Findings of a national dataset analysis on the visits of homeless patients to US emergency departments during 2005-2015

p u b l i c h e a l t h 1 7 8 ( 2 0 2 0 ) 8 2 e8 9

Available online at www.sciencedirect.com

Public Health journal homepage: www.elsevier.com/puhe

Original Research

Findings of a national dataset analysis on the visits of homeless patients to US emergency departments during 2005-2015 K. Lombardi a, J.M. Pines a,b, M. Mazer-Amirshahi c,d, A. Pourmand a,* a

Emergency Medicine Department, George Washington University, School of Medicine and Health Sciences, Washington, DC, United States b Department of Health Policy & Management, Milken Institute School of Public Health, George Washington University, Washington, DC, United States c Department of Emergency Medicine, MedStar Washington Hospital Center, Washington, DC, United States d Georgetown University School of Medicine, Washington, DC, United States

article info

abstract

Article history:

Objectives: To our knowledge, there has been limited description of emergency department

Received 24 February 2019

(ED) visits involving homeless patients over the last decade. Our study aims to analyze US

Received in revised form

national survey data to elucidate the differences between homeless and non-homeless

9 August 2019

patients’ ED visits in terms of patient demographics, resource utilization, and diagnoses

Accepted 6 September 2019

received. Study design: This was a retrospective study using data from the National Hospital Ambulatory Medical Care Survey (NHAMCS) from 2005 until 2015.

Keywords:

Methods: Patient visits were classified as homeless or non-homeless based on survey data;

Homeless

appropriate statistical analyses were subsequently performed to compare these groups in

Emergency department

terms of patient demographics, geography, payment method, resource utilization/diag-

Substance abuse

nostic service use, as well as both psychiatric and non-psychiatric diagnoses received in

Alcohol abuse

the ED.

Schizophrenia

Results: NHAMCS data from 2005 to 2015 were aggregated. In total, 303,326 patient visits were included, which represent an estimated 1.30 billion ED visits over this period. Of these, 2750 encounters were by homeless people, representing 8,781,925 ED visits. Compared with non-homeless visits, homeless patients were disproportionately male, black, non-Hispanic, and seen in large metropolitan areas or the Western/Southern US. Homeless visits were more likely to be related to an injury (47.5% vs. 33.8%), related to an assault (4.2% vs. 1.3%), or self-inflicted (4.8% vs 0.84%). Homeless patients were also more likely to have been seen in the same ED within 72 h (7.3% vs. 3.9%) compared with nonhomeless patients (3.9%, 95% confidence interval [CI]: 3.5e4.4) and were seen an average of 5.7 times (95% CI: 4.7e6.8) in the same ED over the preceding 12 months, with nonhomeless patients seen an average of 3.2 times (95% CI: 3.1e3.4). Homeless patients were more likely to be admitted to the hospital (14.9% vs. 11.2%) and, when admitted, spent an average of 6.3 days in the hospital (95% CI: 5.6e7.1) compared with non-homeless patients

* Corresponding author. Department of Emergency Medicine, George Washington University, Medical Center 2120 L StWashington, DC, 20037, United States. E-mail address: [email protected] (A. Pourmand). https://doi.org/10.1016/j.puhe.2019.09.003 0033-3506/© 2019 The Royal Society for Public Health. Published by Elsevier Ltd. All rights reserved.

p u b l i c h e a l t h 1 7 8 ( 2 0 2 0 ) 8 2 e8 9

83

at 5.2 (95% CI: 5.1e5.3). In total, 28.4% of homeless patients received a psychiatric diagnosis (95% CI: 25.8e31.2) compared with 5.4% for non-homeless patients (95% CI: 5.2e5.7, P < 0.001). In reference to non-homeless visits, homeless visits showed increased odds of alcohol-related diagnoses (odds ratio [OR]: 17.3, 95% CI: 10.1e29.8, P < 0.001) and substance abuse diagnoses (OR: 8.4, 95% CI: 7.2e9.8, P < 0.001). Homeless visits also exhibited greatly increased odds of diagnosis of schizophrenia (OR: 16.6, 95% CI: 12.6e22.5, P < 0.001) and personality disorders (OR: 15.4, 95% CI: 6.4e36.9, P < 0.001). Conclusions: Less than one in 100 US ED visits in 2005e2015 were made by homeless patients. Compared with the non-homeless, homeless patients had greatly increased rates of ED care for alcohol-related, substance abuseerelated, and mental healtherelated problems, particularly schizophrenia and personality disorders. Homeless patients were also more likely to be seen in the ED within the past 72 h or the past 12 months. Homeless patients were more likely to be admitted to the hospital and, when admitted, exhibited longer stay times. © 2019 The Royal Society for Public Health. Published by Elsevier Ltd. All rights reserved.

