) in clinic-referred youth [5
], but not among general population samples [7
]. Among detained youth, the accuracy of this scale as a screener for MDD ranged from poor [11
] to moderate [12
], making it impossible to draw any conclusions. Findings with regard to the DSM Anxiety Problems scale were far more mixed, with some studies showing that this scale was a moderately to highly accurate screener for anxiety disorder in detained ([12
]; but see: [11
]) and clinic-referred adolescents ([13
]; but see: [6
]), though not in clinic-referred children [13
]. The DSM Attention Deficit/Hyperactivity and Oppositional Defiant Problems scales were found to be an accurate screener for attention deficit/hyperactivity disorder and for oppositional defiant disorder, respectively [12
]. The DSM Conduct Problems scale, finally, accurately screened for a conduct disorder diagnosis in detained youth [12
1.2. Contribution of This Study
The present study will substantially contribute to the literature on the DSM scales in at least four ways. First, the previous section clearly showed that studies on the screening accuracy of DSM scales other than the Affective and Anxiety Problems scales are quite rare. The present study will fill this void.
Second, studies on the DSM scales typically provided confidentiality of the information and anonymity to its participants. Consequently, it is unsure to what extent results from these studies have field utility and, thus, are informative for clinicians who are working with detained youth. When being asked to complete the YSR as part of a clinical protocol, detained youth, indeed, may be reluctant to reveal information (e.g., tendency to act out aggressively) that is unknown to others and can be used against them (e.g., in court). To address this issue, the present study relied on data that were gathered during a clinical protocol, thus, outside of a research context.
Third, a large number of detained youth score high on various scales of a mental health screening tool, and meet diagnostic criteria for two or more psychiatric disorders [14
]. Consequently, testing relations between a DSM scale and its corresponding disorder might be more relevant from a psychometric perspective (i.e., construct validity) than from a clinical perspective. Indeed, it can be argued that may be the best a mental health screening tools can offer when working with detained youth is to do an initial sorting or triage of detained youth into one group that includes almost all of the youth with psychiatric disorders (so they can further be assessed) and one group that has as few youth as possible with psychiatric disorders (and that need no further assessment) [15
]. Therefore, this study also examined how the DSM scales serve as a triage tool.
Fourth, studies on the DSM scales did not test if results differed across youth from various ethnic origins. This is unfortunate (a) because detained ethnic minority youth are overrepresented in detention facilities, and often report less mental health problems than detained ethnic majority youth (e.g., [16
]); and (b) because the screening accuracy of mental health screening tools can vary across ethnic groups (e.g., [14
]). To fill this void, the current study differentiated between youth from various ethnic origins.
Forensic research has resulted in an increased awareness among clinicians and policy makers of the mental health needs of detained adolescents. Although there are many methods for assessing mental health problems in juvenile justice settings [15
], most methods require more time and staff expertise than most youth detention centers can afford. Therefore, brief self-report questionnaires are appealing for youth detention centers, particularly because they can help clinicians to classify incoming youths according to their level of urgency (e.g., [18
]). Unfortunately, the overwhelming majority of forensic studies that tested the psychometric properties or ‘clinical’ usefulness of these self-report tools was conducted as part of a research project where the participants were assured that the data would be anonymous and not impact their cases. Subsequently, there is a need for on-going research into the reliability and validity of self-report tools when they are taken out of the lab and into legal settings where the information may bring actual consequences for the informant. The present study filled a gap in the literature by being the first field study that scrutinized the psychometric properties and clinical usefulness of the DSM scales among detained boys.
This study showed that the internal consistency of the DSM scale scores as indicated by α ranged from 0.57 (Anxiety Problems) to 0.81 (Conduct Problems). These αs were identical to the range of αs reported in a previous study among detained boys [12
], and in line with a study among clinic-referred adolescents [27
]. According to the two other indices (that are less sensitive to the number of items in a scale than α), all six DSM scales scores were at least adequately internally consistent. The finding that the DSM scale scores, overall, are also internally consistent in the three ethnic minority groups bears substantial clinical relevance, as detained youths are most often not from the major ethnicity group of the country where they live and are being detained [28
Although caution is warranted (see Section 2.4
), Moroccan boys had significantly lower DSM scale scores than boys from Dutch, Antillean/Surinamese, and Mixed ethnicity. This finding dovetails well with recent studies showing that Moroccan detained boys reported lower levels of mental health problems than boys from Dutch and other ethnic origins as measured by various tools, including the YSR [30
], the Strengths and Difficulties Questionnaire [16
], and the MAYSI-2 [14
]. Detained Moroccan boys, thus, seem to systematically report fewer mental health problems than boys from other ethnic origin, regardless of the measure being used. Future studies are warranted to test if different item functioning could explain these cross-ethnic differences in screening scale scores. Importantly, these lower mean scores in detained Moroccan boys do not imply that these youngsters do not have mental health needs that must be addressed.
