Article Text
PDF
Abstract
Background Despite pronounced increases in homelessness and mental health problems in the USA over the past decade, further exacerbated during the pandemic, and the higher prevalence of mental health conditions among individuals experiencing homelessness, no study has examined trends in self-injury-related ED visits by individuals experiencing homelessness using up-to-date nationwide data. To address this gap, we aimed to investigate the association of self-injury-related ED visits with homelessness and to examine trends in these ED visits by individuals experiencing homelessness.
Methods We conducted a retrospective secondary data analysis using a nationally representative sample of ED visits by adults aged 25–64 years in the USA from the 2016–2021 National Hospital Ambulatory Medical Care Survey. We examined whether intentional self-injury-related ED visits and hospitalisations resulting from an ED visit were associated with homeless status using survey-weighted multivariable generalised linear regression models and whether trends in such visits changed over the study period.
Results Our analysis covered 419.4 million ED visits from 2016 to 2021. Individuals experiencing homelessness constituted 1.8% (7.4 million) of ED visits. Overall, 1.8% of ED visits (7.7 million) were related to intentional self-injuries. Nearly 1 in every 10 ED visits (9.6%) by individuals experiencing homelessness were related to self-injuries, compared with 1.7% among housed counterparts (p<0.001). The adjusted incidence rate ratio for self-injury-related ED visits was 3.14 (95% CI 2.05 to 4.83) for individuals experiencing homelessness compared with housed individuals. Finally, individuals experiencing homelessness accounted for 12.0% and 11.7% of self-injury-related ED visits in 2020 and 2021, respectively (pandemic years), compared with an average of 8.4% in the previous years.
Conclusion Among adults aged 25–64 years, experiencing homelessness was significantly associated with self-injury-related ED visits, and an increase in the rate of such visits among individuals experiencing homelessness was observed during 2020 and 2021. Future studies should assess longer-term trends in these visits and explore interventions to address the societal, health and mental healthcare needs in order to improve the health outcomes of these marginalised individuals.
- emergency department
- suicide
Data availability statement
Data are available in a public, open access repository. Data were from the National Center for Health Statistics' National Hospital Ambulatory Medical Care Survey, accessed on 10 June 2023 (https://www.cdc.gov/nchs/ahcd/about_ahcd.htm).
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
The prevalence of mental health problems and suicide ideation and attempts was about 25% higher during the pandemic compared with the prepandemic years.
Suicide rates and self-injury-related ED visits are higher among individuals experiencing homelessness compared with the general population, but studies using up-to-date and nationally representative data on this topic are limited.
WHAT THIS STUDY ADDS
Using a nationwide sample of ED visits from 2016 to 2021 among non-elderly adults in the USA, this study found that self-injury-related ED visits were significantly associated with homelessness in multivariable analyses.
This study also documented an increase in the rate of self-injury-related ED visits among individuals experiencing homelessness during 2020 and 2021, compared with 2016–2019.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
Findings highlight the importance of and the need for multilevel interventions that target individuals experiencing homelessness across all societal and healthcare domains, particularly at times of great health threats from events such as pandemics, economic recessions and extreme weather.
Introduction
More than two million people experienced homelessness in 2023 among the member countries of the Organisation for Economic Co-operation and Development, with the UK reporting the highest rates.1 The USA reached a record high of 653 100 people experiencing homelessness on a given night in 2023.1 2
Individuals experiencing homelessness across high-income countries face similar health issues and barriers to accessing housing and medical care.3 Studies have shown that individuals experiencing homelessness have a high prevalence of mental health and substance use disorders, which strongly predispose them to suicide and non-fatal self-inflicted injuries, observed across various countries and healthcare settings.4–9 Suicide ideation and attempts and non-fatal self-inflicted injuries are indicators of major distress and vulnerability.4–9 Suicide and self-harm rates among individuals experiencing homelessness were up to 10 times higher compared with the general population,9–11 and individuals experiencing homelessness were more likely to have suicide-related and self-injury-related ED visits and hospitalisations compared with their housed counterparts.11 12
Recent work found that the overall suicide and non-fatal self-injury rates have increased by about 30% over the past two decades in the USA, with a similar trend being observed in the rate of ED encounters related to suicidal ideation and attempts since EDs are the most common point of care for individuals at suicidal crises.13 14 Suicide-related ED encounters overall have increased, particularly during the pandemic2 5 15; however, the extent to which these increases were similar among individuals experiencing homelessness remains unknown. Despite the valuable contribution of the existing literature on homelessness, suicide and non-fatal self-injury-related ED visits, prior studies have used single-year or single-state data and have mostly focused on adolescents and young adults. Notably, they did not examine trends in suicide and non-fatal self-injury-related ED visits by adults experiencing homelessness, who comprise the majority of individuals experiencing homelessness. As the number of people experiencing homelessness has grown substantially over the past years, it is important to examine how ED visits, especially those related to suicide and self-injury, have changed over time.