Introduction Background Homelessness in the United States is a serious societal and public health concern and significant contributor to emergency department (ED) visits and other healthcare use. According to the US Department for Housing and Urban Development, approximately 550,000 people in 2016 in the US or 0.2% of the US population are homeless.1 In addition, the National Alliance to End Homelessness estimates that up to seven million Americans are currently in living arrangements, such as shared housing, which places them at risk for homelessness.2 Previous research regarding homeless patients seen in the ED has found that individuals experiencing homeless receive considerably less access to preventive and primary healthcare services than the general population. Consequentially, these patients turn to EDs as a primary source of healthcare services.3 Homeless patients are also at higher risk of being frequent users of EDs, meaning that often ED use is not a single event but rather a series of ED visits to different hospitals.4e6 As a population, homeless patients have higher risk of alcohol and other substance use,7,8 with some evidence indicating that up to half of homeless Americans are dependent on alcohol.9 Compounding this, patients experiencing homelessness also have increased rates of behavioral and mental health conditions7,8 which when co-occurring with substance use can lead to significantly increased rates of ED service use and costs.4 Homeless patients with mental health conditions are also more likely to return to the ED within 30 days of discharge and more likely to be readmitted to the hospital.10 A recent study of 755 chronic homeless persons indicated that in a three-month period, 30% visited the ED one or two times and 12% visited three or more times.11 Analyses of Medicaid administrative data have indicated that alcohol-related disorders were the most common reason homeless patients accessed EDs.12 Other studies have indicated tobacco and drug use accounted for more than half of all deaths of homeless

persons.13 Studies of alcohol-dependent homeless patients have indicated a 34% rate of ED use and increased rates of ED use for rape, assault, overdose, and mental health complications.14

Importance Prior studies have used national data from the National Hospital Ambulatory Care Survey (NHAMCS) to study ED visits by homeless patients. Of these studies, Ku et al.15 and Oates et al.16 used data sets from years 2005 and 2005e2006 and identified that the homeless patient population was more likely to be male, be older in age, arrive to the ED via ambulance, and be uninsured; require more than two diagnostic tests; leave against medical advice; or be seen by an intern or resident. Hammig et al.17 used 2007e2010 NHAMCS data sets and indicated that homeless patients had higher odds of presenting with self-inflicted or assault-related injuries. Analyses of 2010 data by Ayangbayi et al.18 have indicated that this patient population was more likely than the general population to present for mental health and substance use disorders. In addition, research by Albert and McCaig19 regarding patients with schizophrenia using 2010 data concluded that these patients were more likely to be homeless. Our study aims to analyze US national survey data to elucidate the differences between homeless and nonhomeless patients’ ED visits in terms of patient demographics, resource utilization, and diagnoses received.

Methods Study design, setting, and study population This was a retrospective study using data from the NHAMCS from 2005 until 2015. The NHAMCS is conducted on an annual basis by the National Center for Health and Statistics, a department of the Centers for Disease Control (CDC). The NHAMCS is a national survey designed to meet the need for objective, reliable information about the provision, and use of

84

p u b l i c h e a l t h 1 7 8 ( 2 0 2 0 ) 8 2 e8 9

ambulatory medical care services in US EDs. The survey uses a multistage probabilistic sample of hospital-based ED visits. The survey collects specific data related to patient demographics and resource utilization including, but not limited to, patient sex, age, residence, diagnostic services provided, payment method, and medications administered (in the ED) and prescribed (from the ED) during the patient encounter. NHAMCS survey design includes detailed survey weights to estimate national estimates from the sample data it makes available. The NHAMCS has a publicly available data set, which does not contain any information that could be used to identify any particular patient, and as such, it does not qualify as human subjects’ research.

