P106: Healthcare utilization among homeless and/or substance using adults presenting to the ED

Abstract

Introduction: Substance use and unstable housing are associated with heavy use of the Emergency Department (ED). This study examined the impact of substance use and unstable housing on the probability of future ED use. Methods: Case-control study of patients presenting to an urban ED. Patients were eligible if they were unstably housed for the past 30 days, and/or if their chief complaint was related to substance use. Following written informed consent, patients completed a baseline survey and health care use was tracked via electronic medical records for the next six months. Controls were enrolled in a 1:4 ratio. More than 2 ED visits during the follow-up was pre-specified as a measure of excess ED use. Descriptive analyses included proportions and medians with interquartile ranges (IQR). Binomial logistic regression models were used to estimate the impact of housing status, high-risk alcohol use (AUDIT) and drug use (DUDIT), and combinations of these factors on subsequent acute care system contacts (ED visits + admissions). We controlled for age, gender, comorbidities at baseline, and baseline presenting acuity. Results: 41 controls, 46 substance using, 91 unstably housed, and 31 both unstably housed and substance using patients were enrolled (n = 209). Median ED visits during follow up were 0 (IQR: 0-1.0) for controls, 1.0 (IQR: 0-3.3) for substance using, 1.0 (IQR: 0-4.0) for unstably housed and 4 (IQR: 2-12.3) for unstably housed and substance using patients. The median acute care system contacts over the same period was 1.0 (IQR 0-2.0) for controls, 1.0 (IQR: 0-4.0) for substance using, 1.0 (IQR: 0-5.0) for unstably housed and 4.5 (IQR: 2.8-14.3) for unstably housed and substance using patients. Being unstably housed was the factor most strongly associated with having > 2 ED visits (b=3.288, p<0.005) followed by high-risk alcohol and drug use (b=2.149, p<0.08); high risk alcohol use alone was not significantly associated with ED visits (b=1.939, p<0.1). The number of comorbidities present at baseline was a small but statistically significant additional risk factor (b=0.478, p<0.05). The model correctly predicted 70.1% of patients’ ED utilization status. Conclusion: Unstable housing is a substantial risk factor for ED use; high-risk alcohol and drug use, and comorbidities at baseline increased this risk. The intensity of excess ED use was greatest in patients who were unstably housed and substance using.