People experiencing homelessness (PEH) attend emergency departments, often presenting with complex needs. The relationship between housing status and access to mental healthcare is not well established.

To compare characteristics, clinical severity and admission rates by housing status, and to examine whether homelessness is associated with admission after adjusting for severity.

We conducted a retrospective cohort study of patients referred to psychiatry at an inner-city emergency department between 1 January and 31 December 2020. Housing status was recorded at presentation. Severity was rated using the Threshold Assessment Grid (TAG, range 0–24). The primary outcome was hospital admission. Group differences were examined using t-tests and χ2 tests. Multivariable logistic regression models assessed variables associated with admission, adjusting for TAG total, age and gender, and then for individual TAG domains.

Of 771 referrals, 282 (36.6%) were PEH. PEH had higher severity (mean TAG 12.3 v. 8.1, p < 0.001) but lower admission rates (10.6 v. 18.0%, p = 0.006). All TAG domains were elevated among PEH, with the largest differences in survival, risk from others and social functioning. Each one-point TAG increase was associated with higher odds of admission (adjusted odds ratio (aOR) 1.34, p < 0.001), as was older age (aOR 1.04 per year, p < 0.001). PEH had lower odds of admission (aOR 0.16, p < 0.001), which persisted following domain-level adjustment (aOR 0.45, p = 0.039).

Despite greater severity, admission rates among PEH were lower following emergency psychiatric assessment. These findings highlight the need for better understanding of how emergency mental health services respond to the complex needs of this population.