The missing link(age): Multilevel contributors to service uptake failure in youths on community justice supervision

Abstract

Objectives:

Youth in the juvenile justice system often do not access needed behavioral health services. We seek to determine, based on our model of the Behavioral Health Services Cascade, the prevalence of substance use Screening, Identification of Need, and Referral to and Initiation of behavioral health treatment in youth undergoing intake, and at what steps their access is most challenged. Characteristics associated with identification of needs and linkage to community services are also examined.

Methods:

Data were drawn from administrative records of 33 community justice agencies in 7 states, participating in NIDA’s JJ-TRIALS initiative (N=8307 youth). We examined contributions of youth, staff, agency and county characteristics to identification of behavioral health needs and linkage to community services.

Results:

Approximately 70% of youths were screened, and more than half were found to be in need of treatment. Among those in need of services, only about a fifth were referred to treatment, yet among those referred, 68% initiated treatment. Overall, fewer than 15% of identified youth initiated services. Multivariable multilevel regression analyses identified an array of contributors to service-related outcomes, with youth’s level of supervision among the strongest predictors.

Conclusion:

Community justice agencies appear to follow an approach that focuses identification and linkage practices on youth concerns other than behavioral health needs, although such needs contribute to re-offending. Implications for practice in behavioral health agencies are discussed. Local agencies should coordinate efforts to support communication in the referral and cross-system linkage process.

INTRODUCTION

Yearly, 17 million US youths are referred to juvenile justice systems.¹ They have significant behavioral health concerns, involving troublesome peer and family relationships,² adverse childhood experiences,³ and mental health problems^{4, 5} compared to the general youth population. Many report frequent use of alcohol and marijuana^6–8 and prescription drug misuse,⁹ with a 34% meeting criteria for Substance Use Disorder, across a range of justice settings.¹⁰ Substance Use Disordered juveniles are nine times more likely to be involved in the justice system.¹¹ Intervening early can break the cycle of continued justice involvement since treatment for behavioral health disorders lowers recidivism risk.^{12, 13} Screening and assessment of service needs are recommended in judicial and justice processing¹⁴ to facilitate service receipt.

Most youths (75%−95%) in contact with the juvenile justice system are handled by community supervision agencies at some point in case processing.^{9, 15} Approximately 85% of justice agencies do not directly provide behavioral health services.⁹ While justice agencies commonly screen youths for behavioral health needs and delinquency risk, their mission does not include conducting clinical assessments to disentangle factors promoting delinquency.

Juvenile justice administrators note that fragmentation in local service delivery systems creates barriers to obtaining behavioral health services⁹ and report difficulty working collaboratively with local service providers.^{16, 17} A large majority of justice-involved youths do not access services, even after identification of need.¹⁸ Failure to access those services exacerbates continued substance use and mental health concerns, contributing to further involvement in the juvenile or adult justice systems.^{10, 19, 20} The high screening rate coupled with a low treatment initiation rate warrants closer inspection of which youths are referred to services.

To navigate between justice and behavioral health service sectors, young people must pass through a series of steps, from screening, on to referral, and to service initiation, engagement, and continuing care, described as the Behavioral Health Services Cascade (“Cascade”).¹⁴ We lack understanding of the degree to which this model accurately captures the justice system’s approach to young persons’ behavioral health needs, or the point(s) where juveniles may depart from the Cascade. Without these data, we cannot implement targeted system- or policy-level changes that address service gaps and missed treatment opportunities.

In order to better understand the interplay of multilevel factors on Cascade steps, we measure contributions of youth, staff, agency and county characteristics to these events. We do so in 33 discrete justice settings, necessitating a multilevel approach to understand these complex service systems; youth are supervised by justice staff whose agencies operate within a system that includes behavioral health providers, all of which are further influenced by their larger community context.²¹

METHODS

JJ-TRIALS Research Cooperative.

