Impact of supportive housing on substance use-related health care utilization among homeless persons who are active substance users

Ann Epidemiol

Purpose: Homeless persons with a substance use disorder (SUD) are at high risk of emergency department (ED) visits and hospitalizations. We evaluated the impact of supportive housing on SUD-related ED visits and hospitalizations among active substance users experiencing chronic homelessness.

Methods: We matched 1558 homeless adults eligible for a New York City supportive housing program who had a SUD (2007-2012) to Medicaid claims data. We examined SUD-related hospitalizations and ED visits 2 years posteligibility. We calculated stabilized inverse probability of treatment weights using baseline demographic and clinical characteristics and used doubly robust estimators to compare rates between persons placed into supportive housing and those eligible but not placed. We also examined outpatient SUD treatment.

Results: Placed persons were hospitalized and visited EDs for substance use at significantly lower rates than persons not placed into housing (adjusted rate ratio, 0.70; 95% confidence interval, 0.56-0.88 and adjusted rate ratio, 0.46; 95% confidence interval, 0.31-0.68, respectively) 2 years posteligibility. Placed persons had increased odds of a new episode of SUD treatment and initiating treatment.

Conclusions: Supportive housing placement was associated with decreases in SUD-related hospitalizations and ED visits and with increased odds of new episodes of SUD outpatient treatment among homeless persons with SUD.

Miller-Archie SA, Walters SC, Singh TP, Lim S. Impact of supportive housing on substance use-related health care utilization among homeless persons who are active substance users. Ann Epidemiol. 2019. Epub ahead of print. PMID: 30846275. DOI: 10.1016/j.annepidem.2019.02.002.

Publication Year: 
2019
Resource Type: 
Peer Reviewed Research
Study Design: 
Other Study Design
Population: 
Complex Patients
Homeless
Outcomes: 
Utilization
Screening Research: 
No