FactHub

Data

Deaths the King County Medical Examiner investigated at housing service sites, 2021-2026, by manner of death

The King County Medical Examiner's public count of deaths it investigated among people whose presumed living situation was a housing service site (permanent supportive housing, income-restricted apartments and recovery housing), by year and manner of death, 2021 to October 2026. Version 2 adds every year the dashboard offers and compares the trend with countywide overdose deaths, all Medical Examiner cases and federally counted permanent housing beds.

Coverage: Calendar years 2021 to 2025, and January 1 to October 29, 2026.

Send feedback on this dataset

What this dataset measures

Built for fact checks of deaths in King County's subsidized and supportive housing. This is the most complete count the county publishes: all manners of death, but only countywide and only for deaths the Medical Examiner investigates. 2021 is the earliest year the dashboard offers for this breakdown.

Geography
King County, Washington
Unit of observation
Count of deaths investigated by the Medical Examiner, by year and manner of death, countywide
Update cadence
Updated by FactHub when the Medical Examiner revises its dashboard; the dashboard itself was last updated October 29, 2026
Curator
FactHub Research Team

Curated data tables

These related tables form this versioned dataset release. Each table has its own downloadable file and field definitions.

Deaths at housing service sites by manner of death, 2025

One row per manner of death, from the dashboard's breakdown for 2025 with the living-situation filter set to housing service site.

Unit of observation: Manner of death

View this table

kcme-housing-site-deaths-2025.csv text/csv · 242 bytes
4 field definitions
FieldTypeMeaningUnitsValue type
manner string Plain-language manner of death. — derived
dashboard_label string The manner of death exactly as the dashboard labels it. — observed
deaths integer Deaths the Medical Examiner investigated in 2025 at housing service sites, countywide. deaths observed
share_of_total number Share of the 379 deaths. fraction derived

Deaths at housing service sites by year and manner of death, 2021-2026

One row per year and manner of death. 2026 runs January 1 to October 29.

Unit of observation: Year and manner of death

View this table

kcme-housing-site-deaths-by-year.csv text/csv · 2.1 KB (2,196 bytes)
6 field definitions
FieldTypeMeaningUnitsValue type
year integer Calendar year of death. — observed
period string The period counted: a full year, or January 1 to October 29 for 2026. — observed
manner string Plain-language manner of death. — derived
dashboard_label string The manner of death as the dashboard labels it. — observed
deaths integer Deaths of this manner, summed across the dashboard's sex categories. Zero means the dashboard shows none. deaths observed
year_total integer The dashboard's total deaths at housing service sites for the year. deaths observed

Housing site deaths beside county deaths and housing beds, 2021-2025

One row per year with the comparison series used to judge the housing site trend.

Unit of observation: Year

View this table

housing-deaths-context-by-year.csv text/csv · 481 bytes
10 field definitions
FieldTypeMeaningUnitsValue type
year integer Calendar year. — observed
housing_site_deaths_all_causes integer Deaths at housing service sites, all manners. deaths observed
housing_site_overdose_deaths integer Overdose deaths among people whose presumed living situation was a housing service site. deaths observed
county_overdose_deaths integer All overdose deaths the Medical Examiner investigated in King County. deaths observed
me_deaths_investigated_countywide integer All deaths the Medical Examiner investigated, every living situation. deaths observed
me_covid_surveillance_deaths integer Deaths investigated as part of COVID-19 surveillance, per the Notes tab; 0 for years the notes list none. deaths observed
me_deaths_excluding_covid_surveillance integer All investigated deaths minus COVID-19 surveillance deaths. deaths derived
hud_psh_beds integer Permanent supportive housing beds, Seattle/King County (WA-500). beds observed
hud_other_permanent_beds integer Other permanent housing beds, Seattle/King County (WA-500). beds observed
hud_permanent_beds_total integer Permanent supportive plus other permanent housing beds. beds derived

Deaths with an unknown living situation, 2021-2025

One row per year: deaths whose presumed living situation was recorded as unknown, for all causes and for overdoses.

