Sage Journals: Discover world-class research

Abstract

Objective:

Homelessness is a pervasive and pressing problem in the United States and is associated with health risks and premature mortality. However, accurately defining and classifying homeless status can be complex, and assessment of homelessness status requires careful consideration of the data collection process. We assessed the validity of the ascertainment of homeless status on death certificates.

Methods:

We quantitatively compared the classification of homeless status from 2023 Multnomah County, Oregon, medical examiner data (considered the gold-standard source, due to the in-depth investigation and extensive data available as a case narrative) with housing status from Oregon Health Authority 2023 vital record (death certificate) data alone. We also described people with an indication of homelessness in the 2023 medical examiner data but not in the 2023 vital record data and considered possible causes of this misclassification.

Results:

A total of 171 of 394 (43%) decedents whom we classified as homeless via medical examiner data had an address listed in the final death certificate; all were significantly younger (aged 42 y) than the 223 people with concordant records (aged 46 y).

Conclusions:

We found that the presence of an address on a death certificate does not rule out homelessness at the time of death, especially among younger decedents. Other jurisdictions should assess the accuracy of their homeless classification data, as more research is necessary to quantify the presence of misclassification.

Keywords

medical examiner homelessness housing

Homelessness is a pervasive and pressing problem in the United States. In 2023, more than 650 000 people were experiencing homelessness on a given night in the United States.¹ Oregon has a large number of people experiencing homelessness (16 242 in a single night in 2023) and the third-highest state rate of unsheltered homeless individuals (48 per 10 000 people).¹ Experiencing homelessness can range from having no regular or adequate nighttime residence to staying at a succession of places (“couch surfing”) or living in a car. Multnomah County, which includes Portland, Oregon, is Oregon’s smallest county by geographic area but is the most populous, with approximately 789 000 residents in 2023.² In 2023, approximately 6300 people were experiencing homelessness on a given night, 63% of whom were unsheltered.³ Affordable housing shortages in Portland have contributed to rising homelessness; in 2024, the overall asking rent increased 3.3% from the previous year, to a total of $1677 per month compared with $1621 per month in 2023.⁴

Homelessness is associated with extensive health risks and, ultimately, premature mortality.^5,6 Collecting data on homeless status is critical to understanding the health risks associated with a lack of housing and ensuring that these risks are identified and addressed.⁷ However, accurately defining and classifying homeless status can be complex and require careful consideration of the data collection process.⁸ Uniquely, Oregon state law requires mandatory reporting of homeless status on death certificates, which allows for analysis of causes of death among people experiencing homelessness.⁹ However, the validity of the ascertainment of homeless status on death certificates has not been assessed.

We compared quantitatively the classification of homeless status from medical examiner data (considered the gold-standard source, due to the in-depth investigation and extensive data available as a case narrative) with housing status from vital record (death certificate) data alone. We also described people with an indication of homelessness in the medical examiner data but not in the vital record data and considered possible causes of this misclassification.

Methods

Homeless Status at Death

In January 2022, the passage of Senate Bill 850 (SB 850) by the Oregon Legislature required mandatory reporting of homeless status on death certificates.⁹ SB 850 amended Oregon Revised Statute 432.133 to read, “For a decedent who was homeless at the time of death, including a decedent who was homeless but receiving care at a hospital or other institution, a report of death required to be submitted under this section must indicate that the decedent’s residence address was ‘domicile unknown’.”⁹ The medical examiner is responsible for certifying the cause and manner of death for an official death certificate, whereas other demographic information, including residential address on the death certificate, is obtained by the funeral home (Figure 1).

Figure 1.

Description of data sources used to define homeless status in medical examiner data and vital records data, Multnomah County, Oregon, 2023. Dashed lines indicate potential areas where misclassification could occur. Hexagons indicate electronic databases. Abbreviations: EMS, emergency medical services; fire, fire and rescue services; law, law enforcement (patrol and detectives); MDI, medicolegal death investigator; ORS, Oregon Revised Statute; OVERS, Oregon Vital Events Registration System; SB 850, Senate Bill 850 (mandatory reporting of housing status on reports of deaths)

Data Source 1: Medical Examiner Record Review

Oregon Revised Statute 146.090 requires investigation of any apparently accidental deaths or deaths resulting from injury.¹⁰ Oregon has a semi-centralized medical examiner system, where the state medical examiner maintains statutory authority but counties perform their own death investigations. The Multnomah County Medical Examiner’s Office has recorded information on homeless status at death since 2011.¹¹ To classify cases as likely experiencing homelessness at death, we obtained data from the Multnomah County Medical Examiner’s Office for deaths occurring in 2023 from MDILog (Occupational Research & Assessment Inc), the case management software used by all county and state medical examiner staff to record information related to death investigations. The Multnomah County epidemiologist pooled data into a single dataset for review using 4 criteria:

The case was flagged as domicile unknown in the database.