patients compared with the general population, bivariate logistic regressions were performed and odds ratios subsequently calculated, with homeless visits serving as the dependent variable and non-homeless as the reference variable. To perform this for variables coded as numerical or string in the NHAMCS, responses were dichotomized to ‘yes’ or ‘no.’ In accordance with NCHS guidelines, an estimate was considered to be reliable if it was derived from 30 or more records and had a relative standard error of 30% or less.23 All analyses were performed using STATA, version 15.0, statistical analysis software (StataCorp LP, College Station, TX).

Results Study protocol Demographics NHAMCS survey data from years 2005e2015 were combined into a single data set. Homelessness was defined through the NHAMCS residence variable. In this section, cases are classified as homeless if they ‘have no home’ (e.g., lives in a public space) or the current place of residence was a homeless shelter.20 These data are extracted by CDC field staff based on the reported housing status in individual facility-based medical records. Other available responses for this variable include private residence, nursing home, other institution, other, or unknown. The option ‘other institution’ was only available in years 2007 and 2008. Measures of accuracy of the NHAMCS residence variable in appropriately identifying homeless individuals have not been described. However, prior studies that defined variables similarly to the NHAMCS residence variable have been validated by chart review, matching residential addresses to homeless shelters.21,22 For the purposes of this study, and similar to prior research regarding homelessness using the NHAMCS, all other housing groups were collapsed to a single comparison group, referred to in this article as the ‘non-homeless population.’ Comparison between visits classified as homeless and non-homeless were made for variables intrinsic to the NHAMCS survey describing patient demographics (gender, race, ethnicity) as well as geographic region and metropolitan area size and payment method. These comparisons were also performed for variables detailing the visit content details and diagnostic service use as well as patient medical history.

Analysis To account for survey clustering and produce unbiased national estimates, sample weights designed by the National Center for Health Statistics (NCHS) to adjust for geographic region, urban/rural/ownership designations, and nonresponse were applied to the aggregated data. To identify the most common primary diagnoses, cross-tabulations were performed between the variables residence and diagnoses applied during homeless and non-homeless visit. This was also performed between residence and payment methods to identify the most common methods of remuneration for each residence type. Differences were assessed using the Chisquared test and pairwise contrasts, with unpaired t-tests with an alpha of 0.05 as the threshold for statistical significance. To quantify the relative odds of outcomes for homeless

Between 2005 and 2015, a total of 303,326 visits were included in the NHAMCS, and these were representative of more than 1.30 billion visits to US EDs during this time. Of these, 2750 data entries were homeless persons, which are representative of a national estimate of 8,781,925 ED visits during this period. The proportion of homeless of the total ED population ranged from 0.38% in 2005 to a peak of 0.88% in 2011, with a general upward trend. Table 1 illustrates bivariate relationships between homeless status and variables related to patient demographics, visit content details, and diagnostic service utilization. Compared with the non-homeless population, ED homeless visits were more likely to involve patients who were male, black, non-Hispanic and to visit an ED in the South or West regions of the US. Homeless ED visits were also comparatively more likely to occur at facilities in large central metropolitan areas. Regarding the method of payment for ED services, at only 6.2%, homeless patients were substantially less likely to have private insurance than the general population. Another notable difference was the 28.5% rate of selfpay for services among homeless vs 16.8% for the general population. Homeless ED visits had similar rates of insurance by Medicaid as other the non-homeless (Table 1).

Resource utilization Homeless patients were more likely to be admitted to the hospital than patients in the non-homeless population. In addition, they showed increased rates of admission to the hospital and, when admitted, stayed an average of one day longer than non-homeless patients. Homeless ED visits were less likely to receive imaging of any kind, which included Xray, computed tomography, and magnetic resonance imaging. They showed similar rates of general chemistry laboratory values and urinalysis resource utilization as the general population. Homeless visits also received comparable average numbers of diagnostic procedures and non-diagnostic procedures (Table 2). The analysis of injury-related ED visits by homeless patients demonstrated notable distinguishing features. Nearly half of all visits by homeless patients were related to an injury of any type, a rate that was notably higher than that of non-homeless visits. Overall, homeless patients exhibited a similar average number of chronic conditions as the non-homeless (Table 2).

