Improving Identification of Patients Experiencing Homelessness in the Electronic Health Record: A Curated Registry...

Identification of persons experiencing homelessness (PEH) within healthcare systems is critical to facilitate patient and population-level interventions to address health inequities.

We created an enhanced electronic health record (EHR) registry to improve identification of PEH within a safety net healthcare system.

We compared patients identified as experiencing homelessness in 2021, stratified by method of identification (i.e., through registration data sources versus through new EHR registry criteria).

Sociodemographic and clinical characteristics, healthcare utilization, engagement with homeless service providers, and mortality.

In total, 10,896 patients met the registry definition of a PEH; 30% more than identified through standard registration processes; 78% were identified through only one data source. Compared with those identified only through registration data, PEH identified through new registry criteria were more likely to be female (42% vs. 25%, p < 0.001), Hispanic/Latinx or Black/African American (30% versus 25% and 25% vs. 18%, p < 0.0001), and Medicaid or Medicare beneficiaries (74% vs. 67% and 16% vs.10%, respectively, p < 0.0001). New data sources also identified a higher proportion of patients: at extremes of age (16% < 18 years and 9% ≥ 65 years vs. 2% and 5%, respectively, p < 0.0001), with increased clinical risk (31% with CRG 6-9 vs. 18%, p < 0.0001), and with a mental health diagnosis (56% vs. 42%, p < 0.0001), and a lower proportion of patients with a substance use diagnosis (39% vs. 54%, p < 0.0001) or criminal justice involvement (8% vs. 15%, p < 0.0001). Newly identified patients were more likely to be engaged in primary care (OR 2.03, 95% CI 1.83-2.26) but less likely to be engaged with homeless service providers (OR 0.70, 95% CI 0.63-0.77).

Commonly utilized methods of identifying PEH within healthcare systems may underestimate the population and introduce reporting biases. Recognizing alternate identification methods may more comprehensively and inclusively identify PEH for intervention.