Police-mental health co-response versus police-as-usual response to behavioral health emergencies

Abstract

Background: People with mental illness are overrepresented in United States (US) criminal legal systems. In response, alternatives to traditional police response to behavioral health emergencies have become more common, despite limited evidence for their effectiveness. We conducted the first randomized controlled trial of a police-mental health co-response team to determine program effectiveness relative to a police-as-usual response on key outcomes identified by community stakeholders. Methods: Between January 2020 and March 2021, we randomized behavioral health emergency calls for service in one of six police districts in Indianapolis, Indiana to receive a co-response or police-as-usual response during operational hours between 10 a.m. and 5 p.m. Mondays - Fridays. Eligible calls for service were determined via pre-specified phrases indicating a behavioral health incident over the police dispatch radio. Researchers then communicated random assignment with the co-response team to indicate whether they should respond or withhold. Logistic and negative binomial regression were used to assess group differences in emergency medical services (EMS) events within 12 months of the randomized incident along with jail booking, outpatient encounters, and emergency department visits. Findings: We randomized 686 calls for service with co-response completed in 264 cases and police-as-usual response in 267 cases. The overall rate of attrition was similar across conditions and the final sample included 211 co-responses and 224 police-as-usual responses. We found no significant differences in any EMS event (odds ratio [OR]: 1.26; 95% confidence interval [CI]: 0.85-1.88, p = .246) or event counts (incidence rate ratio [IRR]: 0.85; 95% CI: 0.52-1.37, p = .504). We also found no differences in secondary outcomes (jail booking, outpatient encounters, and emergency department visits). Discussion: A police-mental health co-response team model was not more effective than traditional police response on key outcomes. Co-response team models, such as the one reported here, may unintentionally foster emergency services utilization among persons with behavioral health needs. Without a functioning national mental health system, communities in the US will continue to struggle to identify solutions to meet the needs of community members with complex behavioral health issues.