Outreach Models for Substance Use Disorder in Homeless Populations: A Systematic Review and Network Meta-Analysis

OBJECTIVE: The relative effectiveness of outreach models used to address substance use disorder among homeless populations remains unclear. This systematic review and network meta-analysis aimed to compare the efficacy of various outreach interventions in improving substance use outcomes among homeless populations. METHODS: A systematic search of studies (published January 1990-January 2024) in the PubMed, Embase, PsycInfo, and Cochrane Library databases was conducted. A network meta-analysis was executed to compare outcomes of substance use frequency (self-reported days of substance use across a specified time) and severity (as determined by validated assessments) at the 1- and 2-year follow-ups. Treatment effects were standardized by using mean differences, and interventions were ranked by using P scores. RESULTS: Of the 716 studies screened, 35 studies met the inclusion criteria. The community reinforcement approach (CRA) demonstrated the highest effectiveness across the follow-up points (P scores=0.81-0.83), showing the strongest effect in reducing substance use (standardized mean difference=-0.28, 95% CI=-0.66, 0.10) among homeless populations. Combining assertive community treatment with intensive case management showed marked improvement at the 2-year follow-up (P score=0.79), whereas critical time intervention maintained strong performance across the follow-ups. Substantial heterogeneity was observed across the analyses (I²=76.3%-84.7%). CONCLUSIONS: Comprehensive and integrated approaches, such as CRA and combined interventions, may most effectively address substance use disorders among homeless populations. Although substantial heterogeneity in the results indicated varying effectiveness of the interventions, the findings supported the implementation of multifaceted interventions that address individuals' immediate and long-term recovery needs and provide robust and reliable support systems.