The case had missing residence information in the database.

The case had key terms such as “transient” in the narrative report.

The case listed a shelter as the place where the death occurred.

The Multnomah County epidemiologist assessed medicolegal death investigator case narratives, supplemental information (eg, law enforcement reports, emergency medical services reports), and identification information for each case to determine which investigations supported the classification of homelessness via the US code (eg, primary residence was a transient encampment).¹² For unclear cases, the epidemiologist looked for more information in both vital records (eg, the final residential address was domicile unknown) or worked with an analyst with access to the Homeless Management Information System database, which includes data on homeless service access (eg, day services, rent assistance, creation of a client profile).¹³ These cases were classified as homeless if the address was listed as domicile unknown or evidence indicated that homeless services were accessed in the past 5 years. Five years was chosen as a conservative estimate; if there were doubts about the final homeless status, the epidemiologist erred on the conservative side and classified cases as likely not experiencing homelessness.

Data Source 2: Vital Records Data

The Multnomah County epidemiologist obtained vital records for all 2023 deaths from Multnomah County’s database. Vital records are updated weekly by the Oregon Health Authority and can be matched to medical examiner data via the medical examiner case identifier. Homeless status is captured in vital records through a complex process (Figure 1). Although homeless status is recorded in the medicolegal death investigator case narratives, which are transmitted to the funeral home, funeral home staff complete the demographic portion of the death certificate and may have alternative information that conflicts with the medical examiner report, which may lead to conflicting homeless status fields in medical examiner and vital records.

Analysis

The Multnomah County epidemiologist merged the 2 data sources (vital records data and medical examiner data) into a single analytic dataset using the medical examiner case number as the matching variable. We considered medical examiner homeless status as the definitive status. The final dataset included 394 medical examiner case records classified as homeless. We reported deaths when we found disagreement between the medical examiner homeless status and the presence of any address in the vital records other than domicile unknown (ie, we measured discordance). To describe the reasons for disagreement between the data sources, we collected additional information on these discordant cases. Specifically, the Multnomah County epidemiologist abstracted data from the death certificate on next of kin and informant names (first, middle, last), informant relationships to decedents, and informant mailing addresses.¹⁴ Because informant information is not contained in the electronic data extract used for routine analysis, the epidemiologist examined every copy of the uploaded death certificate that was retained as part of the complete medical examiner case record.

We used t tests to compare the mean age of decedents with concordant and discordant records. Additionally, among medical examiner cases categorized as homeless but with an address in the final death certificate, we used t tests to compare the mean age of records in which the decedent’s address matched the next-of-kin address to records in which the decedent’s address was another location. We defined 2 means as significantly different if the corresponding t test had a P value < .05. We conducted all analyses using SAS version 9.4 (SAS Institute, Inc).¹⁵

The Multnomah County Health Department Project Review Team reviewed this project and determined it was public health surveillance that did not require ethical or institutional review.

Results

Of the 394 medical examiner deaths classified as homeless in the dataset, 223 (57%) had domicile unknown in the residential address field in the vital record data and were considered concordant (Table). However, the remaining 171 deaths classified as experiencing homelessness in the medical examiner data review (43%) had some other address listed in the final death certificate (discordant). Of the 171 records identified as discordant, 78 (46%) addresses matched the next-of-kin address, and 93 (54%) did not.

Table.

Age comparisons among all medical examiner cases categorized as experiencing homelessness at the time of death (n = 394) and medical examiner cases categorized as domicile unknown but with an address in the final death certificate (n = 171), Multnomah County, Oregon, 2023

Case type	No. of cases	Mean age, y	t test	P value ^a
All medical examiner cases categorized as experiencing homelessness at the time of death			−3.07	.002
Discordant homeless status with vital records	171	42
Concordant homeless status with vital records	223	46
Medical examiner cases categorized as experiencing homelessness but with an address in final death certificate			2.56	.01
Mismatch: decedent address did not match next-of-kin informant address	93	44
Match: decedent address matched next-of-kin informant address	78	39

Using the t test, with P ≤ .05 considered significant.

