uniteTOlove.ca
This is a living research library. Claims carry verification statuses. A documented set of about 234 findings, where a derived page carries a source's claim without its original hedge, is known and queued for remediation, so read any affected line alongside its source backgrounder and check the corrections log for status. Found a mistake? Let us know.

Mortality and Overdose Data for People Experiencing Homelessness in Toronto — Backgrounder

DRAFT Research backgrounder

Briefs for this backgrounder

Status: DRAFT (crosses opioid-crisis-public-health-response / A4 and homelessness-encampments / C2★ issue index rows) (page under Domain H — Homelessness & Housing-First Deep Dive; this is the owning page for shelter-death/mortality statistics corpus-wide — other documents in this project should cite this backgrounder rather than restate mortality figures independently) · Nearest this library's issue index row: A4 opioid-crisis-public-health-response (public health unit + provincial funding/SCS approvals) and C2★ homelessness-encampments (shelter capacity/PiT). Version: v1.0 · Date: 2026-07-14 · What this page draws on: carried-forward (carried forward from this page’s own sources docs, a formally registered claim citations as recorded) and New load-bearing findings (this review's live-discovery findings, inline source quotes, not yet through this library’s formal verification process). Claim coverage as of 2026-07-14: 2 carried-forward documents (this library's prior synthesis document (Mortality Overdose Data), homelessness-mortality-overdose-data.md [csrl2 variant]) citing CL-015, CL-792, CL-791, CL-783, CL-029, CL-424, CL-300, CL-474, CL-90683, CL-698 as recorded in those documents — checked directly against this library's internal records register/the claims register this review: none of these IDs currently resolve (no match), so they are cited here at the same carried-forward, not-independently-claims register-confirmed status the master briefing documents themselves carry. 9 new 2026 primary-source findings from this review's live discovery (NEW-2026-MORT-1 through NEW-2026-MORT-9), each with an inline source quote, not yet through this library’s formal verification process. Coverage: breadth not formally checked in this review.

Written per this library's standard page structure, a later review, 2026-07-14. This document's original contribution is a live re-verification of the page’s most load-bearing figures — the cumulative shelter-death total, the 2025 annual figure, and the current opioid-overdose trend — directly against the City's and province's own current primary sources, because this page is corpus-wide load-bearing and other documents (including shelter-system-capacity-strain.md, via CL-793) already cite figures from it.

Scope

This page’s neutral scope question: what do Toronto's and Ontario's own mortality and overdose-surveillance datasets show about deaths of people experiencing homelessness and shelter residents specifically, and what is the current (2025-2026) trend? This document covers: TSSS's Deaths of Shelter Residents dataset (since 2007); Toronto Public Health's broader Deaths of People Experiencing Homelessness dashboard; the City's shelter-specific fatal/non-fatal opioid overdose surveillance; the Ontario Office of the Chief Coroner's province-wide opioid toxicity mortality data, including the homeless-specific subset; the international Housing First mortality-outcomes literature; and gendered/age-stratified mortality findings from peer-reviewed academic sources. It hands off, rather than duplicates: general Housing First program design and cost-effectiveness (homelessness-housing-first-model); shelter capacity and occupancy generally (shelter-system-capacity-strain); encampment-specific fire-death data (flagged as a genuinely distinct, not-yet-covered category below rather than absorbed into this document's main scope). This document is the corpus-wide canonical statement of Toronto shelter-death and homelessness-mortality figures — other documents citing a shelter-death or overdose figure should cite this backgrounder rather than restate the underlying data independently.

Current state

The headline shelter-resident death figures: a live discrepancy found and flagged, not silently resolved

This is the single most important finding of this review. This page’s carried-forward master briefing (this library's prior synthesis, dated July 2, 2026) states the cumulative shelter-resident death total since 2007 as 825, based on the City's 2024 Annual Review. A direct live fetch this review of the City's current page — now titled the 2025 Annual Review of Statistics on Deaths of Shelter Residents, page date modified April 28, 2026 — gives a different cumulative figure:

"Since 2007, there have been 884 total shelter resident deaths reported to TSSS, an average of 49 deaths per year... 59 deaths occurred in 2025, which remains unchanged from 2024."

Source: City of Toronto, "2025 Annual Review of Statistics on Deaths of Shelter Residents," https://www.toronto.ca/city-government/data-research-maps/research-reports/housing-and-homelessness-research-and-reports/deaths-of-shelter-residents/annual-review-of-statistics-on-deaths-of-shelter-residents/, page date modified 2026-04-28. Directly fetched and read 2026-07-14. [NEW-2026-MORT-1]

This confirms the figure CL-793 (cited by shelter-system-capacity-strain.md's backgrounder) already carries — "59 shelter-resident deaths in 2025, cumulative 884 since 2007" — is the City's current, live, correct figure. The page’s own carried-forward master briefing's 825 figure is now stale: it reflects the 2024 Annual Review (2007-2024 cumulative), not the current 2025 Annual Review (2007-2025 cumulative). The arithmetic reconciles cleanly: 825 (2007-2024) + 59 (2025) = 884 (2007-2025) — this is not a data revision or correction, it is simply one additional year's deaths added to the running cumulative total, and this backgrounder treats "884, cumulative through 2025" as the current correct figure going forward, per the Deep-20 corpus's own citation. Any document still citing "825 total since 2007" is citing the 2024 Annual Review's cumulative figure, one year stale as of this review — not wrong for its own stated period, but superseded by the current 2025 Annual Review.

The same primary source gives additional current detail not in the carried-forward master briefing:

"The shelter system provided nearly 44% more bed nights in 2025 as in 2021, so while deaths in 2025 were 55% lower than 2021 totals, when allowing for the larger service population, they were 68% lower... Males make up 60% of the shelter population, but 75% of decedents for 2021-2025... Of the 884 deaths of shelter residents reported to TSSS since 2007: The average number of deaths per year has been 49. The majority of decedents have identified as cis gendered male: 684 or 77%. The average (mean) age at death across all genders has been 52 years. 442 deaths (or 50%) occurred within a shelter/respite facility whereas 442 deaths (or 50%) occurred off-site."