In this study, we used a nationwide sample of ED visits from 2016 to 2021 among non-elderly adults in the USA to examine the association between suicide-related and non-fatal self-injury-related ED visits and homelessness and to explore trends in such visits over time. Our objectives were (1) to describe and compare all-cause ED visits by homelessness status (whether an individual experienced homelessness during the ED visit or not); (2) to describe and compare suicide-related and non-fatal self-injury-related ED visits by homelessness status; (3) to estimate the association between suicide-related and non-fatal self-injury-related ED visits and homelessness; and (4) to examine trends in suicide-related and non-fatal self-injury-related ED visits by individuals experiencing homelessness. We contribute to the literature by using up-to-date nationwide data that incorporate peripandemic years when overall suicide deaths in 2021 and 2022 in the country increased.16 The results of this study can assist public health efforts in addressing the societal and healthcare needs and improve the health outcomes of these marginalised individuals.
Methods
Study design, population and data source
We conducted a retrospective secondary data analysis of the 2016–2021 National Hospital Ambulatory Medical Care Survey (NHAMCS), the most recent years available at the time of the study.17 The NHAMCS is administered annually by the National Center for Health Statistics using a three-stage probability sampling design and collects information on a nationally representative sample of ED visits and resulting hospitalisations from approximately 500 EDs in non-institutional general and short-stay hospitals nationwide.17 Each participating ED is randomly assigned to a 4-week reporting period, and trained interviewers visit the sampled EDs prior to their participation to verify eligibility and explain the procedures.17 Data are collected from medical records, patient interviews and hospital electronic health records by trained personnel using standardised forms to ensure consistency and accuracy.17 Multiple sociodemographic patient-level, visit-level and facility-level characteristics are collected and weighted to produce national estimates.
We included all ED visits and hospitalisations resulting from an ED visit by non-elderly adults aged 25–64 years because these individuals represent 77% of all people experiencing homelessness and also account for more than 65% of all suicides in the USA. In addition, this approach enabled us to reduce heterogeneity in the study sample and strengthen our multivariable regression analysis at the visit level, since the non-elderly are different from children and seniors in various dimensions, particularly from a clinical standpoint.5 12 After applying survey weights, we included a total of 419.4 million ED visits across the study period by this population overall in our analyses.
Dependent variable
The main outcome variable of interest was whether the ED visit or hospitalisation through the ED was related to an intentional suicide, self-harm and non-fatal self-injury during the study period (0=no, 1=yes). To identify encounters related to suicide, suicidal thoughts, self-harm and non-fatal self-inflicted injuries (described from here forth as self-injury-related), we used five diagnoses (available as International Classification of Diseases, 10th Revision (ICD-10) codes) and five reasons for visits that were available in each ED visit, similar to previous work.11 We identified self-injury-related ED visits using the ‘5820.0’ reason for visit code, which includes reasons such as self-hanging, slashed or laceration of wrists, self-stabbing and intentional overdose for each encounter. In addition, visits where any of the available ICD-10 diagnosis codes (up to five ICD codes per visit available) included codes recommended by the Centers for Disease Control and Prevention as being suicide attempt or intentional self-harm encounters (X71–X83, R458, T36–T50, T51–T65, T71, T1491) were also classified as self-injury-related ED visits.18 Finally, to ensure the validity of our outcome, we used a variable that was available in the data which indicated whether an injury, overdose or poisoning was intentional. Only ED visits classified both as self-injury-related using the reason for visit code or the ICD-10 diagnosis codes and as intentional were flagged as being self-injury-related.
Independent variables
Our primary independent variable of interest was whether an individual experienced homelessness during the ED visit (0=no, 1=yes). The NHAMCS included information about patients’ residence, which was used to define an ED visit as a visit by an individual experiencing homelessness. Residential data recorded in the medical records were extracted by trained staff, and patients who reported having no home (eg, living on the street) or residing in a homeless shelter were combined and categorised as individuals experiencing homelessness. If the patient’s residence was coded as ‘homeless/homeless shelter’, we defined such ED visits as visits by individuals experiencing homelessness. We also used the ICD-10 Z59.0, Z59.1 and Z59.8 codes across all available diagnosis codes to further augment the identification and categorisation of ED visits by homelessness.