For the 78 discordant records in which the decedent address matched the next-of-kin address, the most frequent next-of-kin type was parents (n = 52; 67%), followed by siblings (n = 10; 13%) and children (n = 8; 10%); the 8 additional records listed another residential address (Figure 2). Of the 93 discordant records where the decedent address did not match the next-of-kin address, 15 (16%) records had an address that matched a shelter location.

Figure 2.

Next-of-kin types (top) and other address types (bottom) for decedents classified as domicile unknown but with an address in the final death certificate (n = 171), Multnomah County, Oregon, 2023.

Age Comparisons

The average (range) age at death for all 394 cases was 44 (19-78) years. Of the 394 medical examiner cases, the 171 cases with a discordant address on the death certificate (mean age, 42 y) were significantly younger than the 223 cases where there was not a discordant address (mean age, 46 y; t = −3.07; P = .002) (Table). Furthermore, for the subset of cases with discordant addresses (n = 171), those whose address matched their next-of-kin informant address on the death certificate (n = 78; mean age, 39 y) were significantly younger than the 93 whose address did not match their next-of-kin informant address on the death certificate (mean age, 44 y; t = 2.56; P = .01).

Discussion

We assessed the discordance of homeless status between medical examiner data and vital records data among deaths in Multnomah County, Oregon, in 2023. We found that vital records data undercounted the number of people who were homeless when they died, especially among younger decedents, when compared with the medical examiner data (gold standard). Nearly half of these discrepant records appeared to be attributable to next of kin providing their own address on the final death certificate. Because this misclassification—resulting from analyzing vital records only—would underestimate the true number of people dying while experiencing homelessness, it is important to understand and quantify these findings. Although the use of medical examiner data to provide enhanced surveillance in our jurisdiction is well documented,^16-19 to our knowledge, our study is the first to compare homeless status in medical examiner data with homeless status in the final death certificate data.

We found that when discordance occurred between the medical examiner homeless status and the vital records homeless status, the decedent was likely to be younger. Although it is well known that people experiencing homelessness encounter considerable social stigma due to their situation,²⁰ documentation is scarce on any postmortem stigma that next of kin or other key informants might encounter when they are asked to provide the decedent’s address for a final death certificate. A study of qualitative interviews with death investigators in New Mexico revealed that people who enter death data can be influenced by perceived stigma.²¹ Furthermore, the presence of identification can add confusion, especially for homeless young people who may still have parental or caregiver information on file, even if they are homeless and/or only receive mail at the address.²¹ As Green noted in that study, “A person’s address is not necessarily where they are living.”²¹ Green further noted that history, place, and time can interfere with decision-making about homeless status. The current analysis revealed that more than one-third of the cases classified as likely homeless in medical examiner reports would have been shown as being housed in the vital records data. This misclassification needs to be more fully realized. As Greene noted, “Decisions to identify people as homeless are made in relation to death circumstances and are often decided from the perspectives of people who are not trained in homelessness or housing issues.”²¹

The United States currently has no national standard for counting deaths among people experiencing homelessness, and this lack of knowledge has implications for prevention efforts.²² A layered approach to protecting people currently experiencing homelessness while also supporting the end of homelessness has been proposed.²³ An important consideration in this framework is the enumeration of the true number of individuals who die while experiencing homelessness. Multnomah County has long produced reports on homeless mortality, and other local jurisdictions (eg, Maricopa County, Arizona; Los Angeles County, California; and Fulton County, Georgia) do so as well.^24-27 A review of these homeless mortality reports noted that most data sources used medical examiner or coroner reports, which are more likely to capture deaths that are apparently accidental and violent deaths compared with other causes of death.^23,27 As a result, nonaccidental and nonviolent deaths are much more likely to be undercounted in mortality reports, which may incorrectly result in a deemphasizing of other health conditions that affect mortality among people experiencing homelessness. Ample evidence indicates that people experiencing homelessness are dying early from preventable causes, such as heart disease, at rates higher than those of the unhoused population.²⁸ Although SB 850 in Oregon has improved the ability to collect data on nonviolent deaths among people experiencing homelessness,⁹ our main data source remains medical examiner reports. However, even with the risk of undercounting deaths among people experiencing homelessness, homelessness is associated with high levels of mortality, and state vital records systems remain an important source of this information.

