Background
Based on national studies in several countries, youth retained in juvenile justice facilities display high levels of mental health problems, often so severe that they meet criteria for at least one psychiatric disorder (for a review see: [
1,
2]). In the U.S. [
3,
4] and the Netherlands [
5], standardized mental health screening procedures have become routine upon entry into juvenile justice programs to determine the need for emergency mental health services and for additional comprehensive assessment. The present study explored whether brief mental health screening tools, when used shortly after a youth’s entry into detention settings, might be of value for alerting staff to a youth’s potential for future violent offending, thus suggesting the need for more definitive evaluation for risk of harm.
Mental health screening tools, of course, are not developed for that purpose. They are designed to identify youth whose mental health symptoms suggest the need for further assessment to determine need for mental health services [
6]. But if these tools generate mental health screening scores that are related to future violent behavior, this could be of value. Routine evaluation for risk of aggression is not standard practice immediately up a youth’s entry to detention centers, which might be unfortunate since the juvenile justice system has not only an obligation to meet the mental health needs of youth in its custody, but also to protect other youth, detention staff, and the community from harm.
Theory and research on the general relation of mental disorders and violent offending among youth offer mixed expectations regarding a mental health screening instrument’s potential capacity to predict one from the other. Some of the common risk factors for youth offending (e.g., irritability, impulsiveness, substance use) are also symptoms of disorders of youth (e.g., related to depression, traumatic stress, attention deficit disorder, or substance use disorders). Consistent with this, some studies have found that symptoms of psychiatric disorders co-vary with reliable risk predictors of violence (e.g., [
7,
8]). Other studies have found a small to moderate positive association between psychiatric disorder and future violent offending, although with much inconsistency in the specific disorder or disorder categories that were related to future violence (e.g., [
9‐
11]). Therefore, one might expect to find at least modest relations with future violence because some scales of mental health screening tools include items referring to alcohol and drug use, impulsivity or irritability or anger. This would not suggest that mental health screening tools can serve as strong predictors of violence for judicial decision-making purposes. If modest relations were found, the value would be in the tools’ ability to
alert detention staff to engage in further in-depth violence risk assessment to determine whether the youth offers a prospect of danger to staff, other youth in detention or, if released, to others in the community.
However, only a few studies have examined this relationship among criminal justice-involved youth using screening tools. For example, using the Massachusetts Youth Screening Instrument-Second Version (MAYSI-2; [
12]), anger and thought disturbance were related to later aggression during detention [
13,
14], whereas alcohol/drug use and anger were not predictive of violence after being released [
15,
16]. Using the Strengths and Difficulties Questionnaire (SDQ; [
17]), others found that mental health symptoms (e.g. emotion problems and hyperactivity) were not related to violent recidivism [
18]. Unfortunately, firm conclusions are precluded because the studies differed greatly in the variety of mental health symptoms that were considered (e.g. the aforementioned MAYSI-2 studies merely used one or two out of the six clinical scales), the outcome of interest (violence during or after detention), and the control variables included in the analyses. To better inform the science and practice of forensic mental health assessments [
19], the present study examined the relation of mental health screening data that were gathered in the context of a clinical protocol for all youth entering two all-male youth detention centers in the Netherlands. The data base included substantial numbers of detained youth from three ethnic origins (Dutch, Moroccan and Surinamese/Antillean). We examined the relation of mental health symptoms and future violence in these three groups (and in a fourth group of “other” ethnicity) separately, for four reasons. First, past studies indicated differences between various ethnic groups in levels of mental health problems (e.g., [
5,
20]) and recidivism (e.g., [
21]). Second, people of different ethnic origins may respond to mental health screening and assessment tools differently because of variations in openness to acknowledging symptoms (e.g., [
22]). Third, prior work suggested that the relation between mental health problems and future criminality differ across ethnic groups [
18,
23]. Fourth, notwithstanding that two of the ethnic groups (i.e., Dutch and Surinamese/Antillean) were quite specific to juvenile justice settings in the country in which the data were obtained, examination of ethnic differences was expected to contribute more generally to the literature on the relevance of ethnicity in mental health screening and violent risk assessment [
24].
Specifically, the present study used two mental health screening tools (one supplementing the other) to explore whether their scores were related to future arrests for violent behavior. We hypothesized that some symptoms identified on the screening tools would be related modestly to future arrests for violent behavior, but that those relations would vary (in type of symptoms and strength of the relations) for different ethnic groups. Our efforts were exploratory in the sense that we did not form hypotheses regarding specific symptoms or specific ethnic differences.