Covariates
We included patient-level sociodemographic, contextual and clinical characteristics and visit-level characteristics and factors in our analyses, which might differ between those experiencing homelessness and housed individuals, as well as alternatively explain self-injury-related ED visits.6 19 At the patient level, we included age; sex; race/ethnicity; the five most prevalent comorbidities in the population assessed; HIV infection/AIDS status, which is known to be disproportionately higher among those experiencing homelessness compared with the stably housed population in the USA and independently associated with suicide death; and the number of chronic conditions, defined as none, one or two, or three or more.20 We also included health insurance coverage/expected payor information: private health insurance, Medicaid (public insurance for low-income people), Medicare (public insurance for those aged 65+ and the disabled of any age), uninsured and other/unknown. Clinical characteristics related to the specific ED visit included the five most common diagnoses conditional on housing status (ie, five for housed; five for those experiencing homelessness), and whether the visit was related to injury or trauma, overdose/poisoning or adverse effect of medical/surgical treatment which was already specified and available in the data. We also included a categorical variable indicating whether the ED visit was a follow-up encounter to the same ED or not and whether the patient was seen in the same ED in the previous 72 hours, to explore healthcare-seeking and utilisation patterns which could also be associated with self-injury-related ED visits, such as unplanned revisits. The metropolitan statistical area status and the region of the hospital were also included.
Statistical analyses
We initially conducted a descriptive analysis of all-cause ED visits stratified by homelessness experience. We then conducted a similar analysis for all self-injury-related ED visits, also stratified by homelessness experience. Comparisons across variables were conducted using χ2 tests for categorical variables and t-tests or Mann-Whitney U tests for numerical variables. To estimate the association between an ED visit related to self-injuries and homelessness experience, we used a generalised linear regression model with a log-link function and a gamma distribution selected using a modified Park test, controlling for relevant patient-level and visit-level clinical covariates. We also included region and year fixed effects to control for unobserved differences (ie, the COVID-19 pandemic in 2020 and 2021).
To evaluate the robustness of our findings, we conducted two sensitivity analyses. We first replicated the multivariable analysis using the months and year of visit as time fixed effects. We did not include them in our main model due to seasonal adjustment-related issues at the monthly level per the NHAMCS documentation. We conducted an additional robustness check by excluding repeat ED visits within the previous 72 hours or ED visits that were follow-ups to the same ED to minimise selection bias.
Finally, we examined annual trends in ED visits overall and stratified by self-injuries and by homelessness experience. All analyses were adjusted for multiyear survey weights. Data management and statistical analyses were conducted in Stata V.17.0. We followed the Strengthening the Reporting of Observational Studies in Epidemiology guidelines to ensure the inclusion of all relevant information.21
Patient and public involvement
Patients or the public were not involved in the design, conduct or reporting of this study.
Results
Descriptive characteristics of all ED visits overall and stratified by homelessness experience status
Overall, our study included 419.4 million ED visits (table 1). Of these ED visits, most were by non-Hispanic white individuals (57.6%) and had Medicaid (32.1%) or private health insurance plans (29.8%) as the primary expected payor. Almost one-third (28.6%) of overall ED visits were related to injuries/trauma, overdose/poisoning or adverse effects of medical/surgical treatment. The most common diagnoses were chest pain (5.1%), abdominal pain (3.5%) and headache/migraine (2.4%), while the most common comorbidities were hypertension (28.2%), mental health and behavioural disorders (23.4%), and diabetes (14.0%).
Descriptive characteristics of ED visits overall and stratified by homelessness experience of adults 25–64 years of age from 2016 to 2021
Individuals experiencing homelessness accounted for almost 7.4 million ED visits (1.8%) (table 1). Compared with ED visits by those with housing, ED visits by individuals experiencing homelessness were disproportionately more likely to be by men (70.0% vs 44.2%, p<0.001) and Medicaid beneficiaries (55.4% vs 31.7%, p<0.001). Compared with their housed counterparts, individuals experiencing homelessness were up to 10 times more likely to have primary diagnoses and comorbidities related to alcohol or substance use-related disorders and mental health and behavioural disorders (p<0.001 for all), while HIV infection/AIDS was more than three times higher among those experiencing homelessness (2.7% vs 0.8%, p<0.001).
Descriptive characteristics of self-injury-related ED visits overall and stratified by homelessness experience status
In total, 7.7 million ED visits were related to self-injuries across the 5 years, accounting for 1.8% of all ED visits (table 2). We observed that alcohol and/or substance misuse, abuse or dependence (38.5% vs 13.5%) and mental health and behavioural disorders (52.3% vs 23.4%) were more prevalent in ED encounters related to self-injuries compared with overall ED visits. Individuals experiencing homelessness accounted for 9.3% of all self-injury-related ED visits, which was almost five times higher compared with their share of ED visits overall (1.8%).