85

p u b l i c h e a l t h 1 7 8 ( 2 0 2 0 ) 8 2 e8 9

Table 1 e Emergency department demographic characteristics of homeless and non-homeless adult ED visits 2005e2015. Item

Homeless Number (thousands) percent

Total visits Age (years)

Gender Race

Ethnicity Geography

Metropolitan area

Insurance

18e25 26e44 45e64 > 64 Male Female White Black Other Hispanic Non-Hispanic Northeast Midwest South West Large central metro area Large fringe metro area Medium metro area Small metro area Non-metro area Private insurance Medicare Medicaid Workers compensation Self-pay No charge Unknown/other

796 3221 3447 814 6161 2621 6196 2254 331 1101 7681 1634 1691 2347 3110 1193 514 572 572 153 1634 991 2859 83 1823 587 1088

9.1 36.7 39.3 9.3 70.2 29.9 70.6 25.7 3.8 12.5 87.5 18.6 19.3 26.7 35.4 35.6 15.3 17.0 4.6 11.5 11.3 17.5 32.6 0.94 20.8 6.7 10.9

Non-homeless 95% CI 7.5e10.9 34.1e39.2 36.7e41.8 7.7e11.2 67.3e72.8 27.2e72.8 67.4e73.5 22.9e28.7 2.7e5.2 10.6e14.8 85.3e89.4 15.2e22.6 14.9e24.5 22.8e31.1 31.3e39.7 29.3e42.4 11.3e20.4 12.6e22.8 2.7e7.6 7.7e17.1 8.5e14.9 15.2e20.0 29.2e36.1 0.39e2.3 18.1e23.7 5.0e8.8 8.7e13.5

Number (thousands) percent 156,738 365,170 276,959 196,898 581,687 708,728 939,357 303,834 47,225 231,548 1,113,515 231,547 298,025 502,316 258,527 136,537 100,719 125,119 35,001 88,611 478,720 230,685 384,692 15,073 207,794 17,393 48,874

12.2 28.3 21.5 15.3 45.1 54.9 72.8 23.6 3.7 13.7 86.3 17.9 23.1 38.9 20.0 27.0 19.9 24.8 6.9 17.5 37.1 17.9 29.8 1.2 3.7 1.4 5.7

95% CI 11.2e12.4 27.9e28.7 21.1e21.9 14.8e15.8 44.7e45.4 54.6e55.3 70.7e74.8 21.6e25.6 3.2e4.2 12.5e15.1 84.9e87.5 15.7e20.5 20.2e26.5 35.1e42.9 17.5e22.8 23.0e31.4 17.2023.0 18.6e32.2 4.4e10.7 12.0e24.9 36.2e38.1 17.3e18.5 28.7e30.9 1.1e1.3 36.2e38.1 1.1e1.7 4.9e6.7

CI, confidence interval; ED, emergency department.

Homeless and non-homeless patients received a comparable average number of medications in the ED and at discharge. However, non-homeless patients were substantially more likely to be administered an opioid analgesic during their stay, or as their first medication, and they also showed higher rates of receiving opioid analgesic medications at discharge. Homeless patients were less likely to receive a non-steroidal anti-inflammatory drug (NSAID) and more than twice as likely to receive a benzodiazepine (Table 2).

ED diagnoses Analyses of diagnoses given during homeless and nonhomeless visits demonstrated significantly divergent patterns, with homeless patients receiving diagnoses related to psychiatric illness and substance abuse at rates that greatly exceeded that of the non-homeless while experiencing diagnoses quite common in the non-homeless visits at very low levels. For non-homeless visits, the most common diagnoses were psychiatric-related and particularly those related to alcohol or substance use/abuse. Of note, nearly 30% of homeless visits received a psychiatric diagnosis of any type, a rate nearly six times that of the non-homeless (Table 3). In addition, nearly 20% of homeless patients during ED visits received diagnoses related to alcohol or substance use/abuse, a rate which was also approximately six times that of the nonhomeless patients. The most common diagnosis given during homeless ED visits during the study period was non-

dependent use of drugs at nearly 16% of these visits, a rate which was nearly eight times that of non-homeless visits (Table 3). Logistic regression analysis results further indicated the magnitude of these trends, with results indicating greatly increased odds, at times an order of magnitude or greater with which homeless visits received these diagnoses (Table 4).