Unit of observation: Year

View this table

unknown-living-situation.csv text/csv · 132 bytes
3 field definitions
FieldTypeMeaningUnitsValue type
year integer Calendar year. — observed
all_cause_unknown_living_situation integer Deaths the Medical Examiner investigated with PresumedLivingSituation = Unknown, all manners. deaths observed
overdose_unknown_living_situation integer Overdose deaths with an unknown living situation. 2024 and 2025 are total overdose deaths minus the four known categories. deaths observed

Deaths among people living outdoors, in shelters or in vehicles, 2021-2025

One row per year: all-cause deaths the Medical Examiner investigated with PresumedLivingSituation = Outdoors, homeless shelter, or vehicle.

Unit of observation: Year

View this table

outdoors-shelter-deaths.csv text/csv · 93 bytes
2 field definitions
FieldTypeMeaningUnitsValue type
year integer Calendar year. — observed
outdoors_shelter_vehicle_deaths_all_causes integer All-cause deaths among people whose presumed living situation was outdoors, a homeless shelter or a vehicle. deaths observed

Deaths among people living in private homes, 2021 to 2025

One row per year: deaths the Medical Examiner investigated with PresumedLivingSituation = Privately owned or rented residence.

Unit of observation: Year

View this table

private-residence-deaths.csv text/csv · 253 bytes
5 field definitions
FieldTypeMeaningUnitsValue type
year integer Calendar year. — observed
private_residence_deaths_all_causes integer All-cause deaths; 2021 includes most of that year's COVID-19 surveillance cases. deaths observed
private_residence_accident_not_traffic integer Accidental deaths other than traffic, summed across the dashboard's sex categories. deaths observed
private_residence_natural integer Natural deaths, summed across the dashboard's sex categories. deaths observed
private_residence_overdose_deaths integer Overdose deaths among people living in a privately owned or rented residence. deaths observed

Saved analyses

Deaths at housing service sites by manner of death, 2025

Question: How many deaths did the King County Medical Examiner investigate in 2025 among people living at housing service sites, and by what manner?

Result: 379 deaths in 2025: 200 accidental (not traffic), 164 natural, 8 suicides, 3 homicides, 3 traffic, 1 undetermined. Countywide only; no breakdown by building or provider.

Operation: Counts by manner of death from the dashboard; share = deaths / 379

Scope: King County, countywide, 2025

View this analysis

  • Only deaths the Medical Examiner investigated; natural deaths are undercounted.
  • Housing service site mixes supportive, income-restricted and recovery housing.
  • Countywide only.

Deaths at housing service sites by year and manner, 2021-2025

Question: How many deaths at housing service sites did the Medical Examiner investigate each year, and of what kind?

Result: Totals: 309 (2021), 389 (2022), 487 (2023), 433 (2024), 379 (2025). Accidents other than traffic rose from 129 to 200; natural deaths stayed between 164 and 170 a year.

Operation: Counts by year from table deaths-by-year, grouping suicide, homicide, traffic and undetermined deaths as Other

Scope: King County, countywide, full years 2021-2025

View this analysis

  • Only deaths the Medical Examiner investigated; natural deaths are undercounted.
  • Housing service site mixes supportive, income-restricted and recovery housing.
  • Counts, not rates: no resident count is published.

Deaths at housing service sites, January 1 to October 29, 2026

Question: How many deaths at housing service sites has the Medical Examiner counted so far in 2026?

Result: 290 deaths through October 29, 2026, 19 fewer than the 309 in all of 2021.

Operation: Sum of 2026 rows in table deaths-by-year

Scope: King County, countywide, 2026 to October 29

View this analysis

  • Partial year; cases are still being added and 13 had a pending manner.

Change from 2021 to 2025: housing site deaths, county deaths and housing beds

Question: How did deaths at housing service sites change from 2021 to 2025 compared with county deaths and the supply of permanent housing?

Result: 2021 to 2025: overdose deaths at housing sites +68% (114 to 192); all deaths at housing sites +23% (309 to 379); county overdose deaths +28%; permanent housing beds +25%; other Medical Examiner deaths countywide, excluding COVID-19 surveillance, -3%.

