UCLA

Homelessness affects millions of Americans each year, and many homeless individuals have complex healthcare needs that place high demands on the United States healthcare system. This dissertation explores adverse health services utilization outcomes of the homeless population using the Healthcare Cost and Utilization Project’s 2014 State Inpatient Database and the State Emergency Department Database from Florida, Maryland, Massachusetts, and New York.

The first paper, “Association between homelessness and opioid overdose and opioid-related hospital admissions/emergency department visits,” examines opioid overdose and opioid-related hospitalization/emergency department (ED) visit risks of homeless patients compared to a low-income housed comparison group. The study used multivariable linear probability models with hospital fixed effects for dichotomous outcomes. Outcomes were also stratified by sex and race/ethnicity. The study found that homeless patients had substantially higher risks for both outcomes, and the non-Hispanic white female homeless patients were particularly the highest risk group in this patient population. Implementing screenings for homelessness and opioid abuse may be critical for curbing the opioid epidemic in this population.

The second paper, “Frequent emergency department use among homeless individuals: High risk of opioid-related diagnoses and adverse health services utilization outcomes,” compares opioid outcomes, mechanical ventilation, mortality, and hospitalizations of homeless patients who had 4 or more ED visits, 2-3 ED visits and 1 ED visit in 2014, and identifies predictors for higher rates of ED use. Multivariable linear probability models with hospital fixed effects were used for the main analyses and a negative binomial regression model with hospital fixed effects was used for predicting higher rates of ED use. The study revealed that homeless patients who are high ED users were more likely to be hospitalized and have other adverse outcomes. These findings encourage targeted interventions for the high-utilizer homeless population to reduce the burden of serious outcomes and costs for the patient and society.

The third paper, “Association between homelessness and delivery hospitalization outcomes: a multi-state population-based study,” explores delivery hospitalization outcomes of pregnant homeless versus non-homeless women. This project used inpatient data from FL, MA and NY and overlap propensity-score weighing method and regression adjustment. Compared to non-homeless women treated within the same facility, homeless women had higher likelihoods of experiencing placental abnormalities, preterm labor and higher delivery hospitalization costs. A large majority of homeless women were treated in government-owned safety-net hospitals with lower average delivery costs compared to non-homeless women, who were mainly treated in not-for-profit hospitals. These findings highlight the importance of screening pregnant women for social needs, including homelessness, as well as developing policies that encourage partnerships between healthcare providers and community resources, such as local social housing programs.

Taken together, all three papers highlight the adverse health outcomes of the homeless population and the need to quickly identify homeless patients and refer them to appropriate care.