Discussion The CDC has collected data regarding housing status through the NHAMCS survey since 2005; with the release of the 2015 data, the analyses in this study represent the most up-to-date and comprehensive NHAMCS analyses on this subject. Through this, we sought to update, augment, and add depth to prior research. Past analyses of the 2005 and 2006 NHAMCS survey by Ku et al.15 indicated that homeless patients were more likely to be uninsured; treated for acute injury, alcohol abuse or other drug use, and psychiatric issues; and more likely to have been seen in the ED within the past 72 h.15 These dynamics, as well as the preponderance of substance abuse and psychiatric diagnoses in the homeless population, has also been confirmed in studies that examined previous years of the NHAMCS.3,19,24,25 However, our analyses also noted the significant magnitude of these relationships, particularly the large discrepancy between homeless and non-homeless visits regarding alcohol dependence syndrome, non-dependent abuse of drugs, and schizophrenic psychoses. These data

86

p u b l i c h e a l t h 1 7 8 ( 2 0 2 0 ) 8 2 e8 9

Table 2 e Visit content details, medication administration, and resource utilization among homeless and non-homeless adult ED visits 2005e2015. Item

Visit content details

Resource utilization

Medications

Patient medical historya

Homeless

Visit injury related Visit assault related Injury self-inflicted Average pain scale at triage Seen by attending physician Seen by resident/intern Seen by advanced practice provider No follow-up planned Left against medical advice Seen in the same ED as the last 72 h Times in the same ED in the past 12 months Admitted to hospital Wait time to see provider (minutes) Time boarded in minutes (if boarded) Length of visit in the ED (minutes) Length of stay in days (if admitted) Died in the ED Average total diagnostic procedures Average total procedures in the ED Any blood chemistry analysesa Any imaging X-ray MRI CT Given opioid analgesic Given opioid analgesic as the first medication Given opioid at discharge from the ED Given NSAIDs Given benzodiazepines Number of medications given Number of medications given at discharge Triage level Total chronic conditions Cancer Chronic obstructive pulmonary disease Congestive heart failure Coronary artery disease Cerebral vascular disease/stroke Type two diabetes Depression Dementia Patient transferred from another facility

Non-homeless

Percent/ number

95% CI

Percent/ number

95% CI

47.5 4.2 4.8 7.3 86.6 15.9 14.3 9.2 1.9 7.3 5.7 14.9 62.5 132.3 265.6 6.3 0.0094 4.9 2.3 47.3 38.5 29.3 0.35 14.3 16.7 8.1 6.1 18.3 8.5 2.5 2.0 3.3 2.1 2.0 6.3 1.4 0.33 1.3 4.1 22.6 0.15 0.26

44.4e50.1 3.2e5.3 3.7e6.1 6.9e7.7 84.1e88.9 13.3e18.9 11.9e17.0 7.5e11.0 1.2e2.8 6.0e8.8 4.7e6.8 13.0e16.9 56.7e68.4 84.9e180.7 251.4e280.0 5.6e7.1 0.0017e0.0053 4.6e5.3 1.5e3.1 44.0e50.5 35.6e64.4 26.7e32.0 0.17e0.69 12.6e16.3 14.6e18.9 6.7e9.8 4.9e7.5 16.0e20.8 7.1e10.2 2.3e2.6 1.8e2.2 3.2e3.4 1.8e2.3 1.1e3.5 4.1e9.5 0.82e2.2 0.16e0.70 0.82e2.2 1.9e8.7 16.9e29.7 0.02e0.11 0.071e0.97

33.8 1.3 0.84 6.9 87.2 8.9 16.3 8.1 1.0 3.9 3.2 11.2 51.3 115.7 201.3 5.2 0.0051 4.7 2.0 38.1 45.9 34.0 0.64 14.9 24.3 12.1 11.8 23.6 3.9 2.5 1.9 3.3 2.0 3.1 4.3 2.9 5.5 2.7 3.9 8.8 1.2 0.79

