Skip to main content
NCBI home page
JAMA Network logoLink to JAMA Network
. 2026 Oct 2;9(10):e2637190. doi: 10.1001/jamanetworkopen.2026.37190

Disparities in Heat-Related Injuries Among People Experiencing Homelessness

Lucie Richard 1,✉, Luis Palma 2, Stephen W Hwang 1,3,4, Refik Saskin 2
PMCID: PMC13633527  PMID: 42826024

Abstract

This cohort study examines whether people experiencing homelessness in Ontario, Canada, have higher rates of heat-related injury than people not experiencing homelessness.

Introduction

Climate change is rapidly increasing the frequency and intensity of extreme heat events, posing a significant public health threat.1 While the health impacts of extreme heat are well-documented in the general population,2 the disproportionate impact on people experiencing homelessness in Canada is unknown, despite this population facing disproportionate risks due to structural vulnerabilities, such as lack of access to climate-controlled environments and greater underlying health conditions that impair thermoregulation or heat stress awareness. In this study, we assess disparities in heat-related injury seen in emergency departments (EDs) between people experiencing homelessness (PEH) and a frequency-matched cohort of people not experiencing homelessness (PNEH) across Ontario.

Methods

This retrospective population-based cohort study used linked administrative health databases in Ontario, Canada, from April 1, 2018, to March 31, 2025. Datasets were linked using unique encoded identifiers and analyzed at ICES, an independent, nonprofit research institute whose legal status under Ontario’s health privacy legislation allows it to analyze health care and demographic data without individual consent. Use of data was authorized under section 45 of this legislation, for which reason ethics and informed consent was not required. We followed STROBE reporting guidelines.

We identified PEH using a validated case definition3 from the sample of patients 16 years or older with at least 1 ED visit during the study period. To ensure the strong age and sex skew in Ontario’s homeless population did not confound results, PNEH controls were frequency-matched 1:1 to PEH within each fiscal year on age distribution, sex distribution, and region distribution. We identified ED visits between April 1 and September 30 having any ICD-10-CA diagnosis code for heat-related injury, including heatstroke or sunstroke, heat syncope, heat exhaustion or fatigue, heat edema, and other effects of excessive natural heat occurring with evidence of exposure to heat (codes T670-T677, T678, or T679 if paired with e-code X30 or X32).

Each fiscal year, we calculated the rate of heat-related injury visits as number of heat-related ED visits over the number of PEH or PNEH (respectively) identified that fiscal year, per 100 000. Annual and pooled rate ratios with 95% CIs indicating significance at the .05 level were estimated using a Poisson distribution. Owing to outcome rarity, only pooled rate ratios and 95% CIs were estimated for the City of Toronto subsample. Provincial rate ratios were calculated using the same method (events per month over annual population at risk) and plotted graphically. All analyses were conducted at ICES using SAS Enterprise Guide 7.1 (SAS Institute).

Results

Each group included 144 527 individuals, with 98 887 (68.4%) males, 18 190 (12.6%) youths aged 16 to 24 years, 72 569 (50.2%) adults aged 25 to 44 years, 24 085 (16.7%) adults aged 45 to 54 years, and 29 683 (20.5%) adults 55 years and older. Overall, 351 (0.03%) and 101 (0.04%) heat-related injury visits were identified for PEH and PNEH, respectively. Annual rates among PEH were consistently higher than among PNEH, ranging from 2.65 (95% CI, 1.40-5.03) to 4.55 (95% CI, 2.49-8.30) times higher, with a pooled rate ratio of 3.50 (95% CI, 2.81-4.37). Toronto’s pooled rate was nearly double the provincial average (6.14 [95% CI, 3.14-12.02]) (Table).

Table. Pooled and Annual Rate of Heat-Related Injury per 100 000 Population Among PEH and Matched PNEH, Ontario and Toronto Subsample, 2018 to 2025a.

Fiscal year Heat injury rate per 100 000 (95% CI)
PEH PNEH Rate ratio, PEH vs PNEHb
Ontario
2018-2019 29.3 (22.3-37.9) 6.4 (3.4-11.0) 4.55 (2.5-8.3)
2019-2020 15.1 (10.5-21.1) 5.7 (3.0-9.7) 2.65 (1.4-5.0)
2020-2021 28.5 (21.8-36.6) 8.4 (5.0-13.2) 3.41 (2.0-5.8)
2021-2022 17.3 (12.5-23.4) 5.3 (2.8-9.1) 3.24 (1.7-6.0)
2022-2023 17.0 (12.4-22.9) 5.0 (2.7-8.6) 3.40 (1.8-6.3)
2023-2024 19.0 (14.2-24.7) 4.2 (2.2-7.3) 4.53 (2.4-8.5)
2024-2025 19.5 (14.8-25.2) 6.4 (3.8-9.9) 3.07 (1.8-5.2)
Pooled rate ratio
Ontario NA NA 3.50 (2.8-4.4)
Toronto subsample NA NA 6.14 (3.1-12.0)
Open in a new tab

Abbreviations: NA, not applicable; PEH, patients experiencing homelessness; PNEH, patients not experiencing homelessness.

a

PNEH were frequency matched 1:1 to the PEH group by age, sex, and health region. Fiscal years run from April 1 to March 31 the following year. Numerators were calculated between April 1 and September 30, while denominators were calculated across the full fiscal year.

b

Rate ratios and 95% CIs were estimated following a Poisson distribution.

