Preventing opioid use among justice-involved youth as they transition to adulthood: leveraging safe adults (LeSA)

The specific aims of the LeSA Project are to: (1) test the effectiveness of the TBRI intervention and its different support formats – no in-home TBRI support, Structured Coaching, Responsive Coaching – on (a) preventing initiation and/or escalation of opioid use, (b) public health and public safety outcomes, (c) putative change mechanisms, including self-regulation, psychosocial functioning, and relationship with a safe adult; (2) document factors related to TBRI implementation and sustainment; and (3) estimate the start-up and implementation costs and cost-effectiveness of TBRI support options relative to standard re-entry practices in achieving lower rates of opioid initiation and reductions in healthcare and JJ system costs.

Primary hypotheses are as follows:

TBRI is a whole-child, attachment-based, trauma-informed intervention that promotes emotional regulation through interaction with responsive adults. The model is grounded in attachment theory and research [42, 43], emphasizing mindful awareness and engagement between adults and children as a means of building trust and resilience [37]. TBRI is also designed to address relational trauma, for instance, in the three pillars of trauma-wise care: felt-safety, connection, and self-regulation [39, 40]. The three evidence-based principles of TBRI include connection, empowerment, and correction [37]. Each principle includes a set of specific strategies that address social, behavioral, emotional, physiological, cognitive, and ecological aspects of children’s development and well-being (see Table 1).

The primary TBRI intervention is an adapted version of the TBRI Caregiver Curriculum [33, 37] and includes several components (see Tables 2-3): (1) Youth Trainings: 9, 45-min youth-only modules; (2) Caregiver Trainings: a 1-h individual introductory module and 9, 90-min caregiver-only modules; and (3) Nurture Groups: Youth and caregivers participate together in 4, 1-h joint-roleplay sessions. Both Youth and Caregiver Trainings are designed to be delivered in a group setting with three to five in each group; however, missed sessions can be delivered as individual sessions. Additional monthly “booster” nurture group sessions are delivered in instances where the youth’s discharge is extended beyond the original expected date.

For the purpose of this project, TBRI intervention components are delivered by 1–2 Research Assistants (RAs) who are also trained as TBRI Practitioners. All TBRI Practitioners have a minimum Bachelor’s Degree in human services areas, more than 2 years of clinical experience and fieldwork in high-risk juvenile and adult populations, receive a minimum of 20-h onboard training and skills development as well as ongoing training/supervision. All caregiver sessions are delivered virtually to enable caregivers who do not live close to the site where their youth resides. Sessions are synchronous, and participants are provided an “orientation” to the online platform prior to the first session. Youth sessions are delivered in person at the facility by RAs, with support from non-correctional staff (e.g., clinical staff). Nurture groups are delivered in a group format with youth joining virtually as a group from the facility and caregivers joining individually from their homes. Sites determine when and where youth sessions will be delivered, including which staff to involve (i.e., therapeutic staff who could participate in intervention activities. Staff are provided training on logistics and strategies to address potential behavioral issues and support a youth in need.

In addition to the features described above, TBRI also serves as a multi-systemic caregiving model that has been successfully implemented in a wide variety of professional contexts [44] [45] [46] [47]. TBRI can be implemented by anyone trained in the tools and strategies, and therefore provides a framework and impetus for empowering all adults, regardless of status or education, to be catalysts of change for youth. Therefore, TBRI can be implemented across an entire JJ system to facilitate the use of trauma-informed approaches in all interactions. For the purpose of this project, the implementation of TBRI principles and strategies within the broader residential context is referred to as the organizational TBRI intervention.

The LeSA Project uses an effectiveness/implementation Hybrid Type 1 Design [48] to simultaneously test the effectiveness of TBRI for preventing non-medical use of opioids among JJ-involved adolescents and gain insight into facility-level obstacles that must be overcome if the intervention is to be implemented and sustained as part of JJ re-entry protocols. The overall study design includes (a) a delayed-start at the site level and (b) a randomized controlled trial at the participant level.