33.3e34.2 1.2e1.4 0.78e0.93 6.8e7.0 86.0e88.3 7.8e10.7 15.0e17.6 7.2e9.0 0.093e1.1 3.5e4.4 3.1e3.4 10.6e11.9 49.3e53.3 109.8e121.6 196.8e205.8 5.1e5.3 0.0043e0.0062 4.5e4.8 1.8e2.2 37.2e39.0 45.1e46.6 33.3e34.6 0.58e0.71 14.4e15.4 23.5e25.2 11.5e12.6 11.3e12.5 22.6e23.5 3.7e4.1 2.4e2.5 1.8e1.9 3.1e3.4 1.9e2.1 2.8e3.3 4.0e4.7 2.7e3.1 4.9e6.2 2.4e2.9 3.3e4.6 8.0e9.7 0.1e1.3 0.62e0.01

Pvalue <0.001 <0.001 <0.001 <0.001 0.26 <0.001 0.005 <0.001 <0.001 <0.001 <0.001 <0.001 0.04 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 0.61 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 0.25 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001

CI, confidence interval; CT, computed tomography; ED, emergency department; MRI, magnetic resonance; NSAIDs, non-steroidal anti-inflammatory drugs. a Only available from 2009 to 2015.

also indicated that although homeless visits were overrepresented in certain non-psychiatric/behavioral diagnoses, they were comparatively less likely to utilize the ED for diagnoses which were otherwise common for non-homeless visits such as upper respiratory infections, acute pharyngitis, or cardiac dysrhythmia. The disproportionate use of diagnostic services by the homeless population has also been identified in previous studies using the NHAMCS.16 The clearest conclusion that augments past studies is the centrality and magnitude of alcohol and substance abuse as well as mental illness to the homelessness epidemic. Although past research using the NHAMCS has noted similar

trends, our analyses elucidated their magnitude over the life span of the collection of these data and with a more detailed scope. Of particular interest were the substantially greater odds at which homeless patients during ED visits were diagnosed with conditions related to alcohol abuse, substance abuse, and mental/behavioral health concerns than the nonhomeless. These disparities were echoed by the increased rate in which liver function tests were performed and blood alcohol levels were measured for homeless visits, as well as the greatly increased rate of benzodiazepine administration. Considering past research that has indicated the high financial burden of providing services related to these health

87

p u b l i c h e a l t h 1 7 8 ( 2 0 2 0 ) 8 2 e8 9

Table 3 e Frequency of common psychiatric and non-psychiatric diagnoses for homeless and non-homeless ED visits 2005e15. Item

Homeless

Non-homeless

P-value

Percent

95% CI

Percent

95% CI

28.4 18.8 1.2 0.97 17.7 1.3 1.8 15.7 2.1 3.4 3.1 0.28 3.2 0.42

25.8e31.2 16.6e21.2 0.74e2.0 0.53e1.8 15.7e20.0 0.79e2.2 1.3e2.5 13.8e17.7 1.3e3.4 3.0e5.1 2.3e4.3 0.01e0.74 2.2e4.6 0.19e0.91

5.4 2.7 0.079 0.065 2.6 0.17 0.18 2.3 0.40 0.26 0.59 0.037 1.5 0.029

5.2e5.7 2.5e2.9 0.064e0.097 0.051e0.083 2.4e2.8 0.15e0.19 0.16e0.21 2.2e2.5 0.39 0.24e0.29 0.54e0.64 0.028e0.049 1.4e1.6 0.022e0.040

<0.001 <0.001 <0.001 <0.001 <0.001 0.0104 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 0.0109

5.7 0.94 0.51 0.33 0.11 0.28 0.69 0.12 0.84

4.6e7.2 0.57e1.6 0.22e1.1 0.17e0.70 0.036e0.30 0.13e0.61 0.32e1.5 0.36e0.39 0.45e1.6