Descriptive characteristics of self-injury-related ED visits overall and stratified by homelessness experience among adults 25–64 years of age from 2016 to 2021
Across all ED visits by individuals experiencing homelessness, 9.6% were related to self-injuries compared with 1.7% among their housed counterparts (p<0.001). Self-injury-related ED visits by individuals experiencing homelessness were more likely to be by men (80.3% vs 55.3%, p<0.001) and Medicaid beneficiaries (60.0% vs 36.7%, p<0.001), compared with visits for similar reasons by housed individuals. The prevalence of having mental health and behavioural disorders (65.0% vs 50.9%, p<0.001) and alcohol and/or substance misuse, abuse or dependence (73.8% vs 34.9%, p<0.001) was also higher in self-injury-related ED visits by those experiencing homelessness compared with housed ED users.
Trends in ED visits overall and stratified by self-injuries and by homelessness experience status
Overall and self-injury-related ED visits followed a similar trend throughout the study period and were slightly lower in 2021 compared with the previous years (figure 1). However, the share of ED visits by individuals experiencing homelessness was 52.9% and 23.5% higher in 2020 and 2021, respectively, compared with 2019 (1.7% in 2019, 2.6% in 2020, 2.1% in 2021), while the trend was relatively stable from 2016 to 2019. Similarly, among all self-injury-related ED visits, individuals experiencing homelessness accounted for 12.0% and 11.7% of these visits in 2020 and 2021, compared with an average of 8.4% between 2016 and 2019.
Trends in ED visits overall and stratified by self-injuries and by homelessness experience.
Association of self-injury-related ED visits with experiencing homelessness
The estimates of the multivariable regression analysis are presented in table 3. Overall, experiencing homelessness was independently and significantly associated with having a self-injury-related ED visit compared with being housed (incidence rate ratio (IRR): 3.14, 95% CI 2.05 to 4.83). Beyond homelessness experience, the strongest factors associated with having a self-injury-related ED visit were having mental health and behavioural disorders, having Medicare compared with private health insurance coverage, alcohol and/or substance misuse, abuse or dependence, and three or more chronic conditions compared with none. The estimates of the sensitivity analyses were similar, suggesting robust findings (online supplemental appendix table A1).
Supplemental material
Multivariable regression estimates of the associations of experiencing homelessness and other sociodemographic and clinical characteristics with self-injury-related ED visits
Discussion
Using nationally representative data across all ED visits and the resulting hospitalisations from 2016 to 2021 among adults aged 25–64 years in the USA, we found that individuals experiencing homelessness accounted for a disproportionately larger share of self-injury-related ED visits compared with their housed counterparts. They also accounted for higher shares of both overall and self-injury-related ED visits during the first 2 years of the pandemic, particularly in 2020, compared with the previous years. Our analysis shows that experiencing homelessness was associated with a higher risk of having a self-injury-related ED visit, even after controlling for various confounders.
The observed larger increase in the proportion of overall ED visits and self-inflicted injury-related ED visits by individuals experiencing homelessness overall could be partly attributed to the rising number of such individuals during the pandemic due to increased rates of unemployment and job losses, particularly after March 2020, with some states recording doubling rates in the number of individuals experiencing homelessness after 2019.2 22 This increase in self-inflicted injury-related ED visits among those experiencing homelessness during the early pandemic period could be further attributed to the exacerbated vulnerability of the new and the previous, already at-risk population of individuals experiencing homelessness due to pandemic-induced stressors.2 22 Studies using longer-term data are needed to provide additional evidence on whether the observed patterns persisted beyond 2021.