Source: same as NEW-2026-MORT-1. [NEW-2026-MORT-2]

The pre-2020 shelter-resident death trend, and average age at death by year

This page’s inherited master briefing a year-by-year shelter-resident death series back to 2007 (24, 26, 16, 21, 21, 18, 16, 30, 45, 33, 35, 26, 48 for 2007-2019 respectively), continuing 74 (2020), 132 (2021), 110 (2022), 91 (2023), 59 (2024), and 59 (2025), with a partial-year 2026 figure (12 deaths, January-July, not comparable to full-year totals) [From this library’s earlier research from this library's prior synthesis document (Mortality Overdose Data)]. A pre-COVID pattern worth naming, per the master briefing's own analysis: 2015 (45) and 2019 (48) were already elevated years relative to the 2007-2014 average (~21/year) — the COVID-era spike (74 in 2020, 132 in 2021) was a sharp acceleration of an already-rising trend, not a sudden departure from an otherwise-flat baseline [From this library’s earlier research from this library's prior synthesis document (Mortality Overdose Data)]. Every year in this series shows an average age of death between 47 and 58 — people dying of shelter-related causes in their late 40s to late 50s, decades below general Canadian life expectancy (roughly 82); 2021, the deadliest year on record, also shows one of the youngest average ages (47). Male decedents' average age is consistently close to the overall average (48-62 across the series); female decedents show more year-to-year volatility (37-64), and 2021's female-decedent count (31, average age 48) was itself more than double most pre-COVID years' entire decedent count, male and female combined [From this library’s earlier research from this library's prior synthesis document (Mortality Overdose Data)].

The broader TPH dashboard figures, 2019-2024: the wider count behind the shelter-specific total above

Distinct from the TSSS shelter-resident-specific figures above, Toronto Public Health's wider "Deaths of People Experiencing Homelessness" dashboard — which includes deaths on the street and at other locations, not only in shelters — recorded 128 deaths in 2019 (pre-pandemic) and 223 deaths in 2021 (an upward revision from an earlier, since-superseded figure of 216), of whom 132 were shelter residents specifically, a shelter-resident subset that fell to 110 in 2022 [From this library’s earlier research from this library's inherited source document (homelessness mortality and overdose data), original sourcing: Toronto Public Health news release, RES-742]. In February 2025, after integrating Ontario Office of the Chief Coroner data the earlier counts had not captured, TPH revised its 2022 death count from 189 to 331 (a 75% increase) and its 2023 count from 150 to 300 (a 100% increase) — a genuine retroactive upward revision, distinct in kind from the shelter-resident cumulative total's 825-to-884 update above, which is additive (one more year counted), not a correction of prior years [From this library’s earlier research from this library's inherited source document (homelessness mortality and overdose data) and this library's prior synthesis document (Mortality Overdose Data), original sourcing: TPH press release, February 2025, RES-0248]. For the first half of 2024, TPH recorded 135 deaths, with an interim median age at death of 50 for men and 36 for women, and found 54% of deaths involved drug toxicity [From this library’s earlier research from this library's inherited source document (homelessness mortality and overdose data), original sourcing: RES-0248/TPH dashboard]. This interim H1-2024 age-at-death figure (50/36) was itself superseded within this page’s own master briefing by a full-year-2024 figure (50/38, 55% drug toxicity) once complete-year data became available — the sibling page homelessness-gender-based-violence (verified directly this review) carries the full-year-2024 figure as its own canonical mortality statistic, sourced to the same underlying TPH dashboard and to Ali et al.'s 2026 BMC Public Health study. Canonical-figure note: two of this page’s own carried-forward master briefing (this library's prior synthesis document (Mortality Overdose Data) Part 3E, and this library's inherited source document (homelessness mortality and overdose data) itself) separately assert that the BMC Public Health study's own reported figure for women is 36, not 38 — but this conflicts with the homelessness-gender-based-violence page’s own primary master briefing, which states the same BMC study's full-year-2024 figure is 50/38 and explicitly frames 36 as a different, earlier, H1-2024 TPH interim estimate, not the BMC study's own number. This document treats 38 as the corpus-canonical figure for the BMC study's full-year finding (consistent with the sibling page, and with this same document's Part 2 text, which itself states the H1-2024 estimate of 50/36 was "superseded" by a full-year 50/38 figure) and treats the "36" attribution to the BMC study specifically, appearing in this page’s own Part 3E addendum, as an internal inconsistency within the inherited master briefing rather than a fact to restate at face value — the H1-2024 TPH interim figure of 36 is accurately reported above as what it is (an interim, since-superseded TPH estimate), not reintroduced as the BMC study's own figure.

Excess-mortality studies: two Toronto cohort studies, a generation apart

Two Toronto-based cohort studies, a generation apart, quantify excess mortality risk directly rather than only counting raw deaths. A 2009 BMJ study (Hwang et al.) followed a shelter cohort for 11 years and found overall mortality 3.5 times the rate of Canada's lowest income quintile — not the general population as a whole, a more conservative and arguably more meaningful comparison group [From this library’s earlier research from this library's inherited source document (homelessness mortality and overdose data), original sourcing: Hwang SW et al., BMJ, 2009, RES-0111]. An earlier 2000 JAMA study (also Hwang), stratified by age, found the excess risk was steepest for the youngest adults: 8.3 times the general-population rate for ages 18-24, compared with 3.7 times for ages 25-44 and 2.3 times for ages 45-64 [From this library’s earlier research from this library's inherited source document (homelessness mortality and overdose data), original sourcing: Hwang SW, JAMA, 2000, RES-0069]. Read together with this document's age-at-death figures above, these two studies show the mortality gap is not just a difference in median age at death, but a steep, age-graded excess-risk multiplier that is worst for the youngest homeless adults specifically.

2021 historical overdose-rate comparison (journalist's own calculation, not an official City statistic)

A 2022 journalism report (tdotcommunity.ca), citing the same underlying City dataset this document otherwise draws on, documented 1,563 total overdose incidents in shelter settings in 2021 — an 83% increase from 2020 — of which 67 were fatal (up from 46 in 2020). The same reporting calculated a comparative rate of 2,090 overdoses per 10,000 people experiencing homelessness, versus 10 per 10,000 in the general population — roughly 209 times the general-population rate ⚠️ Still being checked: the rate comparison is the journalist's own calculation, not an official City statistic; the underlying 1,563/67 figures are attributed to City data [From this library’s earlier research from this library's inherited source document (homelessness mortality and overdose data), original sourcing: tdotcommunity.ca, April 19 2022]. The same reporting cites understaffing and training gaps as a contributing factor, per Lorraine Lam of the Shelter and Housing Justice Network.