4.4 1.1 0.77 0.95 0.91 0.79 0.064 0.21 0.46

4.2e4.5 0.98e1.1 0.71e0.84 0.88e1.0 0.85e0.96 0.73e0.85 0.048e0.085 0.19e0.24 0.42e0.51

<0.001 0.005 <0.001 <0.001 <0.001 <0.001 0.006 0.06 0.20

Diagnoses e psychiatric/behavioral All psychiatric diagnoses All alcohol or substance related All alcohol related Alcohol dependence syndrome All drug use related Drug use psychoses Substance dependence syndromes Non-dependent abuse of drugs All organic psychoses Schizophrenic psychoses Affective psychoses Paranoid states Neurotic disorders Personality disorders Diagnoses e non-psychiatric/behavioral General symptoms Asthma Acute bronchitis Acute upper respiratory tract infection Acute pharyngitis Arrhythmia Acute myocardial infarction Transient cerebral ischemia Diabetes mellitus CI, confidence interval; ED, emergency department.

issues, the need for alternative methods of treating and managing them in homeless patients is clear.4,5 In addition to confirming the results of previous studies, this analysis indicated that homeless patients seen in the ED were less likely to present for or be diagnosed with symptoms or medical conditions that were otherwise quite common in the general population. These trends are particularly concerning because of their magnitude and in the absence of data that suggests that homeless patients are protected from these conditions. This may indicate that this population is disinclined to seek service for these otherwise common conditions and may be brought to the ED involuntarily or when found to be publicly intoxicated or influenced by substance abuse. This also appears to at least partially contradict established perceptions that homeless patients overuse the ED for primary care services, as all of the diagnoses and reasons for visit which were most under-represented in the homeless population are also exceedingly common in outpatient primary care services.26 The difference in provision of diagnostic services between homeless patients and the general population showed a much more benign magnitude of difference than reason for visit and primary diagnosis. Common laboratory tests were found to be used more frequently for homeless patients and the general population, while diagnostic services such as blood alcohol levels, glucose, and liver function tests were much more likely to be used in homeless patients, presumably due to the high odds of these patients presenting for issues relating to alcohol

and other substance use. Despite this, as well as results that indicated that homeless patients demonstrated similar triage levels and pain scales at presentation at the ED compared with the non-homeless, they demonstrated notably higher levels of admission to the hospital, as well as longer stays when admitted. Furthermore, the magnitude of homeless patients who did not have their services paid for by any official means lends credence to the particular financial burden that providing ED services to this population has been shown to have on facilities and to the price of both ED and non-ED medical services. Results of our analyses regarding medication analysis indicated that homeless patients during ED visits were administered opioid analgesics and NSAIDs at rates that were lower than for non-homeless patients. Although this may be explained by the results detailing the different diagnoses received by homeless patients, further research should be performed to examine if pain is appropriately treated in this ED population. Of similar concern, homeless visits exhibited consistently longer wait times to see provider, overall time spent in the ED, and time boarded in the ED.

Limitations In general, the NHAMCS data are extracted from hospital medical records based on data self-reported by patients. However, depending on the enrollment procedures of individual facilities, these data may not be entered during the patient's visit and thus may be prone to inaccuracy. These

88

p u b l i c h e a l t h 1 7 8 ( 2 0 2 0 ) 8 2 e8 9

Table 4 e Likelihood of select diagnoses and visit content characteristics in homeless visits in reference to nonhomeless in US EDs 2005e2015. Item Primary diagnoses e psychiatric All psychiatric diagnoses All alcohol or substance related All alcohol related Alcohol dependence syndrome All drug use related Drug use psychoses Substance dependence syndromes Non-dependent abuse of drugs All organic psychoses Schizophrenic psychoses Affective psychoses Paranoid states Neurotic disorders Personality disorders

OR

95% CI

p

7.1 8.8 17.3 16.6 8.4 8.3 10.4 8.10 5.2 16.8 5.6 7.7 2.2 15.4

6.2e8.1 7.6e10.2 10.1e29.8 8.6e31.6 7.2e9.8 4.9e14.3 6.8e15.8 6.9e9.3 3.2e8.3 12.6e22.5 4.1e7.6 2.7e22.0 1.5e3.2 6.4e36.9

