Eleven-Month Arrest Outcomes Among Three Crisis Response Models in Michigan

Objective: Mobile crisis and co-response models are increasingly popular mental health crisis response alternatives to law enforcement. In theory, crisis response models could increase behavioral health treatment connections and divert people with mental illnesses from jails. This exploratory study compares post-year arrests of individuals who received services from three crisis response models (co-response, mobile crisis, and office-based crisis) against individuals who received law enforcement-only responses in five communities. Methods: Medicaid billing data and clinician service records identified (n = 474) crisis service recipients of two co-response models, two mobile crisis models, and one office-based crisis model in Michigan. Researchers collected law enforcement reports (n = 690) of mental health crises in the respective jurisdictions. Michigan State Police data revealed subject-level pre- and post-year arrests. Inverse probability of treatment weighting accounted for differences between crisis response and law enforcement groups by age, sex, prior-year arrest, and jurisdiction. A weighted Poisson regression examined each crisis response model's impact on post-year arrest rate relative to law enforcement-only responses. Results: Bivariate analyses supported the inclusion of response types in a regression. All three crisis models showed fewer post-year arrests than their respective law enforcement-only comparisons, but only the mobile crisis response predicted a statistically significant reduced incidence rate of arrest (incidence rate ratio 0.548; p < 0.05). Conclusions: This study is the first to assess post-year arrest outcomes of mobile crisis teams and adds to the null arrest literature of co-response teams. Mobile crisis teams may divert future crisis service pathways away from 911, decreasing law enforcement involvement and arrest risk.