2021 Homes First/Dixon Hall operator-level overdose breakdown (dataset since retired, not independently re-confirmable)

At the individual-location level, the City's now-retired "Suspected Opioid Overdoses in Shelters" open-data source recorded non-fatal opioid-overdose counts at two operators this project profiles closely elsewhere (shelter-operator-financial-profiles): Homes First and Dixon Hall, 2018-2025. Both operators peaked sharply in 2021 (426 for Homes First, 160 for Dixon Hall), matching the system-wide 83% year-over-year increase documented above and, independently, 2021's status as the deadliest year on record in this document's shelter-resident mortality series [From this library’s earlier research from this library's prior synthesis document (Mortality Overdose Data), original sourcing: City of Toronto, "Suspected Opioid Overdoses in Shelters," direct primary source, quarterly by location]. ⚠️ Still being checked: the City's own open-data source for this specific breakdown has since been retired and could not be independently re-confirmed in this review, so this figure is not held to this document's usual verified-citation bar [From this library’s earlier research from this library's inherited source document (homelessness mortality and overdose data)]. These are counts of reported incidents, not a population-adjusted rate — Homes First's higher raw count than Dixon Hall's is consistent with, though not proof of, Homes First's substantially larger bed-night footprint rather than a claim that its individual sites are more dangerous per capita. A first-hand field-observation caveat, preserved here because it materially affects how these figures should be read: a campaign director with direct field exposure to the sector has observed that these numbers are very likely understated, since most overdoses are not formally reported — this is not an independently-verified statistic and is not presented as one, but is consistent with the well-established general pattern that self-reported incident data in stigmatized, criminalized-behaviour contexts tends to undercount true incidence (the same dynamic this document notes elsewhere regarding shelter staff's limited ability to attribute cause of death) [From this library’s earlier research from this library's prior synthesis document (Mortality Overdose Data)]. This page’s own inherited guidance is to cite these figures as a documented floor ("at least 426 reported non-fatal overdoses at Homes First locations in 2021"), never as a ceiling or complete count.

Ontario-wide shelter-specific opioid mortality study (ICES/ODPRN)

A study by the Ontario Drug Policy Research Network (St. Michael's Hospital) and Public Health Ontario, analyzing Chief Coroner and ICES data, found 210 accidental opioid-related deaths within Ontario shelters between January 2018 and May 2022 — more than tripling during the study period (48 deaths in the pre-pandemic period vs. 162 during the pandemic) — while Ontario's total emergency shelter bed capacity grew only 15% over the same window (6,764 to 7,767 beds) [From this library’s earlier research from this library's prior synthesis document (Mortality Overdose Data), original sourcing: ICES/ODPRN, June 18 2024, lead author Bisola Hamzat]. The lead researcher's own framing, preserved directly: shelter residents face "housing instability" and "complex healthcare needs and unique barriers to accessing treatment and harm reduction programs" simultaneously — a compounding-risk finding, not a simple capacity-shortfall finding.

2025 opioid-related shelter deaths: confirmed declining, consistent with the page’s inherited W4-corrected finding

The page’s inherited master briefing (CL-90683) already carries a W4-corrected finding that 2025's 13 fatal shelter overdoses represented a 52% decrease from 2024. A direct live fetch this review of the City's current overdose-surveillance page confirms this figure remains current and gives the most recent quarterly detail:

"In total, there were 13 opioid toxicity/suspect drug-related deaths in 2025, representing a 52% decrease from 2024... Based on preliminary data, there were < 5 suspect drug related deaths of shelter residents in Toronto shelters in Q4 2025."

Source: City of Toronto, "Overdoses in Homelessness Services Settings," https://www.toronto.ca/city-government/data-research-maps/research-reports/housing-and-homelessness-research-and-reports/overdoses-in-homelessness-services-settings/, page date modified 2026-03-06. Directly fetched and read 2026-07-14. [NEW-2026-MORT-3]

The same source, and the City's own shelter-death annual review, independently corroborate a genuine, sustained decline specifically in fatal opioid outcomes:

"Opioid overdose as the suspected cause of death has decreased significantly since 2020 (by 72%) with a consistent downward trend since 2022. This is consistent with data from the Ontario Office of the Chief Coroner."

Source: same as NEW-2026-MORT-1. [NEW-2026-MORT-4]

This directly confirms the page’s own W4-corrected "declining-not-rising" opioid framing remains accurate as of the most current data available (2026-07-14 live fetch): fatal shelter-specific opioid outcomes have been on a consistent downward trend since 2022, independently stated by two different City datasets (the shelter-death annual review and the shelter-overdose surveillance page) that draw on two different underlying data sources (TSSS's own death-report database and the Ontario Office of the Chief Coroner).

Monthly overdose trend by year, cause-of-death share, and the City's harm-reduction response context

This page’s inherited master briefing gives a longer monthly-average view than the quarterly figures above: non-fatal suspected shelter overdoses averaged 74/month (2022), 66/month (2023), 46/month (2024), 47/month (2025), and 96/month for 2026 year-to-date as of the master briefing's writing — roughly double the 2025 monthly average [From this library’s earlier research from this library's prior synthesis document (Mortality Overdose Data)]. Drug toxicity's share of all deaths of people experiencing homelessness has moved across recent years: 47% (2022, original methodology), 53% (2022, revised methodology after OCC integration), 56% (2023), and 55% (full-year 2024) [From this library’s earlier research from this library's prior synthesis document (Mortality Overdose Data) and this library's inherited source document (homelessness mortality and overdose data)]. For context, not as an evaluated finding: the City's Integrated Prevention and Harm Reduction Initiative (iPHARE) was established in December 2020; this page’s inherited master briefing notes the 2025 decline in fatal shelter overdoses (documented above) may be associated with iPHARE and related harm-reduction measures, while cautioning that the relationship between supervised-consumption-site closures, HART Hub transitions, and these shelter-specific overdose trends is more complex than a single number can capture, and should not be conflated with the separate, opposite-direction ambulance-call finding this project's HART Hub material documents elsewhere [From this library’s earlier research from this library's prior synthesis document (Mortality Overdose Data)].