The NIDA-funded Juvenile Justice-Translational Research on Interventions for Adolescents in the Legal System (JJ-TRIALS) Cooperative Agreement includes six Research Centers, each working with four to six local justice agencies providing community justice supervision (e.g., probation) in Florida, Georgia, Kentucky, Mississippi, New York, Pennsylvania, and Texas. Its primary aim was to test the effectiveness of a locally customized intervention to improve delivery of services addressing substance use needs for justice-involved youth by working with justice agencies and their community-based behavioral health partners. We examined youth and staff in 33 county-level sites during the pre-study phase, before the experiment on service delivery improvement. Research Centers received approval from their respective IRBs.²¹ Justice staff provided informed consent for surveys, and a waiver of consent was granted to review de-identified youth records. By design, justice sites (27 intake settings, 6 youth courts) provided community-based services to juveniles whose behavioral health needs were typically addressed by other local providers. Sites varied in sociodemographic characteristics and service availability.

Procedures and Measures

Youth case records.

Data from all new intakes (N=8,307) between March 2014 and August 2015 were captured in this baseline sample²², preceding our offering implementation strategies. The baseline period covered 9–18 months (6 of 7 states contributed 11+ months of data) wherein youths underwent intake evaluations and case disposition and were then either diverted to other services or were placed on community supervision. Justice staff routinely enter information on youths and case management events into agency information systems; through study agreements, information from these systems was made available to researchers, generally by automatic record abstraction; details are available elsewhere.²²

Cascade outcomes were derived from official case records reflecting whether a juvenile: (a) was screened for substance use needs; (b) was determined to need substance use services based on positive screening or urine test result, clinical assessment, alcohol or other drug charges, or court-ordered substance use treatment; (c) was referred to a local service provider by justice staff; (d) initiated those services; (e) engaged in treatment (for at least six weeks); and (f) received care for 90+ days.^{14, 21} Often both substance and mental health services were provided by a single agency, or substance use services were first accessed via mental health service providers; we generically describe these services (and the corresponding service referrals and initiations), as behavioral health services. Screenings measured substance use and broader behavioral health needs; 69% of sites (22 out of 32) relied on evidence-based instruments. Agency records did not consistently document Cascade steps beyond initiation: 23 (69%) sites had information on treatment initiation for more than 95% of cases, only 20 (60%) had data on treatment engagement, and 19 (58%) on continuing care. Accordingly, we restrict reporting to all 33 sites through referral and 23 sites through initiation. Since swift identification and service linkage are key to good clinical practice,²³ we examined if: youths were screened within 30 days of intake and if youths needing services were referred within 30 days of screening.

Youth demographic and offense characteristics were extracted from justice case management records and included: gender, race (white/non-white), age, whether current offenses(s) included alcohol- or other drug- related charge(s), and type of supervision. Higher supervision level included ongoing formal oversight by probation, parole, or Juvenile Drug Court authorities; lower level reflected diversion to community programs or placement on informal community supervision.²⁴ (The dichotomy “formal/informal”, which is frequently used in criminal justice research, does not precisely capture what is meant by “high and low supervision” in this study.) Decisions regarding the needed level of supervision generally reflect offense severity or chronicity.²⁴

Justice agency staff and leadership surveys.

All staff carrying caseloads and their supervisors (n=441, 70.5% response rate) were invited to complete surveys,²⁵ for which informed consent was obtained onsite. Agency leaders completed questionnaires regarding site practices for substance use services and shared practices/collaboration with local providers. Survey responses were aggregated to calculate agency-level measures. Years of experience and caseload size were averaged across respondents within sites. Nine perceived competency items were rated on 5-point Likert scales; scale scores were multiplied by 10 for ease of interpretation, and responses averaged (α=0.85), assessing how well respondents could identify youths’ behavioral health concerns and link them to providers.²⁶ Indicators of organizational characteristics included quality of intra-agency communication (6 items, α=0.86, and job-related stress (4 items, α=0.86).²⁷ Collaborative practices with service providers was based on sharing 12 activities (e.g., pooled funding, joint staffing) with external providers.²⁸

County characteristics.