Operation: (2025 value / 2021 value) - 1 for each series in table context-by-year; 'Other investigated deaths' = me_deaths_excluding_covid_surveillance - housing_site_deaths_all_causes

Scope: King County, full years 2021 and 2025

View this analysis

  • Beds are not residents and leave out most income-restricted and recovery housing.
  • Descriptive comparison; it does not show why deaths changed.

Robustness checks

Housing site deaths per 1,000 HUD permanent housing beds, 2021-2025

Question: Did deaths at housing service sites grow faster than the federally counted supply of permanent housing?

Result: Per 1,000 permanent housing beds, 2021 to 2025: all deaths 39.3 to 38.7 (flat); accidental deaths other than traffic 16.4 to 20.4 (+24%); overdose deaths 14.5 to 19.6 (+35%).

Operation: 1000 x deaths / hud_permanent_beds_total, rounded to one decimal, for all deaths, accidental deaths other than traffic (table deaths-by-year) and overdose deaths

Scope: King County, full years 2021-2025

View this analysis

  • Not a death rate: beds are not residents, and HUD's count leaves out most income-restricted and recovery housing in the Medical Examiner's category.
  • HUD's other permanent housing count jumped in 2022 and 2023 and fell in 2024, which moves the middle years.

Housing site deaths as a share of Medical Examiner cases, 2021-2025

Question: What share of the deaths the Medical Examiner investigated were at housing service sites?

Result: Excluding COVID-19 surveillance cases, housing service sites accounted for 9.1% of Medical Examiner deaths in 2021 and 11.2% in 2025, peaking at 12.5% in 2023.

Operation: housing_site_deaths_all_causes / me_deaths_excluding_covid_surveillance

Scope: King County, full years 2021-2025

View this analysis

  • The Medical Examiner's caseload is not all county deaths.

Does fewer 'unknown' living situations explain the rise?

Question: Could the 2021-2025 rise in housing-site deaths come from fewer deaths being recorded with an unknown living situation?

Result: All deaths at housing sites: +23% as reported, +23% with unknowns split proportionally, +16% (309 to 360) if every death that left the unknown group was a resident. Overdose deaths: +68% as reported, +60% split proportionally, +39% (114 to 159) under the bound, still above the countywide +28%.

Operation: Two tests for 2021 and 2025. Proportional: add unknown deaths to housing sites in proportion to housing sites' share of deaths with a known living situation, then recompute the change. Bound: assume every death that left the unknown group between 2021 and 2025 (19 all-cause, 33 overdose) was a housing-site resident in 2025, subtract them from 2025, and recompute the change.

Scope: King County, full years 2021 and 2025

View this analysis

  • The bound assumes the worst case for the finding, not a likely one.
  • Unknown counts for 2024 and 2025 overdoses are derived by subtraction.

Robustness check for: Change from 2021 to 2025: housing site deaths, county deaths and housing beds

Did deaths move from the shelter group into the housing group?

Question: Could the rise in housing-site deaths come from deaths being coded as housing that would earlier have been coded as outdoors, shelter or vehicle?

Result: Deaths among people living outdoors, in shelters or in vehicles rose 66% (188 to 312) from 2021 to 2025, while housing-site deaths rose 23%. The shelter group grew rather than shrank, so recoding from it into housing can't explain the housing rise.

Operation: Compare 2021 and 2025 all-cause deaths in the outdoors, shelter or vehicle group with housing service site deaths; (2025 / 2021) - 1

Scope: King County, full years 2021 and 2025

View this analysis

  • Only the outdoors, shelter or vehicle group was checked; 'other temporary arrangement' (which can include hotels) was not read for all causes.
  • A rise in both groups doesn't rule out some recoding; it shows recoding wasn't large enough to shrink the shelter group.

Did deaths move from private homes into the housing group?

Question: Could the 2021-2025 rise in housing-site deaths come from deaths being coded as housing service site that would earlier have been coded as a private home, for example because subsidized addresses were identified more often?