Discussion
This study explored whether brief mental health screening tools, when used in youth detention settings, might be of value for staff to identify detained boys at risk for future violence, thus suggesting the need for more definitive evaluation for risk of harm. We found a few significant relationships between MAYSI-2/SDQ scales and future violent arrests, and some were consistent with various past theoretical speculations or studies. For example, the negative relation between Thought Disturbances and future violent arrests in Dutch boys is consistent with prior work on the link between psychotic-like symptoms and future violence arrests among criminal justice-involved individuals (e.g., [
18]). Also, both the positive (Dutch boys) and negative (Surinamese/Antillean boys) prospective relation between Depressed-Anxious and future violent arrest are consistent with theoretical notions that (i) depression in boys is often expressed by aggressive behaviors, which may lead to increased interpersonal conflicts and subsequently increase the risk of contact with the juvenile justice system [
47‐
49], and (ii) depressive feelings, anxiousness and nervousness may protect against future violence because of apathy, lower energy levels and avoiding situations that cause tension [
50,
51].
The most appropriate interpretation of our findings, though, looks to the small number of relations and their small effect sizes. In this light, our results suggest little likelihood that screening for mental health problems in boys who are detained in the Netherlands offers any potential for identifying youth at risk for future violent arrests. Prior work with the SDQ in the Netherlands [
18] and the MAYSI-2 in the U.S. (e.g., [
15]) also did not reveal any consistent relation with officially registered future violent crimes after release to the community, suggesting that our findings are not sample- and country-specific. Possibly the strongest message is that when significant relations between mental health problems and future violence were found, they were almost entirely dissimilar across the four ethnic groups, even to the extent of finding opposite relations for boys in different ethnic groups. This is consistent with some prior work [
18,
23] suggesting that ethnic differences in the relation between mental health problems and future criminality must be considered in future studies.
Strengths of this study include the relatively large number of boys from various ethnic origins who completed well-validated mental health screening tools as part of a clinical protocol, thereby increasing the ecological validity of the findings, and testing the prospective relation between MAYSI-2 and SDQ scores and officially registered future violence using both variable- oriented (Poisson regression) and person-oriented (latent profile analysis) statistical approaches.
Our findings must be interpreted in the context of several limitations. First, both of the tools we used employ youth self-report, and perhaps data from other sources would have found more meaningful relationships. But our purpose was to test the value of data that typically are available at intake to detention centers, and few detention centers have anything other than youths’ self-report during the first few hours or days of their detention. Second, we did not consider institutional misconduct and therefore cannot exclude the possibility that mental health problems, such as thought disturbance and anger-irritability, might predict violence during detention, as was found by others [
13,
14,
52]. Screening tools are influenced not only by enduring traits but also by immediate emotional states, and the latter may be more closely related to immediate (in-custody) aggression than to arrests for violence in the distant future (after release). Third, mental health problems were merely assessed shortly after detention entry. It cannot be excluded that the level of mental health problems decreased during detention, for example, because detention staff adequately responded to their mental health problems. Future research, therefore, is warranted to scrutinize if stability and change of mental health problems are related to future violence. Fourth, it must be acknowledged that prior work demonstrated cross-cultural measurement non-invariance of the SDQ self-report version, suggesting that this tool is not suitable for use in cross-cultural comparisons [
53]. Since the SDQ has rarely been used in detained adolescents, future factor analytical studies in these youths on the SDQ self-report version are warranted [
18]. Fifth, we used official records of past and future arrests for violent offenses, and sometimes youths’ violent behaviors are more extensive than arrest records indicate. This implies that we might have underestimated true violent offending.
The findings in this study have two main implications. First, they suggest that further research explorations of the ability of mental health screening tools to identify youths with future violent tendencies probably will be of little value. Second, we suspect that detention personnel who use mental health screening tools at detention intake already assume that certain scales, such as the MAYSI-2 Angry-Irritable or the SDQ Conduct Problems scales, suggest a heightened likelihood of future aggression. This study discourages detention personnel from making these presumptions, although the results do not rule out the possibility (in light of other past research) of their value for alerting staff to aggressive behavior during the youth’s stay in detention.
Authors’ contributions
OFC requested and analyzed the data. OFC and TG drafted the manuscript together. Both authors read and approved the final manuscript.