County-level measures were obtained from census data,²⁹ including percentage of families with children under 18 years of age living in poverty, percentage of children under age 18 years who were uninsured, and percentage of residents in urban areas. Information on available service providers per 100,000 residents came from the Centers for Disease Control and Prevention Children’s Mental Health portal.³⁰

Analytic plan

First, we examined independent and dependent (i.e., Cascade outcomes) variables for the overall sample and by site. We examined four Cascade steps (screening, in need of treatment, treatment referral, treatment initiation), and two time-related outcomes (screened within 30 days of intake and referred within 30 days of screening). Overall rates were calculated relative to all youths in the intake cohort. On the other hand, “within-step” rates were conditional upon positive events in the prior step, reflecting interest, for example, in examining referral only for those “in need”. We next examined correlations among independent variables, to ensure that multicollinearity would not compromise stability of regression analyses, and bivariate associations for each independent variable (youth, agency/county-level measures) and Cascade outcome.

Multivariable multilevel regression analyses used the same predictors for each outcome, although the contribution of alcohol and other drug-related charges was only examined for screening, because it was included in the definition of “in need”, and therefore in subsequent conditional Cascade steps. Measures included in each step were selected on an a priori basis to examine associations with Cascade outcomes (rather than data-driven modeling). In order, we entered (a) youth characteristics, (b) justice staff characteristics, followed by (c) agency organizational characteristics, and (d) county characteristics. Given the data’s hierarchical nature (youth data nested within county), we used SAS PROC GLIMMIX³¹ for the nested data structures, separating within-county and within-person variance from between-county and between-person variance.^{32, 33}

Missing data.

In a related JJ-TRIALS evaluation of 31,308 youth records from these sites,²² data were available on a median of 49 out of 72 (68%) core items that justice partners all reported as available in their records (e.g., demographics, offense-related and Cascade events). Data were missing because they were not collected, not coded similarly, and/or inaccessible. Cascade events were interpreted as “yes” versus “other”, whereby “other” included all “missing/no indication of event” data, providing conservative estimates of retention rates at each Cascade step.

RESULTS

Contributory Characteristics

Youth characteristics.

Youths were predominately male (74%), with approximately half white (49%). About a fifth were facing alcohol- or drug-related charges and a quarter violence-related charges (see Table 1). Slightly more than half received closer justice supervision.

Table 1.

Sample characteristics (N=8,307)^*

	n	%
Youth characteristics
Male	6164	74.2
Age (mean ±sd)	14.9±1.6
White (vs non-White)	3997	49.4
Hispanic*	1698	21.6
Current charges were alcohol or other drug-related	1362	18.5
Current charges were violence-related	2147	26.4
On higher level of community supervision	4426	55.4
Site-level justice staff characteristics	mean±sd	range
Average years of experience of justice staff	15.3±3.2	(6.8 – 20.2)
Average number of youths on justice staff caseloads	16.3±6.5	(2.3 – 31.5)
Perceived competency to identify youth behavioral health needs and service linkages	38.0±1.9
Site-level juvenile justice organizational characteristics (mean, sd)
Intra-agency communication	30.4±4.1
Job stress	35.1±3.9
Shared/collaborative practices with substance use providers	4.8±2.8
County characteristics	mean±sd
Percent of families with children below the Poverty Line	18.2±6.3
Percent of families with children <18 years who are uninsured	7.0±3.4
Percent urban	86.8±14.2
Number of psychiatrists, licensed social workers, and psychologists per 100,000 youths	37.4±27.8

^*Percentage based on slightly reduced N due to missing data.

Justice agency staff and organizational characteristics.

The average staff member had just over 15 years of experience and supervised 16 youths. The average perceived competency to identify youth behavioral health needs and link juveniles to services score was above the midpoint (38 of 50 points). Intra-agency communication scores averaged slightly above the midpoint (30 of 50 points), reflecting perceived need for better communication. Staff reported relatively high job stress (35 of 50 points). Finally, agency leadership reported collaborating with behavioral health partners in approximately five (of 12) different activities, most commonly “sharing information on youth needs for treatment services” (90.9%) and “joint staffing/case reporting consultations” (88.9%).

County characteristics.