Limitations

Our analysis had several limitations. First, our data were from a county medical examiner system and may not be applicable to other jurisdictions that have a coroner’s office, a decentralized medical examiner office, or a combination of the 2, and our methodology may not be possible to replicate in jurisdictions with different data collection systems and policies. Second, while death investigators do their best to collect as much information as possible about housing status, they may not be able to fully describe each situation. Third, classification of homeless status at death for this analysis required manual review of hundreds of case narratives, which could lead to errors in classification. However, the epidemiologist reviewing the cases is experienced in this field, used both vital records and supplemental information (eg, law enforcement reports), and collaborated with an analyst in a supporting program to check the Homeless Management Information System in unclear circumstances. Fourth, Oregon law requires the reporting of homeless status on death certificates. Although this is a strength in the context of this analysis, these results may not be generalizable to other jurisdictions. The magnitude of undercounting the number of deaths among people experiencing homelessness in states that do not require the reporting of homeless status may be even greater than what we have quantified here.

Conclusions

Despite these limitations, the data systems we used provided a unique opportunity to quantify the discordance between the recording of domicile unknown in medical examiner data and vital records data. While each of these systems may individually undercount the number of deaths that occur among people experiencing homelessness, additional misclassification of homeless status is occurring among vital records and that misclassification differs by age. It is essential that data analysts, public health programs, and policy makers who use vital statistics data recognize that they may underestimate the true count of people experiencing homelessness at death, especially among younger people. We encourage other jurisdictions to assess the accuracy of their homeless classification data, as more research is necessary to quantify the presence of misclassification. This research is imperative to validate the need for more funding and attention to public health efforts surrounding homelessness.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Jaime Walters, MPH

Eric Hall, PhD, MPH

References

de Sousa

Andrichik

Prestera

Rush

Tano

Wheeler

. The 2023 Annual Homelessness Assessment Report (AHAR) to Congress. Part 1: Point-in-Time Estimates of Homelessness. US Department of Housing and Urban Development, Office of Community Planning and Development; December 2023. Accessed November 1, 2024. https://www.huduser.gov/portal/sites/default/files/pdf/2023-ahar-part-1.Pdf

US Census Bureau. American Community Survey demographic and housing estimates. 2023. Accessed November 1, 2024. https://data.census.gov/table/ACSDP1Y2023.DP05?q=multnomahcountyoregon&y=2023

Zapata

Greene

Niehaus

, et al. 2023 Point in Time Findings Report: Count of People Experiencing Homelessness in Clackamas, Multnomah, and Washington Counties, Oregon. Homelessness Research and Action Collaborative, Portland State University; 2024. Accessed November 1, 2024. https://pdxscholar.library.pdx.edu/cgi/viewcontent.cgi?article=1047&context=hrac_pub

Portland Housing Bureau. State of Housing in Portland 2024. Part 2: Housing Market Affordability. 2025. Accessed June 23, 2025. https://www.portland.gov/phb/documents/state-housing-portland-2024-part-2-housing-market-affordability/download

Funk

Greene

Dill

Valvassori

The impact of homelessness on mortality of individuals living in the United States: a systematic review of the literature. J Health Care Poor Underserved. 2022;33(1):457-477. doi:10.1353/hpu.2022.0035

Fowle

Routhier

Mortal systemic exclusion yielded steep mortality-rate increases in people experiencing homelessness, 2011-20. Health Aff (Millwood). 2024;43(2):226-233. doi:10.1377/hlthaff.2023.01039

Waddell

Saldana

Schoonveld

, et al. Infectious diseases among people experiencing homelessness: a systematic review of the literature in the United States and Canada, 2003-2022. Public Health Rep. 2024;139(5):532-548. doi:10.1177/00333549241228525.

Meehan

Waddell

Marx

, et al. Considerations for defining homelessness in public health data collection. Public Health Rep. 2024;139(4):412-416. doi:10.1177/00333549231215850

Oregon Health Authority. Senate Bill 850: domicile unknown. Accessed January 5, 2025. https://www.oregon.gov/oha/ph/birthdeathcertificates/registervitalrecords/pages/senate-bill-850.aspx

10.

Oregon Revised Statutes. Chapter 146: investigations of deaths, injuries and missing persons. 2023. Accessed September 11, 2025. https://www.oregonlegislature.gov/bills_laws/ors/ors146.html

11.

Multnomah County Health Department. Domicile unknown: medical examiner review of deaths among people experiencing homelessness in Multnomah County in 2011. 2012. Accessed January 5, 2025. https://multco.us/file/domicile_unknown_report%3A_2011/download

12.

Cornell Law School, Legal Information Institute. 42 USC §1302: general definition of homeless individual. Accessed January 5, 2025. https://www.law.cornell.edu/uscode/text/42/11302

13.