To address the logistical burden of protocol implementation and to control for the influence of other exogenous factors [49], participating sites are assigned to one of three Study Waves and proceed through three sequential study stages accordingly. Wave membership is determined by stratifying by state (3 sites in one, 6 in the other) and ranking by the number of youth served in descending order. For the state with 3 sites, one is assigned to each wave, with the highest populated site starting in Wave 1. For the state with 6 sites, two are assigned to each wave, with the highest and fourth-highest populated sites starting in Wave 1. Stage 1 - SRP data collection launches as soon as sites are ready. Stage 2 study activities are delayed by 1 month.

The effectiveness trial includes a comparison of youth enrolled in SRP to youth enrolled in SRP plus TBRI, whereby each facility serves as its own control. Because caregivers may need support after their youths’ release, three versions of TBRI are compared, with youth in the SRP + TBRI cohort being randomly assigned to one of three support formats: (1) TBRI Training only (no in-home TBRI support), (2) TBRI Training + Structured Coaching (4 sessions), and (3) TBRI Training + Responsive Coaching (2+ sessions, triggered by youth need/risk).

The effectiveness trial proceeds in 3 stages: Standard Reentry Practice (SRP) only (baseline control), SRP + TBRI, and Facility Sustainment (see Fig. 2). Stage 1 - SRP: sites engage in standard practice; study activities include youth and caregiver assessments over 18 months. Each site serves as its own control. A total of 120 youth/adult dyads are recruited across the nine sites. The length of SRP varies between 8 months (Wave 1) and 12 months (Wave 3). Stage 2 - SRP + TBRI Intervention: sites continue to engage in standard practice; TBRI intervention is delivered to youth and caregivers enrolled in the study; youth and caregiver assessments are conducted over an 18-month period. A total of 240 youth/adult dyads are recruited across the nine sites. Stage 3 - Facility Sustainment: youth/caregiver recruitment, baseline assessment, and interventions cease; youth and caregiver follow-up interviews continue; responsibility for implementing primary and secondary TBRI intervention components is transferred to the JJ facility.

At the participant level (see Fig. 3), for both SRP and SRP + TBRI stages, youth and safe adults are recruited/consented 2–3 months prior to discharge, and individually complete baseline at recruitment and follow-up interviews at 3- (within 2 weeks of discharge), 6-, 12-, and 18-months post-baseline. For dyads recruited during the Stage 2 – SRP + TBRI Intervention period, the primary TBRI intervention sessions begin after baseline assessment while the youth are in the residential facility. At the point of discharge, dyads are randomly assigned to one of 3 TBRI in-home support conditions, with probabilities equal to 1:3 within each site. This overall design enables comparison of TBRI versus SRP plus a randomized control trial comparing TBRI support conditions.

The implementation overlay includes three types of research activities occurring throughout all three project stages: (1) documenting facility use of trauma-informed strategies and contextual factors (see Measurement and outcomes) that facilitate or impede the use of these strategies, (2) supporting facility efforts to become more trauma-informed and in their use of TBRI practices during daily interactions with youth (see Support for the organizational TBRI intervention), and (3) building the necessary infrastructure to sustain the primary and secondary TBRI intervention components after the project ends (see Sustaining the primary and secondary TBRI intervention).

Prior to the launch of Stage 1 - SRP, the research team and facilities work together to explore each facility’s needs and desire to implement trauma-informed strategies such as TBRI. Contextual factors and use of trauma-informed strategies are documented through surveys of JJ facility staff at four time points throughout the study: the beginning of Stages 1, 2, and 3, and the end of Stage 3 (see Fig. 2). This information is used to customize support for facilities seeking to implement TBRI across their system and to understand how contextual factors—specifically how the use of trauma-informed care at a facility level—moderate the effectiveness of the primary and secondary TBRI intervention.