Most justice agencies (86.8%) were in “urban” counties. Almost a fifth of families with children in these counties live below the poverty line; most children under 18 years have health insurance, including Medicaid. Finally, counties have an average of 37.4 service providers for every 100,000 persons.

Cascade outcomes.

Table 2 displays rates for each Cascade step, with rates highly variable across sites. Across 33 sites, just under three quarters of youths were screened (range=7.1%−100%) and more than half were found to need treatment (range=13.4%−95%). Among those in need, about a fifth were referred to treatment (range=3.7%−24%). If referred to treatment, rates for initiation and engagement were relatively high: two-thirds of those referred initiated treatment (range=7.9%−100.0%), almost half who initiated treatment engaged in services (range=5.9%−94.1%), and more than a quarter who initiated treatment remained in continuing care (range=10.0%−75.0%). On average, youths were screened within 104 days of justice intake. For those needing services, referral occurred within an average of 26 days.

Table 2.

Behavioral health cascade rates for overall sample

	n	%
Substance use screening	5942	71.5
Days between juvenile justice intake and screening (mean±sd)a	103.5±145.2
Screened within 30 days of juvenile justice intake (vs no)	4477	76.4
Screen Positive	2252	27.1
In Need	4294	51.7
Behavioral health referral	1203	14.6
Referral within In Need	940	21.9
Days between substance use screening and referral (mean±sd)b	25.7±59.7
Referred within 30 days of screening (vs no)	375	43.9
Treatment Initiation	638	8.9
Treatment initiation among those referred	572	67.5
Engagement	271	4.5
Engagement among those who initiated treatment	271	45.7
Continuing Care	152	2.6
Continuing care among those who engaged in treatment	152	26.4

^aRange=0–981 days;

^bRange=0–821 days.

Predicting Cascade Steps

Screening.

About 42% of the youth were screened with an evidence-based instrument (19% with MAYSI, 10% with YASI, 8% with PACT, 3% with SASSI, 2% with CRAFFT and 1% with GAIN). Youths with alcohol- or drug-related charges were about 40% more likely to be screened (OR=1.37, p=0.002); those receiving higher levels of justice supervision were almost three times as likely to be screened (OR=2.92, p<0.001; see Table 3). Finally, youths living in counties where fewer children had health insurance were approximately 60% more likely to be screened (OR=1.62, p=0.001).

Table 3.

Adjusted odds ratios predicting behavioral health cascade events and days until substance use screening and referral

