Wraparound for High-risk Families with Substance Use Disorders: Examining Family, Child, and Parent Outcomes

No registry is involved.

The Research Questions include:

1. Do families receiving Wraparound intervention achieve higher parenting and family functioning outcomes than treatment as usual?
2. Do children of Wraparound families achieve higher levels of child well-being, safety and permanency than those receiving treatment as usual?
3. Do parents receiving Wraparound intervention achieve greater reductions in addiction severity, trauma symptoms, and more sustained recovery than those receiving treatment as usual?
4. Does participation in a Wraparound lead to greater engagement in services and development of family strengths?
5. What factors lead to higher levels of engagement and better outcomes?
6. What are the barriers and facilitators associated with engagement and outcomes?

The data collection and analytic plan described in this document include only those activities planned for research questions 1-3.

Recruitment. Enrollment into the study will begin fall 2020 and run through spring 2023 for a total of 30 months. The project has multiple referral sites spanning medical centers, addiction treatment or recovery centers, child protection, family courts, family resource centers, and other social service agencies. The investigators plan for a rolling recruitment process, aiming for 40 participants enrolled per year. (Wraparound case-loads should not exceed 10 families and there are 2 Wraparound coordinators).

Group balance (size, Division of Children Youth and Families involvement, other demographics) will be monitored by the evaluation team at least semi-annually (which should be sufficient given the 12-month intervention) and any adjustments to the sampling procedures will be reviewed with the Data Safety Monitoring Board and national cross-site evaluation liaison before executing. When cases from either group are lost to follow-up or Wraparound caseloads drop below 10 families per Wraparound coordinator at any given time, the investigators will randomly select families from a short-term wait list or if necessary, re-engage referral sites to enroll and consent at least four families from which to randomize into the intervention and treatment-as-usual groups to restore target sample size.

Data Collection. Outcomes will be measured on five domains: Family Functioning (primary), Child Well-Being, Child Safety, Child Permanency, Parent Recovery. Outcome measures for the domains of family functioning (primary outcome; parenting and parent depression), child wellbeing (child behaviors/functioning), and parent recovery (addiction severity, trauma symptoms) will be collected from participants in both study arms at baseline, 6 months (mid-point), and 12 months (post-intervention/treatment as usual). The battery of outcome measures will take around 60 minutes to complete at each of the 3 time points. Child safety and permanency data, as well as parent engagement in substance use disorder treatment and recovery services will come from New Hampshire Department of Health and Human Services administrative records.

The project coordinator or Wraparound coordinator will conduct the baseline Wraparound data collection on both intervention and treatment-as-usual families. Midpoint and endpoint assessments will be conducted by members of the evaluation team to help reduce bias. Midpoint and endpoint assessments will not be blinded.

The investigators will also gather state-level Department of Health and Human Services administrative data on all recovery domain adults and all children involved in cases whether in the Wraparound intervention or treatment as usual groups. The investigators will request all available data for adults (from when they turned age 18 through the end of the grant - Sept. 2024) and lifetime data (from birth through end of grant) for children who are a part of the case. The investigators estimate 2 children per family. Not all enrolled families will have had past involvement with the Division of Children Youth and Families. The investigators estimate that 75% of cases will have past or current Division of Children Youth and Families involvement and have administrative data on children. A data share agreement between New Hampshire Department of Health and Human Services, Dartmouth-Hitchcock Health, and Dartmouth College is underway.

Data Management. All participant data will be stored in a Health Insurance Portability and Accountability Act-, and Service Organization Control 1/2/3-compliant, password-protected, encrypted database serving as the central project data management system. Each case and case member (parent, child, other guardian/caregiver) will be assigned a unique case identifier at the time of enrollment and consent into the study. The only people who will have credentialed full access to the data will be members of the evaluation team. Other project team members will have limited credentialed access to be able to enter data on children and parents and view data reports (individual or aggregate). A Data Share Agreement is in progress to assist with administrative data sharing between Dartmouth-Hitchcock Health, Dartmouth College, and the New Hampshire Department of Health and Human Services.

Analyses. Given the complexity and multi-level nature of the data in this study (e.g., children nested in families; families referred from different access points), the investigators will employ a range of analytic methods.

Child-Level Outcomes: The investigators will use generalized linear mixed models (GLMM) to evaluate the effect of the Wraparound intervention on child well-being over time and compared to the children in the treatment as usual group. They will use generalized linear mixed modeling, which models both fixed and random effects and provides flexibility for modeling outcomes with binary or non-normal distributions and clustered or grouped data. Link functions within generalized linear mixed models will model outcomes measured on binary or dichotomous scales (e.g., logistic/logit link), continuous scales (e.g., regression/identity link), and counts (e.g., Poisson regression/log link). Most of the child well-being measures will be continuous in nature, but some safety and permanency data will be binary. Analyses will also control for potential covariates associated with the different referral/access point (e.g., those involved and not involved with New Hampshire Division of Children Youth and Families).

Child well-being measures differ by age group. One measure is used with very young children (Infant-Toddler Sensory-Profile, children 0-18 months), and two versions of another measure (Child Behavior Checklist) are divided by age (10 months - 5 years; 6 -18 years). Scores from the two versions of the Child Behavior Checklist can be combined.

Parent- and Family-Level Outcomes: The investigators will use Repeated Measures Analysis of Variance and McNemar's Chi-square to demonstrate the effects of program participation on parent (e.g., recovery measures, parenting skills) and family outcomes. Where covariates are important to model, they will use repeated measures Analysis of Covariance and logistic regression. Multiple regression will identify factors that explain family engagement and perceptions of support.

Adverse events. The study team, Wraparound coordinators, and evaluators (during midpoint and endpoint assessments and check-ins) will closely monitor adverse events. They will report any adverse event (e.g., suicide attempt, drug overdose) immediately to the Institutional Review Board and Data Safety and Monitoring Board (already established) to carefully consider whether the study impacted these events, how to best support the participant, and how to otherwise report and mitigate the events.

Change management. Not applicable. No registry.

Sample size and power. With a minimum of 80 parents for whom self-reported family functioning and recovery outcomes are expected in each group, the investigators have ample power (0.99) to detect a medium effect size across three time points (η2=.06), assuming moderate correlations (0.3) between the 3 time points and alpha of .05. For parent administrative data gathered from the New Hampshire Bureau for Drug and Alcohol Services, the investigators estimate 1 parent per family, leading to 80 per group. However, the investigators expect 90% (n=72 per arm) of these parents to have administrative data available in the Bureau for Drug and Alcohol Services. This analysis will be powered at .99 to detect a medium effect size with an alpha of 0.05.

For children, the investigators estimate having 65 children in each group for whom the same child well-being measure (Child Behavior Checklist) will be available. (the investigators will only collect this measure on 1 focal child per family). For this child well-being outcome, the investigators estimate power at .99 to detect differences of medium effect size (η2=.06) across three time points assuming moderate correlations (0.3) in the measures between time points and an alpha of 0.05. In terms of child safety and permanency administrative data comparisons, the investigators estimate 2 children per family (n=160; 80 per group). However, the investigators expect about 75% of children to be involved with the New Hampshire Department of Children Youth and Families and only those with this involvement will have administrative data (child safety, permanency) available. The investigators therefore estimate 60 in each group. The analyses are still powered at .99 to detect a medium effect size with an alpha of 0.05.

All outcomes data (at the individual level) will also be used in a national cross-site evaluation to study intervention effects with a combined sample across all grantees.

Plan for missing data. The outcomes analyses will utilize an intention to treat approach to minimize potential for bias based on drop out from the intervention and loss to follow-up, and imputation to account for missing data.