This study also showed that the DSM scales are related to their corresponding disorder. The results showed that boys who meet criteria for a specific psychiatric disorder (e.g., conduct disorder) obtain higher scores on the corresponding DSM scale (i.e., Conduct Problems) than boys who were without that disorder. Likewise, the DSM scales were moderately to highly accurate in predicting their corresponding disorder, though not always significantly. These AUCs were somewhat higher than those reported in a prior study in detained boys on the DSM scales [12
]. Specifically, in this prior study, the AUC for the DSM Attention Deficit/Hyperactivity Problems scale in predicting attention deficit/hyperactivity disorder in the total sample was 0.77 (vs. 0.85 in this study), for the DSM Oppositional Defiant Problems scale in predicting oppositional defiant disorder 0.78 (vs. 0.81), for the DSM Conduct Problems scale in predicting conduct disorder 0.76 (vs. 0.85), for the DSM Affective Problems scale in predicting Any Affective Disorder 0.65 (vs. 0.78), and for the DSM Anxiety Problems scale in predicting Any Anxiety Disorder 0.73 (vs. 0.76). The accuracy of the DSM scales in real-world settings, thus, are at least as good as those stemming from a study that guaranteed confidentiality to its participants [12
]. Overall, these findings suggest that the DSM scales do relatively well in identifying youths with specific psychiatric disorders.
However, the clinical usefulness of the DSM scales for identifying detained boys who require further narrowly-focused assessments can be questioned, though, for several reasons. First, the AUCs for dichotomized DSM scales in the total sample were generally below 0.70, which is unfortunate because clinicians often find cut-points more appealing and easier to use than thinking in dimensional terms [31
]. Second, the high comorbidity rate in detained boys may limit the usefulness in determining what kind of narrowly focussed assessment is needed. Indeed, a boy with comorbid conduct disorder and anxiety disorders may have DSM Conduct and Affective Problems scale scores that fall in the borderline range. Yet, if a clinician would rely on the results being displayed in Table 4
, he or she would only refer the boy for further assessment to ensure whether he meets criteria for CD. This would be unfortunate, because comorbidity of externalizing and internalizing disorders increases the risk of poor outcomes, including suicidal behavior [32
], and implies that interventions are likely to be insufficient when not adequately tailored to their complex needs [1
The present study, finally, examined if DSM scales can be useful for triage purposes. Although the positive and negative predictive values may be more appealing for clinicians, these values can be affected by prevalence rates of the disorders [34
], and may therefore difficult to generalize to other samples. The sensitivity and specificity indices reported in Table 5
do not run this danger, and, for example, showed that 98% of the 209 detained boys who were without any disorder were not in the borderline range on at least two DSM scales. This means that only a small number of youths would have been referred for further comprehensive psychiatric evaluation while this was not warranted. Unfortunately, the sensitivity of the multiple decision rules used in the present study was poor, suggesting that a high percentage of the boys with at least one psychiatric disorder would have been missed if these decision rules were used for triage purposes. Of note, a prior study among detained boys showed that an alternative mental health screening tool that was completed during a clinical protocol served relatively better as a triage tool [2
]. Clearly, further research is warranted to elucidate how well various mental health screenings can be used for triage purposes.
The findings of this study must be considered in the context of several limitations. First, the YSR and diagnostic interviews used different time frames, a difference that may partially explain why some DSM scales did not accurately predict the presence of a disorder (see also: [8
]). Second, the YSR has no DSM scale on substance use that may help to identify boys with one of the most prevalent disorders among detained adolescents, being substance use disorder (e.g., [1
]). Third, the diagnostic interview to assess the externalizing disorders (the DISC-IV) was different from the diagnostic interview to assess internalizing disorders (the DAWBA). Given evidence that diagnoses resulting from these measures are not equivalent [36
], future research on the topic should use at least one diagnostic interview to assess all relevant disorders of interest. Fourth, even though parents (and teachers) of detained youths are often difficult to locate or unwilling or unable to provide reliable information [37
], the sole reliance on self-report still can be considered to be a limitation of the present study. Fifth, only detained males were included in this study, implicating that this is the only population that an inference can be drawn upon. Sixth, because of sample size considerations, Antillean and Surinamese youths were merged together in one group, whilst youth from various other origins were merged together in the Mixed ethnicity group. Although this approach is in line with previous papers, it may have obscured differences regarding the performance of the DSM scales between these ethnic minority groups. Seventh, due to power issues, the AUCs for dichotomized DSM scale scores presented for each of the four ethnic groups are exploratory at best. Finally, graduate students and test assistants who were present during the YSR administration did not calculate the YSR scores (these were computer generated) and were not necessarily the same ones who interviewed the youth with the DISC or DAWBA. Although this occasionally was the case, it is highly unlikely that they were aware of the youths’ YSR DSM scale score, and even if this was the case this would not have affected the DISC and DAWBA diagnoses for the simple reason that YSR information is not considered in the DISC and DAWBA algorithms. However, an anonymous reviewer argued that the index tests (i.e., YSR DSM scales) and the reference standards (DISC and DAWBA diagnoses) were, strictly speaking, not 100% blinded, and that this must be mentioned as a limitation.