State-level homelessness and drug overdose mortality: Evidence from US panel data

Although homelessness is a well-documented risk factor for drug overdose at the individual level, less is known about state-level homelessness and overdose mortality in the United States (US). This study used 2007–2020 panel data for all US states and the District of Columbia, from the following sources: US Department of Housing and Urban Development (homelessness data); Centers for Disease Control and Prevention (drug overdose death counts, population estimates, and opioid prescribing rates); Bureau of Labor Statistics (unemployment rates); and the National Forensic Laboratory Information System (drug seizure data). Two-way (state and year) fixed effects models regressed log-transformed drug overdose mortality rates on homelessness prevalence, in nested models adding demographic composition and unemployment measures, as well as drug supply measures. Models were weighted by state population size, and standard errors (SEs) were clustered at the state level. Homelessness prevalence was significantly and positively associated with rates of drug overdose mortality after adjusting for nationwide trends, time-invariant differences between states, demographic composition, and unemployment rates (b[SE]=0.98[0.36], p=0.009). The positive association between homelessness prevalence and overdose mortality was attenuated at higher levels of fentanyl availability (fentanyl involvement in drug seizures; interaction term b[SE]=−0.02[0.01], p=0.001). The positive association between state-level homelessness and drug overdose mortality suggests that policies and programs to prevent and reduce homelessness represent fundamental elements of a comprehensive response to the US overdose crisis. •Positive association between state rates of homelessness and overdose mortality.•Association persisted after adjusting for demographic and unemployment measures.•Association attenuated at higher levels of fentanyl involvement in drug seizures.