Monthly rates for PEH were also consistently higher than for PNEH (Figure). The difference between groups was statistically significant during only 3 months: July 2018 (PEH: 180.7 [95% CI, 121.9-257.9] vs PNEH: 35.7 [95% CI, 13.1-77.7]), July 2020 (PEH: 188.6 [95% CI, 130.6-263.5] vs PNEH: 55.7 [95% CI, 26.7-102.4]), and July 2024 (PEH: 80.1 [95% CI, 49.0-123.8] vs PNEH: 20.1 [95% CI, 6.5-46.9]).

Figure. Line Graph of Monthly Rates of Heat-Related Injury per 100 000 Population for People Experiencing Homelessness (PEH) and People Not Experiencing Homelessness (PNEH).

Line chart of monthly injury rates by year for P E H and P N E H. Single-panel line graph with two time-series. Vertical axis label at left: Monthly rate of heat-related injuries; tick marks and gridlines at 0, 100, 200, and 300. Horizontal axis: months repeated across years, labeled May, Jul, Sep, Nov, Jan, and Mar, with year labels beneath: 2018, 2019, 2020, 2021, 2022, 2023, 2024, and 2025. Legend in the upper right titled Group with two entries: P E H as a dark teal line and P N E H as an orange line. The dark teal P E H line has seasonal peaks each year, with values near zero in winter months and higher values from late spring through summer. In 2018, P E H rises from near 0 in May to about 60 in June, peaks near 180 in July, then declines to about 35 in August and about 25 in September. In 2019, P E H increases from near 0 in May to about 15 in June, about 40 in July, peaks near 70 in August, then falls to about 45 in September and about 10 in October. In 2020, P E H increases from near 0 in May to about 45 in June and about 70 in July, peaks near 190 in August, then drops to about 30 in September and near 0 by October. In 2021, P E H rises to about 55 in June and about 60 in August, then declines to about 15 in September. In 2022, P E H rises to about 25 in May and about 65 in July, then declines to about 50 in August and near 0 by September. In 2023, P E H rises from about 10 in April to about 25 in May, about 60 in July, peaks near 75 in August, then drops to about 10 in September and about 30 in October. In 2024, P E H rises from near 0 in April to about 30 in May, about 70 in July, peaks near 80 in August, then declines to about 45 in September and near 5 in October. The orange P N E H line remains lower throughout, generally near zero in winter and peaking in summer: about 35 in July 2018, about 30 in August 2019, about 55 in August 2020, about 25 in June 2021, about 22 in June 2022, about 25 in August 2023, and about 30 in July 2024.

Open in a new tab

Discussion

Between 2018 and 2025, PEH had 3.5 times higher ED-presenting heat-related injuries compared with PNEH across Ontario; Toronto had nearly double this disparity. Monthly comparisons also highlighted 3 months with substantial difference between groups, all corresponding to major heat events in Ontario (July 2018: the 2018 North American heat wave that resulted in more than 200 excess deaths in Québec4 and saw humidex peaking at 50 °C [122 °F] in Toronto5; July 2020: the “hottest July ever recorded” in North America,6 with nearly half the month at extreme temperatures; and July 2024: once again, the hottest July ever recorded by the NOAA7). These findings demonstrate that even within a universal health care system, the structural vulnerabilities associated with homelessness yield substantial heat-related morbidity, corroborating and expanding upon previous adjusted estimates from the United States.8,9

Study limitations include the insensitivity of heat-injury diagnostic codes10 relative to all conditions exacerbated by extreme heat. Furthermore, linked administrative data cannot identify patients without valid health card numbers. Finally, this design cannot causally attribute disparities to housing status or address other potential sources of confounding, such as comorbidities or medications that impair thermoregulation.

As the climate continues to warm, particularly in northern regions such as Ontario,1 our findings underscore the importance of heat-response strategies tailored to the needs of PEH. In the short term, these include dedicated homeless-serving cooling infrastructure and other emergency management strategies and, in the longer term, improving stock of climate-controlled housing with appropriate supports.

Supplement.

Data Sharing Statement

jamanetwopen-e2637190-s001.pdf (16.5KB, pdf)

References

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplement.

Data Sharing Statement

jamanetwopen-e2637190-s001.pdf (16.5KB, pdf)

Articles from JAMA Network Open are provided here courtesy of American Medical Association

RESOURCES