Our regression results are in line with previous work that documented up to 10 times higher rates of self-injury-related attempts and subsequent ED visits among individuals experiencing homelessness compared with their housed counterparts using various databases across different countries.4 5 8–12 This association highlights that individuals experiencing homelessness might be predisposed to suicide ideation and attempts and non-fatal self-inflicted injuries, which will often result in an ED visit. Given the existing issues for individuals experiencing homelessness in accessing needed care, particularly for mental health and substance use disorders, many are forced to seek treatment and solely rely on EDs for their care since EDs in the USA are legally required to provide treatment regardless of individuals’ ability to pay.6 19 However, ED visits are costly and not well suited to provide definitive, comprehensive or continuous care and management of individuals with long-term mental health needs and predisposition towards self-harm, particularly when these are amplified by housing issues. Alternative care delivery models specialised in mental health, such as walk-in behavioural crisis stabilisation services, might be a more cost-effective substitute for EDs and have shown promise in some US states.23
Furthermore, our results suggest that as part of provider-level and system-level efforts to address both the experience of people who are unhoused and their increased risks of self-injuries, there is a need for initiatives that overcome providers’ challenges when encountering and managing such patients in the ED. For example, evaluation and management of many individuals experiencing homelessness at risk of self-injurious behaviours will often require individually tailored approaches. Initiatives to address and prevent self-injurious behaviours among those experiencing homelessness would need to avoid stigmatising or coercive actions and respect individuals’ autonomy.22 The importance of increasing access to mental health services among adults experiencing homelessness with mental health problems is also emphasised by the ongoing objectives of the Healthy People 2030 initiative aimed at improving the health and well-being of the population over the next decade.24
Interestingly, the recently updated recommendations of the US Preventive Services Task Force endorsed screening for depression in the overall adult (19 years or older) population while reporting insufficient evidence to recommend universal screening for suicide risk, contrary to recommendations by other stakeholders.25 26 Given the co-occurrence of mental health and substance use disorders and the high self-injury risk rates among those experiencing homelessness, our findings provide evidence on the importance of screening such individuals for self-injurious behaviours, but mostly for preparing responsive healthcare systems to disclosures of self-injurious thoughts and attempts.3 27
Critically, overcoming all these issues will require policy changes that target individuals experiencing homelessness specifically and holistically across all the societal and healthcare domains.28 One-stop approaches that integrate housing, social, primary care and mental health services in a non-judgemental and supportive environment may alleviate the psychosocial burden, enable better healthcare transitions and improve health outcomes among individuals experiencing homelessness at risk of self-injuries.29 Finally, given that individuals experiencing homelessness across high-income countries face similar health issues and barriers to accessing housing and medical care, our findings also have implications for countries with different sociodemographic, contextual and healthcare system characteristics in terms of policy development to raise awareness about the link between homelessness experience and self-inflicted harm and injuries and to address the mental health challenges among those experiencing homelessness.3 Country-specific studies are warranted to investigate whether similar patterns like the ones documented in our study are observed in their populations.
Limitations
First, response or social desirability biases could result in underestimation of homelessness experience, given the identification strategy used for residential status. Nonetheless, prior studies demonstrated high accuracy in using residential addresses and diagnosis codes to identify homelessness.30 Second, we were not able to distinguish suicidal from non-suicidal self-injuries and sheltered from unsheltered individuals experiencing homelessness, which warrants further studies. Third, we also cannot assess the extent to which our estimation of the association between homelessness experience and ED visits was driven by potential ‘outliers’. Finally, it is worth noting that our analysis is at the visit level, not at the patient level, due to the lack of patient identifiers in our data. This limitation does not affect our descriptive statistics, which describe the count of the number of ED visits by homelessness status. However, since we could not account for multiple ED visits by the same patient, the SEs and 95% CIs from our regression analysis using visit-level data might be downwardly estimated, although our IRR estimates remain unaffected. To assess this bias due to within-subject correlation, we conducted a robustness check by dropping repeat ED visits within 72 hours, which yielded similar results, suggesting the bias in SEs is not a major concern.
Conclusions
In conclusion, our findings suggest that self-injury-related ED visits among adults 25–64 in the USA were significantly associated with experiencing homelessness, and an increase in the share of such visits among individuals experiencing homelessness was also observed during 2020 and 2021. The high prevalence of mental health and substance use disorders among individuals experiencing homelessness with suicidal and self-injurious behaviours highlights the need to raise awareness among healthcare providers and to holistically address the societal, health and mental healthcare needs and to improve the health outcomes of these marginalised individuals.
Data availability statement
Data are available in a public, open access repository. Data were from the National Center for Health Statistics' National Hospital Ambulatory Medical Care Survey, accessed on 10 June 2023 (https://www.cdc.gov/nchs/ahcd/about_ahcd.htm).
Ethics statements
Patient consent for publication
Ethics approval
This study involves human participants and was approved as non-human subjects’ research by the University of South Carolina Institutional Review Board (IRB) (no: Pro00128316). This is a retrospective study of de-identified publicly available data.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Handling editor Christopher Humphries
Contributors Concept and design: TG, DY. Acquisition, analysis or interpretation of data: all authors. Drafting of the manuscript: TG. Critical revision of the manuscript for important intellectual content: all authors. Supervision: TG, DY. TG had full access to all the data in the study and takes responsibility for the integrity of the data, the accuracy of the data analysis and the overall content.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.