But non-fatal overdose calls tell a more complicated, less reassuring story

New this review, a genuine tension the page’s inherited framing did not have current data to surface. While fatal shelter overdoses are declining, non-fatal overdose calls show a sharp recent uptick not present in the page’s inherited figures:

"There were 236 calls to Toronto Paramedics from Toronto shelters and respites sites in Q4 2025 to respond to non-fatal overdoses, an 80% increase from Q3 2025 (N=131)... There were 567 suspected non-fatal overdoses in Toronto shelters over 2025, based on calls to Toronto Paramedics. This is a 2% decrease from 2024."

Source: same as NEW-2026-MORT-3. [NEW-2026-MORT-5]

Full-year 2025 (567 non-fatal calls) is nearly flat versus 2024 (a 2% decrease), consistent with the page’s inherited figures. But the Q4 2025 quarter alone shows an 80% jump over Q3 2025 — a genuinely new, unresolved signal this backgrounder flags rather than smooths into the "declining" full-year framing. Two different things are true in the same dataset at once: full-year 2025 fatal outcomes are down sharply (-52%), full-year 2025 non-fatal outcomes are roughly flat (-2%), and the most recent quarter (Q4 2025) shows a sharp non-fatal increase not yet reflected in any annual summary. This document does not resolve whether the Q4 2025 spike is a one-quarter anomaly or the start of a reversal — that is a genuinely open question as of this review.

Provincial context: Ontario's opioid mortality trend is more mixed than Toronto's shelter-specific trend alone suggests

New this review. The Ontario Office of the Chief Coroner's most current quarterly summary (data effective May 7, 2026, covering through Q4 2025) shows a provincial trend that complicates a simple "declining" narrative when read at the province-wide level, as distinct from Toronto's shelter-specific fatal-count trend above:

"The number of opioid-related deaths in the most recent quarter (Q4 2025; 386 deaths) is 7% higher than the deaths in quarter prior (Q3 2025; 360 deaths) (preliminary). Q4 2025 reflects a 17% increase in deaths when compared to Q1 2025."

Source: Office of the Chief Coroner (Ontario) / Ontario Drug Policy Research Network, "Quarterly Update from the Office of the Chief Coroner — Opioid Toxicity Deaths in Ontario," data effective May 7, 2026, https://odprn.ca/wp-content/uploads/2026/05/OCC_Opioid-Mortality-Summary-2025_Q4-May-2026.pdf. Directly fetched and read 2026-07-14. [NEW-2026-MORT-6]

The same source gives the full-year picture, which does show an overall decline from the pandemic-era peak but at a level still well above pre-pandemic baseline:

"In 2025, the mortality rate has decreased by 54% compared to 2021, however, remains 44% higher than in 2016." Full-year opioid toxicity deaths in Ontario: 2,266 (2024) and preliminary 1,398 (2025, likely to rise as ongoing investigations resolve, per the source's own preliminary-data caveat).

Source: same as NEW-2026-MORT-6. [NEW-2026-MORT-7]

The homelessness-specific subset of this same provincial dataset:

"In Q4 2025, nearly 1 in 5 opioid toxicity deaths occur amongst people experiencing homelessness (including unsheltered, emergency sheltered, provisionally accommodated, or at immediate risk of homelessness)."

Source: same as NEW-2026-MORT-6. [NEW-2026-MORT-8]

This is a genuine, disclosed tension this document states rather than resolves: Toronto's shelter-specific fatal-overdose count is down sharply on a full-year basis (13 in 2025, -52% vs. 2024), and the City's own framing attributes the broader decline in suspected-opioid-cause deaths to a "consistent downward trend since 2022" — but the province-wide picture shows Q4 2025 deaths rising versus both Q3 2025 and Q1 2025, and homeless Ontarians continue to make up a disproportionate and undiminishing share (nearly 1 in 5) of all provincial opioid deaths. Toronto's shelter system specifically may be doing better than the provincial trend as a whole; this document does not have a source directly explaining why Toronto's shelter-specific figure and the provincial quarterly figure move in different directions in the same period, and does not speculate.

The drug supply itself: still shifting, still dangerous

New this review, from the same OCC quarterly source. The specific substances involved in Ontario's opioid deaths continue to shift toward higher-potency, harder-to-reverse combinations:

"Fentanyl involvement in opioid toxicity deaths has decreased by 10% between 2024 and 2025... Medetomidine is involved in nearly 1 in 10 opioid toxicity deaths [2025]... Stimulants are involved in nearly 7 in 10 opioid toxicity deaths [2025]."

Source: same as NEW-2026-MORT-6. [NEW-2026-MORT-9]

This is consistent with, and updates with a full-year 2025 figure, the page’s inherited finding on medetomidine's emergence as a xylazine-adjacent contaminant. Fentanyl's declining involvement share (down 10% year-over-year) does not necessarily mean the supply is safer — it may reflect increasing polysubstance complexity (stimulant co-involvement now in nearly 7 in 10 deaths) rather than a straightforward risk reduction, and this document does not resolve which interpretation the evidence best supports.

The international Housing First mortality literature: genuinely mixed, not a clean success story

New this review, addressing this page’s own flagged International-context opportunity. A live search this review for peer-reviewed Housing First mortality literature — distinct from Housing First's better-established housing-stability outcomes — found a genuinely mixed evidence picture, not the clean mortality-reduction narrative sometimes assumed:

A 2015 observational study (Vancouver At Home/Chez Soi-adjacent cohort) found housing-status patterns of cause of death differed between Housing First participants (72% died of natural causes) and a comparison homeless group (49% natural causes), suggesting different mortality patterns rather than a simple lower/higher rate finding. A French randomized controlled trial found the overall 2-year mortality rate was higher in the Housing First group (6.5%) than the treatment-as-usual group (3.4%), though the study's own authors flagged limitations preventing a definitive causal conclusion. A large (2,108-participant) randomized clinical trial of Housing First for homeless adults with mental illness found no statistically significant difference in mortality between the Housing First and treatment-as-usual arms. A systematic review of 26 US/Canadian studies did find Housing First associated with a 37% mortality reduction specifically among HIV-positive clients — a population-specific finding, not a general-population one.