Result: Private-home deaths didn't rise with the county: accidental deaths 925 to 918 (-1%), overdoses 439 to 469 (+7%, vs +28% countywide). Housing sites' share of deaths among housed people rose from 12% to 18% (accidents) and 21% to 29% (overdoses). The dashboard can't tell a real shift from better identification of subsidized addresses. If the housing share had stayed at its 2021 level, housing-site deaths would be up 2% (309 to 316) and overdoses 19% (114 to 136), about as fast as housing beds (+25%).

Operation: Housed = housing service site + privately owned or rented residence. Housing share = housing deaths / housed deaths, for accidental deaths other than traffic and for overdoses, 2021 and 2025. Bound: hold the 2021 housing share fixed, apply it to 2025 housed deaths, and recompute the housing-site change (all causes = 379 minus the accidental deaths removed).

Scope: King County, full years 2021 and 2025

View this analysis

  • The bound is a worst case for the finding: it treats the whole rise in housing sites' share as recoding, although housing beds grew 25%, which by itself would raise that share unless other housing grew as fast.
  • Private-home all-cause deaths in 2021 include most of that year's 620 COVID-19 surveillance cases, so the all-cause private-home change is not comparable; the check uses accidents and overdoses.
  • No published record shows whether or when the Medical Examiner changed how it identifies housing service sites.

Did natural deaths shift toward the housing group the way overdoses did?

Question: If the Medical Examiner got better at recognizing subsidized addresses, deaths of every kind would move from the private-home group to the housing group, not only overdoses. Did the housing group's natural deaths grow relative to private homes as much as its overdose deaths did?

Result: Overdose ratio (housing to private homes): 0.26 in 2021, 0.41 in 2025 (+58%). Natural-death ratio: 0.095 to 0.167 (+76%) as published, or 0.148 to 0.167 (+13%) if all 2021 COVID-19 surveillance cases were private homes. From 2023, when surveillance had ended, to 2025 both ratios held steady or fell. The test can't settle whether coding changed between 2021 and 2023, because the dashboard doesn't say where the surveillance cases lived.

Operation: For each year: housing-site natural deaths / private-home natural deaths, and housing-site overdose deaths / private-home overdose deaths. A second natural ratio subtracts the year's COVID-19 surveillance cases (620 in 2021, 322 in 2022) from private-home natural deaths, as if all were private homes.

Scope: King County, full years 2021 to 2025

View this analysis

  • The Medical Examiner doesn't take all natural deaths, and it investigated extra natural deaths for COVID-19 surveillance in 2021 and 2022; their living situations aren't published.
  • Natural deaths depend on residents' ages and health, which may differ between the two groups and change over time.

Overdose deaths per bed with rapid rehousing counted

Question: Does the finding hold if rapid rehousing beds, which grew fastest, are added to the bed count?

Result: Counting only permanent supportive and other permanent housing, beds rose 25% and overdose deaths per 1,000 beds 35% (14.5 to 19.6). Adding rapid rehousing, beds rose 34% and overdose deaths per 1,000 beds 25% (12.6 to 15.8). Overdose deaths outgrew the housing either way.

Operation: 1000 x housing-site overdose deaths / (HUD permanent supportive + other permanent + rapid rehousing beds), 2021 and 2025

Scope: Seattle/King County CoC (WA-500)

View this analysis

  • Rapid rehousing tenants mostly rent private-market apartments with a time-limited subsidy, so the Medical Examiner may record them as living in private homes; counting their beds probably understates the rate.
  • HUD counts leave out most income-restricted and recovery housing, which the Medical Examiner's group includes.

Sources

  1. King County Medical Examiner's Office, KCMEO Annual Report dashboard: Characteristics of Deaths Investigated by KCMEO (last updated October 29, 2026) 2025 deaths at housing service sites by manner of death (Decedent Characteristics tab, PresumedLivingSituation = Housing service site)

    Retrieved November 7, 2026. Public government data. Counts are public government statistics; FactHub redistributes the counts it read.

    Reliability: Government. The county Medical Examiner's own public dashboard of the deaths it investigates; its notes say the deaths are not representative of all county deaths and that it does not take all natural deaths.

  2. Public Health - Seattle & King County, Methods Report: Documentation and Analysis of Fatal Overdose Data Definition of the housing service site category

    Retrieved November 7, 2026. Public government document. Linked; FactHub repeats only the definition it read.

