Latent Class Analysis of Service Utilization Patterns Among Reentering Criminal Justice-Involved Adults: Longitudinal Mixed Effects Model with Psychological Distress

Background and Purpose: More than 95 percent of incarcerated individuals will be released back into their communities, with an estimated 650,000 individuals reentering from prisons each year. As a result, post-incarceration reentry has seen a considerable increase in scholarly and policy attention. However, holistic services remain a pressing need. Additionally, few studies have utilized person-centered methods, which generate participant profiles based on lived experiences, to examine service utilization among justice-involved adults during reentry. The current study establishes latent classes of service utilization among reentering justice-involved adults and examines associations with subsequent levels of psychological distress using panel data.

Methods: Data includes four waves from Phase 1 of the 5-Key Model for Reentry (N = 536), collected from 2018-2019 as a multisite randomized controlled trial of reentering justice-involved adults from 50 prisons across four states: Florida, Kentucky, Pennsylvania, and Texas. Data were collected after release from incarceration and then 4, 8, and 15 months after release. Service utilization was measured through service receipt across nine domains: life skills, mental health, substance use, relationships, employment, education, physical health, housing, and cognitive skills. Psychological distress was measured through a ten-item version of the Kessler Psychological Distress Scale. We controlled for current mental health disorder, substance use disorder, or co-occurring disorder as well as sociodemographic factors including race/ethnicity, gender, age, income, and state. Latent class analysis (LCA) was used to identify distinct profiles based on reported patterns of service utilization after release. Following, mixed-effects regression models examined whether trajectories of psychological distress differed among service classes.

Results: Examination of LCA fit indices show a three-class model provided the best fit, revealing: a high service use class (9.89%), a low service use class (80.04%), and a high substance use service class (10.07%). The mixed-effects regression model showed a significant decline in psychological distress at waves 2 (b= -3.15, p<.01) and 3 (b= -2.52, p<.05), and that service utilization profiles were associated with longitudinal differences in distress. Compared to those in the high service class, those in the low service class reported significantly lower levels of psychological distress initially (b= -2.06, p<.05), but over time, both classes with higher service utilization observed significantly steeper declines in distress. The low service class revealed more constant, and slightly increasing, levels of distress over time.

Conclusions and Implications: Person-centered methods are an underutilized approach with justice-involved populations, which can be used to better identify heterogeneous outcomes based on lived experiences. Results reveal different patterns of initial service utilization, and that those with higher use patterns experience sharper declines in psychological distress over time. Future reentry research should examine mechanisms that match returning citizens to services and reduce barriers to service access, particularly among historically marginalized communities with lower rates of access. Social workers may benefit from this work as substance use service needs remain high among this population. Furthermore, this research emphasizes the significance of providing additional education and inclusion in the reentry process to better address needs, and highlights how service utilization may positively impact long-term psychological outcomes.