Source: WebSearch synthesis this review, corroborating multiple peer-reviewed sources including Fitzpatrick-Lewis et al. and the French ANRS-Cosmos trial as reported via PMC-indexed articles (https://pmc.ncbi.nlm.nih.gov/articles/PMC4669629/, https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8254224/, https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12314730/). ⚠️ Still being checked: this is a search-summary synthesis across multiple sources, not a single primary document directly fetched and read in full this review — a future pass should independently confirm each study's precise figures against its own published text before this finding is cited at higher confidence than "directionally corroborated across multiple independent search results."

This is genuinely honest absence, not a gap in this review's effort: Housing First's strongest, most consistently-replicated evidence base is housing stability (already documented in this project's homelessness-housing-first-model backgrounder, e.g. the At Home/Chez Soi 73%-vs-31% stability finding), not all-cause mortality reduction specifically. The mortality-specific literature is real, peer-reviewed, and genuinely mixed — including at least one RCT finding higher mortality in the Housing First arm — and this document states that plainly rather than importing an unearned mortality-reduction narrative onto Housing First's genuinely strong housing-outcome evidence base.

The drug supply itself, expanded: adulterants beyond medetomidine, and the provincial scale of the crisis

Beyond the fentanyl/medetomidine/stimulant composition data already given above, this page’s inherited master briefing additional adulterant and provincial-scale findings not yet carried forward into this document. Toronto's Drug Checking Service confirmed a surge in early April 2026 — five deaths believed linked to opioids between April 3-6, four in close proximity in the downtown/midtown area — attributed to increasing contamination and unpredictability in the fentanyl supply, including medetomidine, a veterinary tranquilizer more potent than xylazine [From this library’s earlier research from this library's prior synthesis document (Mortality Overdose Data), original sourcing: CBC News, April 9, 2026]. Carfentanil — up to 100 times more potent than fentanyl — is a real, ongoing contaminant, not a rare event: of 179 Toronto drug-checking samples containing carfentanil between October 2019 and August 2025, 88% also contained at least one other high-potency opioid, and 62% also contained a benzodiazepine-related drug or veterinary tranquilizer, meaning carfentanil almost never appears alone and compounds with other depressants in ways that make standard naloxone response less reliable [From this library’s earlier research from this library's prior synthesis document (Mortality Overdose Data), original sourcing: Toronto's Drug Checking Service, updated January 2026]. This is a provincial, not Toronto-specific, crisis: the North Bay Parry Sound District Health Unit reported opioid-related overdose activity in April 2026 (42 overdoses, 4 fatal) exceeding its previous peak (38 overdoses, September 2020) — the highest since its surveillance program began in 2019 — and Ontario public health units collectively issued 20 separate drug alerts in April 2026 alone [From this library’s earlier research from this library's prior synthesis document (Mortality Overdose Data), original sourcing: North Bay Parry Sound District Health Unit, May 5 2026]. On what supervised consumption sites had been preventing: between March 2020 and May 2024, trained workers at supervised consumption sites across Ontario medically intervened to prevent an estimated 22,000 overdose deaths, and research on the areas immediately surrounding these sites found crime rates remained stable or declined following implementation; a further 10 supervised consumption sites closed in 2025, half of them in Toronto [From this library’s earlier research from this library's prior synthesis document (Mortality Overdose Data), original sourcing: Social Planning Toronto, citing Ontario harm reduction data]. This gives this page’s existing HART Hub/SCS-closure material a concrete before-and-after anchor: not just that overdose calls rose after closures, but a specific estimate of what the closed sites had been preventing.

This document's mortality and overdose material has focused almost entirely on opioids, reflecting where the current acute crisis is — but alcohol use disorder among people experiencing homelessness is a distinct, severe problem Toronto has real, evidence-based policy experience addressing. Alcohol use disorders affect an estimated 30-50% of people experiencing homelessness — a scale comparable to opioid use disorder in this population — frequently compounded by non-beverage alcohol consumption (rubbing alcohol, mouthwash, hand sanitizer) when beverage alcohol is unaffordable or unavailable [From this library’s earlier research from this library's prior synthesis document (Mortality Overdose Data), original sourcing: NEJM Catalyst, 2022, peer-reviewed review]. An 11-year follow-up study (Hwang et al., BMJ, 2009 — the same cohort study already cited above for opioid-adjacent excess mortality) found alcohol-related death mortality ratios of 6.4 times for men and 8.5 times for women among people living in shelters, rooming houses, and hotels, compared to the general population — consistent with this document's broader pattern that homelessness-related mortality is more severely gendered against women than raw death counts alone suggest [From this library’s earlier research from this library's prior synthesis document (Mortality Overdose Data)]. Toronto has real policy history here: Canada's first Managed Alcohol Program (MAP) was created in 1996 at Seaton House, then Toronto's largest shelter, directly in response to the freezing deaths of three homeless men who had been turned away from shelter accommodation because of their drinking; a coroner's inquiry into those deaths recommended the "wet shelter" model — supervised, dosed alcohol consumption on-site — as the policy response [From this library’s earlier research from this library's prior synthesis document (Mortality Overdose Data)]. A subsequent outcome study (Podymow et al., CMAJ, 2006) found the program significantly decreased emergency department visits and police encounters while stabilizing participants' alcohol intake [From this library’s earlier research from this library's prior synthesis document (Mortality Overdose Data), original sourcing: Podymow et al., CMAJ, 2006]. MAPs remain small in scale: there are now "over 40" operating across Canada, but individual Toronto sites like Art Manuel House serve as few as 10 people at a time — a capacity-vs-need gap this document notes without adopting the master briefing's own recommendation-framing on it (see below) ⚠️ Still being checked: the "over 40" national count was not independently re-confirmed this review.