A total of nine residential facilities designed to address the behavioral and emotional needs of justice-involved youth serve as research sites for this project. Three are located in a Midwestern state and 6 in a Southwestern state in the U.S.A. Three are operated by the state JJ system, 4 by local JJ departments, and 2 by an independent agency. All sites are secure residential programs with medium security protocols, all receive referrals from JJ departments across a wide geographic region, and all provide educational, medical, and behavioral health services. At project start, 3 sites had incorporated TBRI into their trauma-informed care approach for 1–3 years, 2 were familiar with TBRI but not using principles and strategies, and 4 were unfamiliar with TBRI.

The project involves two target samples drawn from the nine facilities at rates proportional to current census/staffing levels. The effectiveness trial sample includes a total of 360 youth/caregiver (e.g., parent/guardian, extended family member) dyads recruited from participating JJ facilities over a 3-year period (approximately 10–15 dyads/year per site). Inclusion criteria include youth ages 15–18 at study enrollment who (a) are disposed to community supervision (i.e., probation) and receiving care in a secure JJ facility, (b) have no indication of active suicide risk, and (c) have worked with JJ staff to identify one caregiver that is willing to participate in the study. The implementation overlay sample includes approximately 203 (out of 810 potential participants with a 25% completion rate) JJ staff (facility staff and probation officers; an average of 22–23 per site) who work directly with youth during residential and/or community re-entry. Out of this sample, approximately 90 JJ staff (an average of 10 per site) are recruited for focus groups.

Research sites facilitate the identification and recruitment of family participants (e.g., connecting RAs to families). Brochures, consent and assent forms, assessment battery, and curricula are available in English and Spanish. Bilingual English and Spanish-speaking RAs are available for Spanish-speaking families in all research activities. The research protocol is approved by the Texas Christian University (TCU) Institutional Review Board (IRB) and by site research review committees (if applicable), and the local juvenile board (if applicable). Assessment data are collected via Qualtrics. Protocols are delivered via secure virtual platforms (e.g., Zoom, Microsoft Teams) or in-person, depending on the activity. Caregiver consent, parental/guardian permission for youth participation, and youth assent are obtained in writing by RAs prior to all research activities; youth provide consent in writing if they turn 18 during the study. Youth and caregivers are compensated separately for their time spent in assessments in an amount of $15 e-cards per assessment.

Research staff work with JJ facility leadership to determine participant lists for staff surveys and focus groups. All individuals with direct youth contact (e.g., case managers, educators, direct care staff, probation officers) receive an individualized link to a confidential survey administered via Qualtrics. Recruitment of newly hired staff occurs prior to each assessment. JJ leadership are asked to identify up to 10 individuals representing leadership and line staff, including admin, clinical supervisors, justice probation officers, clinical staff, direct care staff, for in-person focus groups. To avoid potential coercion, introduction of risk to subordinates’ disclosures, and/or biased/invalid focus group data, leadership and line staff are invited for separate in-person focus groups at each site. Consent is obtained prior to the first assessment by an RA and prior to each focus group interview. Staff participants receive an honorarium (e.g., a cup, group meal) for participation.