    Reliability: Government. The county health department's description of how its death data assign a presumed living situation; it defines housing service site as permanent supportive housing, income-restricted units and recovery housing.

  3. King County Medical Examiner's Office, KCMEO Annual Report dashboard: Overview and Notes tabs (last updated October 29, 2026) All deaths the Medical Examiner investigated countywide by year, and COVID-19 surveillance counts

    Retrieved November 11, 2026. Public government data. Counts are public government statistics; FactHub redistributes the counts it read.

    Reliability: Government. The Medical Examiner's own dashboard. The Overview tab gives all deaths it investigated by year; the Notes tab gives how many of those were COVID-19 surveillance cases and says it does not take all natural deaths.

  4. Public Health - Seattle & King County, King County fatal overdose dashboard Overdose deaths by year, countywide and at housing service sites

    Retrieved November 6, 2026. Public government data. Counts are public government statistics; FactHub redistributes the counts it read.

    Reliability: Government. The county's dashboard of overdose deaths investigated by the Medical Examiner, by presumed living situation. The county funds much of the housing in the housing service site category.

  5. U.S. Department of Housing and Urban Development, 2007-2025 Housing Inventory Count by CoC Permanent supportive housing and other permanent housing beds for Seattle/King County (WA-500)

    Retrieved October 10, 2026. Public domain (U.S. government work). Public domain.

    Reliability: Government. A one-night inventory of beds reported each year by the local homelessness system to HUD; it counts beds, not occupants, and excludes most income-restricted and recovery housing.

License: FactHub's selection, cleaning and combined tables are licensed under CC BY 4.0. Source data keeps the terms listed above (terms).

Limitations and reproducibility

Reproducible from public sources: Read by hand from the Medical Examiner's public Tableau dashboard: Decedent Characteristics tab, PresumedLivingSituation filter set to Housing service site, year selector set to each year from 2021 to 2026 (2025 read November 7, 2026; other years November 11, 2026). The dashboard shows each manner of death split by sex; we added the sexes. Each year's manners sum to the dashboard's total for that year. All-case totals by year come from the Overview tab, COVID-19 surveillance counts from the Notes tab, overdose counts from the county's fatal overdose dashboard (as read for dataset king-county-overdose-deaths-by-place, version 1), and beds from HUD's Housing Inventory Count. Dashboards can be revised; later readings may differ.

  • Covers only deaths the Medical Examiner investigated, about 1 in 5 deaths in King County. It does not take all natural deaths, so deaths from illness are undercounted.
  • Housing service site mixes permanent supportive housing with income-restricted apartments and recovery housing; supportive housing alone can't be separated here.
  • A countywide total: the dashboard gives no count by building, provider or city.
  • No count of residents is published, so no death rate can be computed.
  • Presumed living situation is the Medical Examiner's classification from its investigation, not a match against housing records.
  • The dashboard offers 2021 as its earliest year for this breakdown, so there is no all-cause count for 2019 or 2020.
  • 2026 covers only January 1 to October 29 and includes 13 deaths whose manner was still pending; it is not comparable with full years.
  • The Medical Examiner also investigated deaths for COVID-19 surveillance in 2020 to 2022 (620 in 2021, 322 in 2022), which inflates its countywide totals for those years.
  • HUD beds are a one-night count of beds, not residents, and leave out most income-restricted and recovery housing, so deaths per bed is a rough comparison of growth, not a death rate.
  • Housing sites' share of deaths among housed people rose from 2021 to 2025; the dashboard can't show whether some of that came from better identification of subsidized addresses rather than more deaths (see analysis private-residence-recoding-check).

Publications using this dataset

Version details

The exact citing versions, and the version of this dataset each cites.

  • Article king-county-subsidized-housing-deaths-since-2021 — version 9 (published) cites version 7
  • Chart kcme-housing-site-deaths-2025 — version 1 (published) cites version 1
  • Chart kcme-housing-site-deaths-by-year — version 5 (published) cites version 7
  • Chart kcme-housing-deaths-vs-beds-2021-2025 — version 5 (published) cites version 7

Node: dataset:kcme-housing-site-deaths:7