Homeless Services Department. HMIS links for providers. Accessed September 11, 2025. https://hsd.multco.us/hmis-links-for-providers

14.

Oregon Health Authority. Oregon death certificate worksheet for funeral homes. 2023. Accessed January 5, 2025. https://www.oregon.gov/oha/ph/birthdeathcertificates/registervitalrecords/documents/death/deathcertworksheet.pdf

15.

Hodge

Jr Gostin

; CSTE Advisory Committee. Public Health Practice vs. Research: A Report for Public Health Practitioners Including Cases and Guidance for Making Distinctions. Council of State and Territorial Epidemiologists; 2004. Accessed May 7, 2025. https://ldh.la.gov/assets/docs/IRB/resources/Public_Health_Practice_vs_Research.pdf

16.

Walters

Firearm-related deaths in Multnomah County, Oregon, 2010-2016: linking medical examiner data to state vital records data. Public Health Rep. 2020;135(1):40-46. doi:10.1177/0033354919893032

17.

Walters

Mew

Repp

KK.

Transgender and nonbinary deaths investigated by the state medical examiner in the Portland, Oregon, metro area and their concordance with vital records, 2011-2021. J Public Health Manag Pract. 2023;29(1):64-70. doi:10.1097/PHH.0000000000001582

18.

Walters

Repp

Mew

MC.

Alcohol and drug presence in traffic crash fatalities before and after the COVID-19 pandemic: evaluation of the Fatality Analysis Reporting System (FARS) and linked medical examiner–vital records data in Clackamas, Multnomah, and Washington County, Oregon, 2019-2021. Forensic Sci Int Synerg. 2024;8:100468. doi:10.1016/j.fsisyn.2024.100468

19.

Walters

Lewis

Dileo

Evaluation of a suspected overdose form for timely surveillance of fatal drug overdoses, Multnomah County, OR, 2020-2021. J Public Health Manage Pract. 2024;30(4):586-592. doi:10.1097/PHH.0000000000001942

20.

Reilly

Williamson

A systematic review of the effect of stigma on the health of people experiencing homelessness. Health Soc Care Community. 2022;30(6):2128-2141. doi:10.1111/hsc.13884

21.

Greene

RN.

Marked as homeless: reconciling with ambiguities about housing status in death records. Socius. Published online September 3, 2022. doi:10.1177/23780231221121174

22.

Meehan

King

Biggs

Boyer

Berner-Davis

DiPietro

Review of local homeless mortality efforts: a call for standardized data and reporting. J Public Health Manag Pract. 2025;31(3):343-350. doi:10.1097/PHH.0000000000002070

23.

Mosites

Lobelo

Hughes

Butler

JC.

Public health and homelessness: a framework. J Infect Dis. 2022;226(suppl 3):S372-S374. doi:10.1093/infdis/jiac353

24.

Maricopa County, Office of the Medical Examiner. Medical Examiner Annual Report 2022. May 2023. Accessed May 7, 2025. https://www.maricopa.gov/ArchiveCenter/ViewFile/Item/5601

25.

Los Angeles County Department of Public Health, Center for Health Impact Evaluation. Mortality rates and causes of death among people experiencing homelessness in LA County: 2017-2023. March 2025. Accessed May 7, 2025. http://publichealth.lacounty.gov/chie/reports/Homeless_Mortality_Report_2025.pdf

26.

Bentley

. Fulton County medical examiner annual report 2022. 2023. Accessed May 7, 2025. https://www.fultoncountyga.gov/-/media/Departments/Medical-Examiner/2022-Annual-Report042224medicalexam.pdf

27.

National Health Care for the Homeless Council. Homeless mortality data toolkit: understanding and tracking deaths of people experiencing homelessness. 2021. Accessed May 7, 2025. https://nhchc.org/wp-content/uploads/2020/12/Homeless-Mortality-Toolkit-FULL-FINAL.pdf

28.

Baggett

Liauw

Hwang

SW.

Cardiovascular disease and homelessness. J Am Coll Cardiol. 2018;71(22):2585-2597. doi:10.1016/j.jacc.2018.02.077

Misclassification of Homeless Status at Death,Multnomah County,Oregon,2023

Abstract

Objective:

Methods:

Results:

Conclusions:

Keywords

Methods

Homeless Status at Death

Data Source 1: Medical Examiner Record Review

Data Source 2: Vital Records Data

Analysis

Results

Age Comparisons

Discussion

Limitations

Conclusions

Footnotes

Declaration of Conflicting Interests

Funding

ORCID iDs

References