For the effectiveness trial, data come from 3 sources: (1) youth self-report at Months 0, 3, 6, 12, and 18; monthly post-release check-ins, (2) JJ facility records (e.g., recidivism, treatment utilization), and (3) caregiver self-report with a timeframe identical to youth. Table 4 depicts research goals, assessment instruments, data collection timeline, and youth and/or caregiver single or dyadic perspective on each construct. The primary outcome is days to youth initiation of non-medical opioid use: measured by (a) Timeline Followback [51], (b) TCU Drug Screen 5 with Opioid Supplement (TCU DS 5-OS) [52], and (c) Substance Use Involvement items from the HEAL Prevention Cooperative [53]. Secondary outcomes include putative change mechanisms, public health, and public safety outcomes. Putative change mechanisms include youth self-regulation and self-efficacy (e.g., “How confident are you that you will not misuse a prescription opioid in the next 30 days”) [53] and youth/caregiver psychosocial functioning and relationships between youth and caregivers. Self-regulation is measured with the TCU Adolescent Thinking forms on regulation (TCU THK-NUY, TCU THK-PUY; decisions influenced by positive or negative emotion) [54], supplemented with the delayed discounting task (DD task) [55], Difficulties in Emotion Regulation Scale (DERS) [56], and Barkley Deficits in Executive Functioning Scale-CA Short Form (BDEFS-CA) [57]. Psychosocial functioning is measured by the Strengths and Difficulties Questionnaire for youth (SDQ) [58], Generalized Anxiety Disorder Assessment (GAD-7) [59], Patient Health Questionnaire (PHQ-8) [60], PROMIS-Pain [61], and items that evaluate youth’s self-reported efficacy to resist opioids (“How confident are you that you will not misuse a prescription opioid in the next 30 days?”) [53]. Youth/caregiver relationships are measured using the Family Assessment Device (FAD) [62], the Experiences in Close Relationships – Relationship Structures (ECR-RS) [63], and the Attachment, Relational Frustration, and Discipline Practices scales from the Behavior Assessment System for Children Parent Relational Questionnaire (BASC-3) [64].

Public health and safety outcomes include opioid-related health, opioid use overdose, use of other illegal substances, recidivism, and SU or MH treatment records. Opioid-related health outcomes include SU treatment, hospital visits related to SU, diagnosis, overdose, and receipt of Narcan [65]. Recidivism information (re-arrest, re-adjudication), urinalysis results, and treatment records (type of referral, service receipt) come from JJ facility youth records [65]. For these data, sharing agreements between each state-level JJ facility and the research institute are established, specifying how de-identified records are shared, stored, and analyzed, following models from previous collaborations [66]. Background information on youth is extracted from existing JJ youth records (e.g., offense, disposition decision, supervision level, offense history including truancy, and prior SU or MH diagnosis).

Youth and caregivers provide different pieces of background information that serve as covariates or moderators, including demographics (e.g., age, sex, race/ethnicity, employment, education). Caregivers are also asked about their SU (TCU DS 5-OS) [52] and relationship with their own caregiver using the Experiences in Close Relationships-Relationship Structures (ECR-RS) [63]. Other potential covariates include number and type of ACEs [67], youth self-reported social exposure to substance use (e.g., “How often does the adult who is most important to you drink alcohol?”) [53], services youth and family received (adapted from the Child and Adolescent Services Assessment) [68], caregiver motivation in treatment participation (Parent Motivation Inventory; PMI) [69], parenting strain (Caregiver Strain Questionnaire-Revised Short Form; CGSQ-SF) [70], and parental resiliency (Resiliency of Self-Efficacy Scale; RS) [71]. Participants are also asked to respond to two open-ended questions regarding the impact of COVID-19 (e.g., “How is the COVID-19 crisis affecting you or has affected you?”). Monthly phone check-ins with youth assess 8 indicators of any opioid use; any alcohol, other drug use; truancy; trouble with the law, increases in depression, increases in anxiety, increases in the stress in relationship with the safe adult in the past month.

Participants in the SRP + TBRI cohort are asked to report on the utilization of intervention content and intervention fidelity, which are developed specifically for this project. Caregivers’ use of TBRI strategies is assessed by (1) self-report after completion of final TBRI Group Training session and after each TBRI In-Home Training session and (2) RA report upon completion of each TBRI Nurture Group and TBRI In-Home Training sessions. RAs and participants complete a post-session checklist after each session to assess fidelity of content delivery and participant engagement.