Stimulants: a separate, rapidly worsening crisis

While opioids dominate the fatal-overdose data above, methamphetamine and other stimulant use has grown sharply and compounds the crisis in ways specific to homelessness. CAMH has confirmed that emergency department visits for amphetamine use (predominantly methamphetamine) increased more than 700% between 2014 and 2021, with inpatient admissions for amphetamine-related presentations rising nearly 300% over the same period; province-wide, Ontario has seen a 15-times increase in amphetamine-related emergency department visits over the last 20 years [From this library’s earlier research from this library's prior synthesis document (Mortality Overdose Data), original sourcing: CAMH, "Addressing the Mental Health and Addictions Crisis in Toronto"]. Unlike opioid use, methamphetamine use among homeless populations is frequently reported as functional rather than purely recreational: people who are unsheltered report using it specifically to stay awake and protect themselves and their belongings overnight, a documented harm-reduction-adjacent behaviour pattern distinct from typical substance-use framing [From this library’s earlier research from this library's prior synthesis document (Mortality Overdose Data), original sourcing: CATIE; CBC longform reporting]. This matters for how this page’s material should be read: a purely abstinence- or treatment-oriented response misses that some stimulant use in this population is a rational response to the danger of unsheltered sleep, not solely a symptom of addiction. CAMH has specifically called for urgent investment (its own figure: $500,000) in a Rapid Access Addiction Medicine (RAAM) clinic and hub specifically for methamphetamine and stimulant users in Toronto — a concrete, already-articulated ask, named here descriptively as CAMH's own stated position rather than adopted as this document's own recommendation [From this library’s earlier research from this library's prior synthesis document (Mortality Overdose Data)].

A note on the master briefing's own REC M-13/REC M-14 recommendation framing: this page’s inherited master briefing frames the Managed Alcohol Program scale-up and a shelter-system gender-capacity/harm-reduction parity mandate as two developed recommendations (REC M-13, REC M-14) pointing to a POLICY-001 recommendations document. POLICY-001 was not found anywhere in this corpus (checked directly this review), and the underlying gender-capacity finding itself is carried, with a verified handoff, by the sibling homelessness-gender-based-violence page (see below). Consistent with this document's own stated neutrality convention (already applied above to the Finland comparator), the underlying facts are restored above and below; the recommendation-framing itself is correctly excluded from this document's own prose, since recommendations belong in a claim_type:recommendation claims-register row or an L6 card, not a backgrounder — this is a design choice, not a silent drop.

Encampment fire deaths: a distinct mortality category, not covered by the shelter/service-setting data above

This document's mortality data above covers shelter-resident deaths and shelter/service-setting overdoses; it does not, until this restoration, name encampment fire deaths, a genuinely distinct fatality category. A December 2020 City staff report to Council states directly that seven people have lost their lives as a result of encampment fires in Toronto since 2010 [From this library’s earlier research from this library's prior synthesis document (Mortality Overdose Data), original sourcing: City of Toronto, staff letter to Planning and Housing Committee, December 7, 2020]. The same report documents a sharp spike in fire incidents (not necessarily deaths) in 2020 itself: Toronto Fire Services responded to 216 fires in encampments in 2020 (to December 2 that year) — a 218% increase over the same period in 2019 — with one of the seven total deaths since 2010 occurring that year; as of the report's snapshot date, 395 tents were counted across 66 park sites citywide. The report attributes fire risk specifically to open flames, generators, unsafe wiring, and flammable-substance storage in outdoor conditions, and separately flags polystyrene-insulated "pods" as a newly-emerging, especially combustible risk that does not meet Ontario Building or Fire Code structural definitions [From this library’s earlier research from this library's prior synthesis document (Mortality Overdose Data)]. Whether Toronto Fire Services has published an updated fire-count or fatality total since this 2020 snapshot was not checked in this review — see "Open questions" below.

A citation-correction note: the Federal Housing Advocate's "59 deaths in 2025" figure

Canada's Federal Housing Advocate cited "59 deaths...recorded by the City of Toronto in 2025" in a 2026 report. This page’s inherited master briefing traced this figure directly to TSSS's own 2024 Annual Review of Statistics on Deaths of Shelter Residents, which reports 59 deaths of shelter residents specifically in 2024 (not 2025) — meaning the Advocate's citation, as originally traced, appeared to misattribute both the year and the category (shelter-residents-specifically vs. the broader all-homeless-deaths total) [From this library’s earlier research from this library's prior synthesis document (Mortality Overdose Data)]. This document notes an important wrinkle the master briefing could not have known at the time it was written: this document's own live re-verification above (NEW-2026-MORT-1) confirms the City's current 2025 Annual Review states 59 deaths also occurred in 2025, "unchanged from 2024" — meaning the Advocate's headline number is, coincidentally, now also correct for 2025 specifically, even though the category caveat still applies (59 is the shelter-resident-specific figure, not the broader all-homeless-deaths total, which this document does not have a 2025 figure for). The general methodological caution the master briefing raises — that a narrow shelter-resident-specific figure should not be cited without its qualifier, and without confirming which year it actually describes — remains valid practice regardless of this numerical coincidence.

Gendered mortality and shelter-capacity findings: owned by the sibling GBV page

This page’s inherited master briefing (this library's prior synthesis Part 3E) documents a 2026 peer-reviewed BMC Public Health study (Ali et al.) finding 215 total 2024 Toronto homelessness deaths broken down by gender, and a structural shelter-capacity/harm-reduction-access gap by gender (93 adult shelters, only 25% women-only, women-only shelters holding only 18% of bed capacity against a 41% population share, and the lowest harm-reduction-service rate of any shelter type). Verified this review: the sibling page homelessness-gender-based-violence's own backgrounder carries this exact study and all of these figures directly and in full (its "Mortality and cause-of-death gap by gender" section), consistent with that page’s own stated scope, which claims ownership of the gendered mortality/shelter-capacity finding while handing off only general mortality/overdose methodology to this page. This document accordingly treats that content as owned elsewhere rather than restating it here; see this page’s own Scope section, and the canonical-figure note above regarding the 36-vs-38 age-at-death discrepancy this review surfaced between this page’s own inherited master briefing and the GBV page’s.

Finland comparator: strong housing outcomes, no mortality-specific figure located

Finland's Housing First model (Y-Foundation, national policy since 2007-2008) is well-documented for housing-stock and homelessness-count outcomes — homeless shelter/hostel bed counts in Helsinki fell from 2,121 (1985) to 52 (2016) while independent rental units for formerly homeless people rose from 65 to 2,433, and hostel/boarding-house homelessness fell 76% nationally from 2008 to 2017 [same WebSearch pass, corroborating Centre for Public Impact and HUD USER secondary sources; ⚠️ Still being checked: not independently fetched from a Finnish primary source this review]. No mortality-specific figure for Finland's Housing First cohort was located in this review — this is stated as a genuine gap, not filled with an inference from the housing-outcome data, which measures a different thing.

