The association between cold weather and mortality among populations ever and never experiencing homelessness in British Columbia, Canada

OBJECTIVES: To assess the association between cold temperature and mortality among people who had ever experienced homelessness and those who had not in British Columbia, Canada. METHODS: Using administrative health datasets, we identified a cohort of individuals who ever experienced homelessness (EH) between 2013 and 2022. We compared all-cause mortality in this cohort with two cohorts from the same geographic region who had never experienced homelessness (NH): one matched on age and sex only (NH-loose), and another additionally matched on income assistance and underlying health conditions (NH-strict). We examined the association between weekly minimum temperature and risks of all-cause and cause-specific mortality across the three cohorts. RESULTS: All-cause mortality in the EH cohort was twofold higher than in the NH-strict cohort and sevenfold higher than in the NH-loose cohort. Mortality risk in the EH cohort increased significantly when temperatures fell below 0 °C in both the temperate Coastal Region and the colder Inland Region. Although mortality also increased in the NH cohorts, in the Inland Region the temperature threshold associated with a comparable increase in risk was lower than that observed in the EH cohort. CONCLUSIONS: Higher mortality in the EH cohort relative to the closely matched NH-strict cohort suggests that poorer health outcomes among people experiencing homelessness cannot be fully explained by underlying health conditions or socioeconomic disadvantage. The consistent elevation in risk below 0 °C across regions highlights the need to re-evaluate temperature thresholds used in local emergency response plans to better protect marginalized populations. These findings support evidence-based cold weather response planning and upstream policies that address housing instability and homelessness as key determinants of health.