For the implementation assessment, the LeSA JJ staff survey (see Table 5 for a description and timeline) includes information on organizational context (Survey for Juvenile Justice Secure Facilities) [72], staff characteristics (Survey of Organizational Functioning and Leadership, Professional Quality of Life; SOFL, PQL) [73], use of trauma-informed care (Attitudes Related to Trauma-Informed Care Scale; ARTIC) [74], use of TBRI intervention strategies (TBRI Acceptability, Appropriateness, and Feasibility scale), and TBRI Professional Use scale [75], and answer two open-ended questions regarding the impact of COVID-19. Youth participating in LeSA also report their perceived social support from facility staff (with the Berlin Social Support Scale; BSSS) [76]. During focus groups, staff report routine practices in the use of trauma-informed approaches (e.g., “How does your facility address issues that youth may have had due to their traumatic experience(s)?”) and child-centered strategies (e.g., “In what ways does your agency strive to meet the unique needs of youth?”), agency facilitators/barriers to uptake of new interventions (e.g., “What (if any) changes have you made that produced significant changes for youth?”), general penetration of TBRI within the agency (e.g., “What (if any) changes have you made using TBRI that produced significant changes for the staff or your organization more broadly?”), and agency interest/success in sustaining and/or expanding TBRI practices (e.g., “Are you planning to continue TBRI practices at your agency?”). The focus groups will be held once per year over 4 years for approximately 60 min.

For economic analyses, the Drug Abuse Treatment Cost Analysis Program (DATCAP) [77, 78] is used to capture implementation and operating costs of TBRI [79] [78], generating cost summary statistics using data on intervention engagement and youth case flow (e.g., the average annual cost per youth; average cost per treatment episode per youth) [80] [81].

For all analyses, an ITT approach is used, where all study participants are included in analyses. For H1 and H2, comparisons of survival curves for each group are performed using SAS PROC PHREG and PROC GLIMMIX using the log rank test with the Sidak multiple-comparison adjustment, with the primary outcome as the dependent variable. Differences between groups on secondary outcomes and change over time in implementation measures (e.g., TBRI use) are examined using repeated measures ANOVAs (SAS 9.4, PROC MIXED), with study site as random effects and condition, time, and interactions as fixed factors. Because SU rates are generally lower among JJ females, sex as a biological variable is examined as a moderator of differences in outcomes using univariate analyses (e.g., chi square, t-tests, ANOVA); when significant, tests of moderation are undertaken as part of all analyses. Putative change mechanisms are investigated as mediators of the Group – Primary Outcome relationship and structural equation modeling (SEM; using Mplus) is used to examine sex differences in mediation. For H3, changes in the outcomes over time are examined using repeated measures ANOVAs (SAS 9.4, PROC MIXED), with study site as random effects and time and time × site interactions as fixed factors. For H1–3, a log rank test is used to assess study dropout; stochastic multiple-imputation provides maximum likelihood estimates, accounts for uncertainty in missing values, and provides unbiased estimates of the treatment effect if data are missing at random.

Qualitative data from focus groups are transcribed and qualitatively coded (in Atlas.ti) by at least 2 research staff. An iterative process of coding and thematic analysis [84] are used to understand routine practices employed, nuances in the use of trauma-informed and child-centered strategies, facility facilitators/barriers to uptake of new interventions, general penetration of TBRI within the facility, and facility interest in sustaining and/or expanding TBRI practices.

To address H4 regarding economic components, an activity-based micro-costing approach is used to identify, measure, and value resources invested by participating sites, and link costs to primary and secondary study outcomes [85]. Cost comparisons represent the incremental costs of adopting SRP + TBRI relative to SRP alone. Incremental cost represents the difference in facility/provider cost after adopting the intervention vs before adopting [86, 87], and incremental effectiveness is the difference in time to opioid use initiation in TBRI relative to SRP (and between TBRI support conditions). For each facility, differences in costs and incremental cost-effectiveness ratios are estimated using multi-level general linear models (e.g., generalized linear mixed models) to account for clustering at the facility/provider and patient levels. Nonparametric bootstrapping is used to compute confidence intervals [88].