	Screened OR (95% CI)		In Need OR (95% CI)		Referreda OR (95% CI)		Initiated Txb OR (95% CI)		Substance use screening within 30 days of juvenile justice intake OR (95% CI)		Substance use referral within 30 days of screeninga OR (95% CI)
	n=6882		n=7781		n=4026		n=731		n=4723		n=584
Youth characteristics
Male	1.06	(0.89–1.26)	1.39***	(1.24–1.56)	1.18	(0.94–1.48)	1.23	(0.71–2.13)	1.14	(0.90–1.46)	0.97	(0.47–2.02)
Age	0.97	(0.92–1.02)	1.30***	(1.26–1.35)	1.00	(0.94–1.07)	1.04	(0.88–1.24)	1.09*	(1.01–1.17)	1.51**	(1.17–1.94)
White (vs non-white)	0.99	(0.83–1.18)	1.40***	(1.26–1.56)	1.21	(0.99–1.48)	0.89	(0.55–1.43)	0.96	(0.77–1.21)	1.06	(0.56–2.01)
Alcohol or other drug-related charge (vs other)	1.37**	(1.11–1.67)	---	---	---	---	---	---	0.79	(0.61–1.03)	---	---
More (vs less) supervision sanction	2.92***	(2.42–3.54)	2.58***	(2.28–2.92)	4.89***	(3.81–6.28)	1.64	(0.92–2.91)	1.33*	(1.02–1.73)	0.35*	(0.14–0.89)
Site-level justice staff characteristics
Average years of experience of justice staff	1.08	(0.77–1.52)	1.09	(0.89–1.33)	0.80**	(0.70–0.92)	1.07	(0.79–1.43)	0.66	(0.43–1.09)	0.82	(0.60–1.12)
Average staff caseload	0.97	(0.86–1.10)	0.93*	(0.86–1.00)	0.95*	(0.91–1.00)	0.97	(0.86–1.09)	0.97	(0.83–1.13)	0.89*	(0.80–0.99)
Perceived competency to identify behavioral health needs and link youths to services	1.11	(0.68–1.79)	0.91	(0.68–1.18)	1.04	(0.85–1.26)	1.21	(0.76–1.90)	1.33	(0.71–2.56)	1.03	(0.67–1.58)
Juvenile justice organizational characteristics
Intra-agency communication	1.15	(0.79–1.67)	1.02	(0.86–1.20)	0.97	(0.87–1.09)	0.69*	(0.51–0.92)	0.51**	(0.34–0.78)	1.12	(0.86–1.46)
Job stress	1.06	(0.86–1.31)	1.04	(0.92–1.18)	0.97	(0.89–1.05)	0.74*	(0.56–0.97)	0.58***	(0.42–0.80)	1.10	(0.93–1.30)
Shared/collaborative practices with behavioral health providers	1.04	(0.76–1.42)	0.91	(0.75–1.10)	1.10	(0.97–1.26)	0.61***	(0.46–0.80)	1.02	(0.66–1.55)	0.93	(0.68–1.29)
County characteristics
Percent of families with children < the Poverty Line	1.03	(0.92–1.15)	1.02	(0.96–1.08)	1.02	(0.97–1.06)	0.97	(0.88–1.08)	0.84*	(0.73–0.97)	0.98	(0.91–1.07)
Percent of families with children <18 years who are uninsured	1.62***	(1.09–2.42)	1.12	(0.92–1.37)	0.91	(0.79–1.05)	1.37	(0.98–1.91)	1.48	(0.93–2.35)	0.80	(0.60–1.05)
Percent urban	1.04	(0.97–1.11)	1.00	(0.96–1.04)	1.00	(0.98–1.03)	1.05	(0.98–1.13)	1.02	(0.93–1.11)	0.96	(0.91–1.01)
Number of psychiatrists, licensed social workers, and psychologists per 10,000 youths	1.01	(0.97–1.06)	1.00	(0.97–1.02)	0.99	(0.97–1.00)	0.99	(0.95–1.04)	1.05	(0.99–1.10)	0.98	(0.93–1.02)
Percent variance attributable to Site	50.7		27.7		13.5		29.8		60.5		27.5
Percent difference from unconditional model	18.6		10.0		14.7		32.0		10.0		−7.1
Fit statistics
AIC	4109.15		8992.81		3268.31		598.10		2419.64		429.05
BIC	4134.07		9016.75		3291.76		616.27		2444.55		452.51

^aAmong youth identified as being In Need of substance use services.

^bAmong youths referred for substance use services.

NOTES: OR=odds ratio; AIC= Akaike information criterion; BIC= Bayesian information criterion;

^*p<0.05,

^**p<0.01,

^***p<0.001

In Need.

Youth characteristics were the most consistent predictor of service need. Older youths (OR=1.20, p<0.001) were 20% more likely to have identified service needs; males (OR=1.39, p<0.001), and white youths (OR=1.40, p<0.001) were approximately 40% more likely to have identified substance use needs. Those receiving higher levels of supervision were more than 2.5 times as likely to be identified (OR=2.58, p<0.001). The sole remaining significant contributor was caseload size: those under supervision of staff with larger caseloads had slightly lower rates of identified substance use needs (OR=0.93, p=0.042).

Referral.

Youths under higher levels of supervision were nearly five times as likely to be referred to treatment (OR=4.89, p<0.001) compared to those who were diverted or informally supervised. Juveniles from agencies with more experienced staff were less likely to receive referrals (OR=0.80, p=0.002), as were those supervised in agencies with larger caseloads (OR=0.95, p=0.036).

Initiation.