Key tensions / tradeoffs

A corrected, larger cumulative shelter-death total, read two ways. The City's own cumulative shelter-death figure rose from 825 (2024 Annual Review) to 884 (2025 Annual Review) [NEW-2026-MORT-1] — this is one additional year of deaths added to a running total, not a downward retroactive revision (contrast with the TPH broader dashboard's genuine retroactive upward revision of 2022/2023 figures via OCC data integration, already documented in this page’s inherited master briefing via CL-792). Both a "the shelter-death toll has now passed 884" framing and a "2025's annual death count (59) was unchanged from 2024, and the underlying rate fell when adjusted for a larger service population" framing are simultaneously true and sourced from the same document — this document states both without picking one as the more important framing.

Toronto's shelter-specific decline against a rising provincial quarterly trend. Toronto's shelter-system fatal opioid count fell 52% in 2025 and the City frames the broader suspected-opioid-cause trend as a "consistent downward trend since 2022" [NEW-2026-MORT-4] — while the province-wide OCC data shows Q4 2025 opioid deaths up 7% over Q3 2025 and 17% over Q1 2025 [NEW-2026-MORT-6], with homeless Ontarians still comprising nearly 1 in 5 of all provincial opioid deaths, undiminished [NEW-2026-MORT-8]. This document does not resolve whether Toronto's shelter system specifically is bucking a worsening provincial trend, whether the province-wide Q4 uptick will show up in Toronto's own Q1 2026 data once available, or whether the two datasets (Toronto shelter-specific vs. province-wide-including-non-shelter) are simply measuring different enough populations that no direct comparison is warranted.

A sharp Q4 2025 non-fatal overdose spike inside an otherwise-flat full year. Full-year 2025 non-fatal shelter overdose calls were roughly flat versus 2024 (-2%), but Q4 2025 alone was 80% higher than Q3 2025 [NEW-2026-MORT-5] — a genuinely unresolved signal that could reflect a real emerging trend, a data-reporting artifact, or a seasonal pattern this document does not have enough quarters of comparable data to assess confidently.

Housing First's strong housing-outcome evidence, against genuinely mixed mortality-outcome evidence. This page’s own broader corpus (via homelessness-housing-first-model) documents Housing First's strong RCT-level housing-stability evidence. This document's own literature search found the mortality-specific evidence base considerably weaker and more mixed, including at least one RCT finding higher, not lower, mortality in the Housing First arm. Both are true of the same underlying intervention studied for different outcomes — this document does not let the strength of the housing-stability evidence bleed into an unearned mortality claim.

What the evidence does and doesn't support

Well-supported:

Thin or contested:

International context

Treaties/frameworks touched

The right to health under Article 12 of the International Covenant on Economic, Social and Cultural Rights (ICESCR) is the most directly engaged international framework — the UN Committee on Economic, Social and Cultural Rights' General Comment No. 14 (2000) names access to health facilities, goods, and services, including harm-reduction and substance-use treatment access, as components of the right's "core obligations." A homeless-specific mortality gap of the scale documented here (average age at death 52-53 versus general-population life expectancy in the high 70s/mid-80s, per this page’s inherited academic sources) is a direct, quantifiable illustration of unequal enjoyment of this right, though this document does not have a source establishing that Toronto's specific mortality gap has been the subject of any UN-level review or communication.

2-3 best global comparators

Finland's Housing First model (Y-Foundation). Named above — strong housing-outcome evidence, no mortality-specific figure located this review. Included as a comparator on housing-stability grounds, with the mortality-specific gap in the evidence stated honestly rather than papered over.

Portugal's decriminalization-plus-harm-reduction model. Not independently fetched this review, but named in general terms as a frequently-cited comparator for drug-related mortality specifically (as distinct from Housing First's housing-stability focus) — Portugal's 2001 decriminalization of personal drug possession, paired with expanded treatment and harm-reduction access, is widely cited as associated with drug-related mortality rates well below the EU average in subsequent years. ⚠️ Still being checked: not independently confirmed with a current, directly-fetched primary source this review; named as a candidate for a future dedicated pass rather than cited with specific figures here.

Ontario's own Consumption and Treatment Services model, for internal comparison. Not a foreign comparator, but worth naming: this page’s inherited corpus already documents supervised consumption site closures and an associated increase in ambulance calls (per this page’s own inherited cross-reference to HART Hub material) — Toronto's own within-province experience offers a real, if contested, natural comparator for what happens to overdose outcomes when harm-reduction site access changes, without needing to reach for an international comparator at all. This is named here as a scope note rather than developed further, since HART Hub/SCS closure policy analysis belongs to a different page’s scope.

What Toronto/Ontario can steal shamelessly

Finland's model demonstrates a specific, nameable, transferable design principle directly relevant to this page’s own capacity-strain findings (cross-referenced from shelter-system-capacity-strain): permanent housing capacity was scaled up precisely as congregate shelter/hostel capacity was scaled down (Helsinki's hostel beds fell from 2,121 to 52 while independent rental units rose from 65 to 2,433) — a coordinated substitution, not simply an addition. This is stated descriptively; any specific "Toronto should replicate Finland's bed-to-unit substitution ratio" proposal belongs in a claim_type:recommendation claims-register row or an L6 card, never in this document's own prose.

Cui Bono — who profits from this problem persisting

Per this project's this library's live-discovery discipline and this library's standing requirement, this library's internal records was checked directly, and a targeted live-discovery search was conducted this review for any ESTABLISHED or REPORTED finding of an entity profiting from homelessness mortality or the opioid crisis specifically as it intersects Toronto's shelter system.

accountability seed landscape contains no row naming a shelter-mortality-specific or opioid-supply-specific beneficiary in Toronto. It is directly relevant, though, that accountability seed landscape's Auditor General of Ontario homelessness-programs entry (§1) is the same 2021 audit already cited as ESTABLISHED elsewhere in this corpus — a provincial-system-level finding about oversight failure, not a finding naming a specific entity profiting from mortality outcomes.

A targeted search this review for pharmaceutical-industry, naloxone-supply, or drug-testing-service procurement findings specific to Toronto's shelter system did not surface an ESTABLISHED or REPORTED finding meeting the Prime Rule's sourcing bar within the scope of this review.

