Defining “High-Acuity”: Using Linked Administrative Data to Prospectively Identify Children Likely to Enter Restrictive Care Settings

Background: Federal and state agencies have prioritized reducing the use of restrictive care settings for youth in state custody. Despite this, agencies continue to report difficulty securing appropriate family-based or community placements for those whose needs are characterized as requiring more specialized or intensive services. Yet, efforts to prospectively identify this population, sometimes referred to as “high-acuity youth”, are limited. This study seeks to fill this knowledge gap by asking three questions: (1) what are the prevalence and features of medical complexity (MC) among children involved with the CWS; (2) is MC associated with subsequent use of restrictive care, including inpatient hospitalization, juvenile justice detention, group home placement, and institutional care; and (3) Among CWS-involved children with MC, what diagnostic needs are most strongly associated with risk of entering restrictive care?

Methods: Using linked Medicaid, CWS, and juvenile justice data from Pennsylvania, we identified a sample of 73,985 children aged 1 to 17 involved in a CWS case between 2015 and 2019. The Pediatric Medical Complexity Algorithm (PMCA) was used to measure MC and identify the affected body systems (using Medicaid records in the three years preceding the CWS case). We employed multivariate logistic regression models to examine how MC was associated with restrictive care—including inpatient hospitalization, juvenile detention placement, institutional placement, and group home placement—in the year following a CWS case. Among children with MC, three-way associations between affected body system, age, and restrictive placement were examined.

Results: Twelve percent of children were identified as having MC, while an additional 16% were identified as non-complex but with one chronically affected body system. Among children with MC, mental health was the most commonly affected body system, identified in 55% of the group. Given this prevalence, we stratified MC by the presence of a mental health diagnosis when examining MC as a predictor for restrictive care. Relative to children with neither MC nor a mental health diagnosis, those with MC but no mental health diagnosis had nearly three times the odds of a restrictive care placement (odds ratio [OR] = 2.9, p < 0.001). When a mental health diagnosis was present, MC was associated with more than five times the odds of restrictive placement (OR = 5.4, p < 0.001). Finally, a key trend emerged: When body systems, age, and restrictive placement outcomes were examined together, the prevalence of mental health conditions among children in restrictive placements increased significantly with age.

Conclusions and Implications: Complexity as defined by the PMCA is significantly related to placement outcomes. Our findings emphasize the need for expanded mental health treatment options and specialized planning to avoid congregate and/or restrictive placements for vulnerable youth.