Significant contributors to service initiation were confined to agency organizational characteristics. In counties where referred youths were less likely to initiate treatment, staff reported greater levels of intra-agency communication (OR=0.69, p=0.011), greater levels of job stress (OR=0.74, p=0.031), and more collaborative practices with service providers (OR=0.61, p=0.001).

Predicting days to screening and referral.

Older youths (OR=1.09, p=0.024) and those receiving higher levels of supervision (OR=1.33, p=0.033) were more likely to be screened within a month of intake (Table 3). Juveniles seen in agencies reporting greater levels of both intra-agency communication (OR=0.51, p=0.002) and job stress (OR=0.58, p=0.001) were less likely to be screened within 30 days of intake, as were those in counties with more families living in poverty (OR=0.84, p=0.015). Older youths (OR=1.51, p=0.002) and those receiving less supervision (OR=0.35, p=0.028) were more likely to receive service referrals within 30 days of screening. Youths supervised in agencies with larger caseloads (OR=0.89, p=0.035) were less likely to be referred within 30 days.

DISCUSSION

Findings from this large, multisite study corroborate and extend prior research, which consistently shows that justice-involved youths do not receive necessary and appropriate behavioral health services.³⁴ Over 75 % of intakes were screened for those concerns, with over half of those requiring further evaluation or treatment, yet only one in five “identified” juveniles was referred to further clinical assessment or treatment. Examining outcomes step-by-step across the Cascade, our investigation has documented where in this process service gaps occur and identified critical targets for intervention.

Across all sites, while screening was robust, it was not universal, with the unadjusted rate between 7.1% and 100%. The two strongest contributors to being screened were being charged with a substance-related offense and a higher level of justice supervision. Many justice systems target youths for screening who present with an arrest for a substance-related offense and/or who have been adjudicated for a crime, both of which may be viewed as proxies for behavioral health problems. Agency policies do not embrace utilization of screening to identify “hidden populations” of juveniles with such problems.

Even after screening identified behavioral health needs, justice sites demonstrated low rates of referral to community services, and many youths took a very long time to receive referrals. Just over 20% of those “in need” of services were referred to providers; of these, fewer than half were referred within 30 days. In adjusted analyses, referral rates were almost five times higher for those “in need” who were receiving higher levels of supervision, compared to youths receiving lower supervision levels, consistent with longitudinal studies of juvenile detainees.³⁵ Although those at higher supervision levels were more likely to receive service referrals, these referrals occurred less swiftly than for those at lower supervision. The gap in time from screening to referral could be influenced by court processes, since a referral may not be provided until it is included in a formal court probation service plan.

Staff practices resulting in a greater likelihood of referrals for youths at higher levels of supervision (or in decisions to screen them) appear consistent with the Risk-Need-Responsivity (RNR) model.^{17, 36} The RNR framework directs intensive services to individuals at higher risk of serious recidivism, offering fewer interventions to lower risk individuals (presuming that they may desist from future offending without more intensive justice involvement). Staff are likely well aware that failure to comply with a service referral could result in increased supervision or more restrictive placement for these juveniles at lower risk for recidivism.^{37, 38} In a justice agency that both screens for behavioral health needs and follows RNR principles, it is mostly individuals high in both behavioral health need and recidivism risk who are likely referred to local providers, although all those with screening-identified needs require such services, even if their justice supervision status would be considered “low”. While considerable evidence supports RNR-based policies for targeting the values, attitudes, and family factors contributing to criminal activity among higher risk individuals, the relevance of this approach for service referral for individuals with behavioral health needs may not be supported.³⁹

The RNR-informed case management model seems to be at odds with the public health approach underlying the Cascade.¹⁴ This distinction may follow from the different ways health and justice agencies view the outcomes they address. Identification of most health outcomes presumes that treatment is viewed positively by clients, with few social or legal consequences for failure to seek treatment. Circumstances change when health outcomes impact the larger community’s wellbeing, due to contagion (as in school review of children’s vaccination status). For health conditions seen as contributory to further criminal activity, such as substance use, or certain mental health problems,^{10, 19, 20} the justice system is likely to apply punitive consequences for failure to follow through.

