The goal of this Phase II SBIR is to develop and test a novel measurement and feedback system tailored to Wraparound Service Model (WSM), called SMART-Wrap. Aim 3 consists of a controlled efficacy study of SMART-Wrap in which 2 WSM provider organizations (WPO) will participate in a parallel group-randomized trial. Using a cascading recruitment strategy, 24 care teams will be recruited, with 12 teams randomized to the SMART-Wrap intervention versus 12 to "Services-as-Usual" (SAU), stratifying by supervisor to balance clustering effects. Randomization will be done at the care coordinator (CC) level. Each WPO supervisor (n = 12) will have equal numbers of CCs assigned to each condition (approx. 2 CCs for each supervisor), yielding 6 supervisors in each condition (4 per WPO). Care team participants must agree to use the SMART-Wrap prototype as part of their WSM over an 8-month test period and complete measures of service and implementation outcomes. CC use of standardized assessment, plan revision, and other SMART-Wrap promoted behaviors will be assessed via monthly web-based surveys. Investigators will enroll n=5 caregivers (CGs) per CC who will be the primary focus of data collection. This will yield a total recruited sample of N=120 (n=60 per group) caregivers of youth aged 8-17 with SED. Family participants will participate for 4 months total, initiating within a 4-month enrollment period. Family participants will respond to measures regarding service and youth outcomes at baseline and at 4 months.
Investigators hypothesize the following for our primary outcomes: (1) compared to care teams in the control group, care teams in the SMART-Wrap group will demonstrate (a) greater use of data and feedback in service delivery; (b) greater fidelity to the Wraparound process; (c) higher self-reported teamwork, working alliance, and satisfaction with the intervention; and (d) more positive attitudes toward standardized assessment; and (2) compared to the control group, caregivers of youth receiving services from care teams using SMART-Wrap will report (a) greater goal clarity; (b) greater use of MBC strategies; (c) greater satisfaction with services and progress; (d) better fidelity to Wraparound; (e) more effective team functioning; and (f) more positive outcomes, including reduced caregiver stress and improved symptoms and functioning.
Our exploratory hypotheses focus on supervision to examine whether (1) supervisors randomized to SMART-Wrap report (a) a greater proportion of supervision time spent reviewing data on family progress and case conceptualization and strategies; (b) greater alliance with the care coordinator, and (c) greater perception of CC effectiveness. Investigators will also explore (2) the validity of SMART-Wrap measurement by asking whether domains assessed via SMART-Wrap in service delivery (e.g., alliance, progress, satisfaction, functioning) are associated with data collected by the external research team via interviews and surveys.
To evaluate differences on youth and CG outcomes (youth functioning on the Top Problem Assessment (TPA), youth symptoms on the Brief Problem Checklist (BPC), and caregiver strain on the Caregiver Strain Questionnaire (CSGQ), investigators will use a series of 3-level longitudinal hierarchical linear models (separate model per outcome variable) to compare differences in group rates of change and timepoint mean scores based on intervention condition. Investigators will also test for curvilinear time trends over five timepoints of data on the TPA.
To evaluate implementation outcomes, investigators will conduct the following analyses. A series of 2-level cross-sectional HLMs with families nested within CCs will be run to assess CG and CC Wraparound fidelity on the Wraparound Fidelity Index, Brief Version as a function of intervention condition. Separate 3-level longitudinal HLMs will be run to test for differences, as a function of intervention condition, on CC satisfaction on the Therapist Satisfaction Inventory, attitudes toward standardized assessment on the Attitudes Towards Standardized Assessment scale, and percentage of time during supervision meetings spent (1) reviewing data on family progress and on (2) case conceptualization and strategies on the Supervision Process Questionnaire (SPQ) (with levels for timepoint, CC, and supervisor on the SPQ). Longitudinal HLMs will be run exploring the rate of change on use of data/feedback and other MFS targets on the Monthly CC Report (MCR) and Monthly CG Report (MPR), and investigators will test for curvilinear trends over 5 months. Investigators will explore differences on CG and CC alliance on the Working Alliance Inventory, satisfaction, and attendance using HLMs. Dropout is binomial; thus, investigators will use Hierarchical Generalized Modeling, log-link function to estimate odds of dropout.
To evaluate exploratory analyses, investigators will conduct the following analyses. With two HLMs, investigators will predict the trajectories of youth functioning (e.g., the TPA), while accounting for time-varying changes in MCR and MPR total scores. To explore the validity of SMART-Wrap measurement, investigators will associate SMART-Wrap measures with other measures. Using Pearson r and Intraclass Correlations (ICCs), investigators will explore the relationship between timepoint-matched CC scores on youth functioning in SMART-Wrap and CG-rated scores on the TPA, CGSQ, and BPC. Investigators will also explore Pearson r correlations between ratings of progress as scored in SMART-Wrap, and HLM-calculated rate of change on the caregiver-rated TPA. Investigators will examine the relationship between parent satisfaction as recorded in SMART-Wrap and Parent Satisfaction scores. Second, investigators will study how the use of the system may be significantly related to improvement on several outcomes, including fidelity, working alliance, and youth problems and symptoms. Predictors will include CC and supervisor use of SMART-Wrap system elements, data completeness, and time in system. For these analyses, investigators will use HLMs as appropriate to the DVs (longitudinal or cross-sectional) Across analyses, investigators will model data missing at random using full maximum likelihood estimation to account for attrition bias and investigators will control for multiple comparisons using false discovery rates.