Table: empty this review.

Honest explanation: this page’s scope (mortality and overdose surveillance data) is measurement and public-health-response infrastructure, not itself a service-delivery or contracting relationship with an obvious extraction vector comparable to, for instance, the financialized-rental-housing findings accountability seed landscape documents elsewhere. The unregulated drug supply itself — specifically, whoever profits from the fentanyl/medetomidine/carfentanil supply chain killing people in Toronto's shelter system — is a genuine, serious Cui Bono question this document's scope (City/provincial surveillance data) is not positioned to answer; it would require organized-crime/supply-chain investigative reporting this review's discovery did not locate meeting the Prime Rule's sourcing bar, and this is named as a real, serious, and unresolved gap rather than a non-issue.

Indigenous context

⚠️ W3 DEFERRAL (genuine, not a placeholder): this page’s own inherited master briefing (Part 3E of this library's prior synthesis) already establish that women and gender-diverse people experiencing homelessness in Toronto show a dramatically higher overdose share of mortality (76% and 80% of deaths, respectively, versus 48% for men) — Indigenous-specific mortality overrepresentation within Toronto's homelessness-mortality data is a distinct, real, and directly relevant angle this document's own discovery pass this review did not have scope to develop with the same rigor as the gendered findings above. This is flagged honestly as deferred to a dedicated W3 Indigenous-context pass rather than authored thinly here — a homelessness-mortality backgrounder that gestures at Indigenous overrepresentation without a properly sourced, dedicated treatment would be worse than naming the gap plainly.

Open questions / data gaps

Claim-index appendix

carried-forward (carried forward from this page’s own sources docs, cited as-is; IDs do not currently resolve in this library's internal records register/the claims register, checked directly 2026-07-14):

New load-bearing findings (this review, source quotes below, not yet through this library’s formal verification process):

---

Source quotes (NEW-2026-MORT-1 through NEW-2026-MORT-9)

NEW-2026-MORT-1 — Corrected cumulative shelter-death total.

"Since 2007, there have been 884 total shelter resident deaths reported to TSSS, an average of 49 deaths per year... 59 deaths occurred in 2025, which remains unchanged from 2024."

Source: City of Toronto, "2025 Annual Review of Statistics on Deaths of Shelter Residents," https://www.toronto.ca/city-government/data-research-maps/research-reports/housing-and-homelessness-research-and-reports/deaths-of-shelter-residents/annual-review-of-statistics-on-deaths-of-shelter-residents/, page date modified 2026-04-28. Accessed 2026-07-14.

NEW-2026-MORT-2 — Detailed breakdown.

"The shelter system provided nearly 44% more bed nights in 2025 as in 2021, so while deaths in 2025 were 55% lower than 2021 totals, when allowing for the larger service population, they were 68% lower... Males make up 60% of the shelter population, but 75% of decedents for 2021-2025... Of the 884 deaths of shelter residents reported to TSSS since 2007: The average number of deaths per year has been 49. The majority of decedents have identified as cis gendered male: 684 or 77%. The average (mean) age at death across all genders has been 52 years. 442 deaths (or 50%) occurred within a shelter/respite facility whereas 442 deaths (or 50%) occurred off-site."

Source: same as NEW-2026-MORT-1.

NEW-2026-MORT-3 — 2025 fatal overdose count.

"In total, there were 13 opioid toxicity/suspect drug-related deaths in 2025, representing a 52% decrease from 2024... Based on preliminary data, there were < 5 suspect drug related deaths of shelter residents in Toronto shelters in Q4 2025."

Source: City of Toronto, "Overdoses in Homelessness Services Settings," https://www.toronto.ca/city-government/data-research-maps/research-reports/housing-and-homelessness-research-and-reports/overdoses-in-homelessness-services-settings/, page date modified 2026-03-06. Accessed 2026-07-14.

NEW-2026-MORT-4 — Sustained decline confirmation.

"Opioid overdose as the suspected cause of death has decreased significantly since 2020 (by 72%) with a consistent downward trend since 2022. This is consistent with data from the Ontario Office of the Chief Coroner."

Source: same as NEW-2026-MORT-1.

NEW-2026-MORT-5 — Q4 2025 non-fatal spike.

"There were 236 calls to Toronto Paramedics from Toronto shelters and respites sites in Q4 2025 to respond to non-fatal overdoses, an 80% increase from Q3 2025 (N=131)... There were 567 suspected non-fatal overdoses in Toronto shelters over 2025... This is a 2% decrease from 2024."

Source: same as NEW-2026-MORT-3.

NEW-2026-MORT-6 — Provincial Q4 2025 uptick.

"The number of opioid-related deaths in the most recent quarter (Q4 2025; 386 deaths) is 7% higher than the deaths in quarter prior (Q3 2025; 360 deaths) (preliminary). Q4 2025 reflects a 17% increase in deaths when compared to Q1 2025."

Source: Office of the Chief Coroner (Ontario) / Ontario Drug Policy Research Network, "Quarterly Update from the Office of the Chief Coroner — Opioid Toxicity Deaths in Ontario," data effective May 7, 2026, https://odprn.ca/wp-content/uploads/2026/05/OCC_Opioid-Mortality-Summary-2025_Q4-May-2026.pdf. Accessed 2026-07-14.

NEW-2026-MORT-7 — Provincial full-year context.

"In 2025, the mortality rate has decreased by 54% compared to 2021, however, remains 44% higher than in 2016." Full-year deaths: 2,266 (2024), preliminary 1,398 (2025).

Source: same as NEW-2026-MORT-6.

NEW-2026-MORT-8 — Homeless-specific provincial share.

"In Q4 2025, nearly 1 in 5 opioid toxicity deaths occur amongst people experiencing homelessness (including unsheltered, emergency sheltered, provisionally accommodated, or at immediate risk of homelessness)."

Source: same as NEW-2026-MORT-6.

NEW-2026-MORT-9 — Drug supply composition, 2025.

"Fentanyl involvement in opioid toxicity deaths has decreased by 10% between 2024 and 2025... Medetomidine is involved in nearly 1 in 10 opioid toxicity deaths [2025]... Stimulants are involved in nearly 7 in 10 opioid toxicity deaths [2025]."

Source: same as NEW-2026-MORT-6.