A large proportion of youths in contact with community justice agencies are diverted or handled informally. In 2010, nationwide, of 1,368,200 estimated delinquency cases, 68% were either not processed under court oversight or were placed on informal supervision.²⁴ Providing access to behavioral health services for juveniles with identified needs can successfully prevent further offending.^{12, 13} Addressing youth service needs while avoiding possible further justice consequences for non-attendance requires balancing health and justice perspectives for lower risk youths. For these juveniles, referrals can be offered and supported without punitive legal consequences for non-compliance. Policies to support prevention can further community needs as well as justice agency goals (i.e., preventing reoffending).

In developing a co-ordination plan, justice and behavioral health agencies should consider the principles underlying each partner’s decisions to refer or treat particular individuals. Justice agency policy that results in only youths with profiles that most seriously point to future criminal activity (i.e., those with a history of more serious offending, antisocial attitudes or peers, fewer family supports) being sent on to local providers, has implications for clinical practice. These youths are less likely to initiate and engage in offered services,⁴⁰ and their multiple problems compound the challenges in addressing their diagnostic concerns for provider agencies. Once initiating behavioral health services, justice-involved youths often drop out or experience service gaps.⁴¹ Both lack of parental buy-in and family dysfunction are likely to be associated with lower adolescent behavioral health service use.^{16, 42, 43} Challenges such as these strongly underscore the need for interagency coordination to support service linkage.^{16, 41}

This study contributes to understanding of how staff- and community-level features underlie practices along the Cascade. For example, net of other features, youths with identified behavioral health needs were more likely to get service referrals if they were seen at supervision agencies with smaller caseloads, and when supervised by staff with fewer years’ experience at this job. Effectively managing the service referral process, addressing potential barriers and engaging in practices that support service initiation (e.g., confirming that appointments are kept^{18, 25}) require considerable staff effort. More recent hires may have been exposed to training that stresses the importance of understanding the behavioral health needs of justice system youths.⁴⁴

While we are unable to pinpoint directionality, some associations likely reflect the consequences of lower rates of service utilization: after referral, for juveniles who did not initiate treatment (compared to those who did), staff reported greater levels of intra-agency communication, more job stress, and more collaborative practices with local providers, perhaps as they struggle to address non-attendance. While contributors to optimal agency practice deserve further study, advance inter-agency planning that clarifies which youths will be referred, to which partner agencies, and how attendance will be monitored may actually reduce effort as agencies spend fewer resources determining how to approach youth needs after the fact.

Limitations.

Common to most descriptive investigations of agency services, this study has a number of limitations. Chief among these might be the high level of missing data in administrative records and the non-random selection of participating states and local sites. Our conservative approach to estimation²² was applied to address the former; despite the latter, participating sites nonetheless showed wide ranges on most measures, demonstrating the great variability across such systems nationwide.

Conclusions.

This multisite study found that justice agencies identified youths’ treatment needs but fell short in making referrals that should result from that identification. The strongest predictor of youth participation in behavioral health services was their level of supervision. This suggests that justice agencies may utilize a model for service referral (RNR) that may not support individuals with behavioral health needs, as opposed to those with serious risk of re-offending, although both systematic screening and referral are possible within these settings. Failure to provide juveniles at lower levels of supervision with needed treatment referrals is worrisome, given their high numbers and the value of those services in preventing further justice system involvement. Increased cross-agency planning should facilitate systematic referral of those whose needs have been identified, reducing unmet youth service needs, and, in turn, enhancing both public safety and public health.

Highlights.

• In this large (N=8307 youth) multisite investigation of youths in contact with community justice agencies, over 75 % of intakes were screened for BH concerns, with over half of those requiring further evaluation or treatment.

• Only one in five “identified” juveniles were referred to further clinical assessment or treatment, and fewer than half of those were referred within 30 days.

• Community justice agencies appear to follow an approach to practice that focuses identification and linkage practices on youth with criminogenic (e.g., legal history), rather than BH needs, although both contribute to re-offending.