<0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001

Diagnoses e non-psychiatric General symptoms Asthma Acute bronchitis Acute upper respiratory tract infection Acute pharyngitis Arrhythmia Acute myocardial infarction Transient cerebral ischemia Diabetes mellitus

1.3 1.0e1.7 0.02 0.9 0.54e1.5 0.69 0.65 0.28e1.5 0.31 0.34 0.17e0.69 0.002 0.11 0.038e0.32 <0.001 0.36 0.16e0.78 0.009 3.4 1.6e7.5 0.002 0.333 0.55 0.17e1.8 1.8 0.99e3.5 0.054

Visit content details Seen in the same ED within 72 hrs Injury related Assault related Self-inflicted injury Admitted

1.8 1.9 3.1 5.3 1.2

1.5e2.3 1.6e2.1 2.3e4.1 4.0e6.9 1.0e1.4

<0.001 <0.001 <0.001 <0.001 0.05

CI, confidence interval; ED, emergency department.

facts also make it difficult to assess the true incidence of homelessness seen in US EDs as patients may be disinclined to accurately report their housing status because of embarrassment, lack of awareness regarding the nature of their housing status, or the transient nature of their housing. The information may also have not been collected during their visit. Furthermore, the NHAMCS does not adjust when a person visits the ED and is included in the survey multiple times, as homeless patients have been seen to be more likely to frequently visit the ED; this serves as a potential source of error.14 The NHAMCS defines a person as homeless if the patient ‘has no home’ (e.g., lives in a public space) or the patient's current place of residence was a homeless shelter.20 This definition does not include persons who are in arrangements such as shared housing, whom some researchers consider to be homeless. This variable also excludes some members of the population who may be transiently housed or who are at risk of homelessness, a population which is estimated to be up to seven million.2 US military veterans are represented disproportionately in the homeless population, with national estimates that they represent at a rate of 24.8 per 10,000 as opposed to a rate of 18.3 per 10,000 in the general population.2

Conclusions When compared with the general population, homeless patients were found to have much greater rates of service for alcohol-related, substance abuseerelated, and mental healtherelated problems. Hospital staff must be cognizant of these comorbidities when treating homeless patients, and patients often require mental health, substance abuse, and social service referral. Ultimately, EDs may need additional resources to optimally care for these socially and medically complex patients.

Author statements Ethical approval This study used publicly available database. Thus, it does not require ethical approval.

Funding This study was not funded.

Competing interest The authors have no conflicts of interest to declare.

references

1. Henry M, Watt R, Rosenthal, et al. The 2016 Annual Homeless Assessment Report (AHAR) to congress. 2016. https://files. hudexchange.info/resources/documents/2016-AHAR-Part-1. pdf. [Accessed 19 December 2018]. 2. National alliance to end homelessness. The State of Homelessness in America. 2016. http://endhomelessness.org/ wp-content/uploads/2016/10/2016-soh.pdf. [Accessed 10 December 2017]. 3. Brown RT, Steinman MA. Characteristics of emergency department visits by older versus younger homeless adults in the United States. Am J Public Health 2013;103(6):1046e51. https://doi.org/10.2105/AJPH.2012.301006.  n CLK, Byrne TH, Lin W, Bharel M. Cost of 4. Mitchell MS, Leo health care utilization among homeless frequent emergency department users. Psychol serv 2017;14(2):193e202. 5. Adams JG. Emergency department overuse. JAMA 2013;309(11):1173e4. 6. Jackson TS, Moran T, Lin J, Salhi BA. Prevalence of homelessness and housing insecurity in an urban emergency department. Ann Emerg Med 2017;70(4):S167. https://doi.org/ 10.1016/j.annemergmed.2017.07.307. 7. Salhi BA, White MH, Pitts SR, Wright DW. Homelessness and emergency medicine: a review of the literature. Acad Emerg Med 2018;25(5):577e93. https://doi.org/10.1111/acem.13358. 8. Doran KM, Rahai N, McCormack RP, et al. Substance use and homelessness among emergency department patients. Drug Alcohol Depend 2018;188(February):328e33. https://doi.org/ 10.1016/j.drugalcdep.2018.04.021. 9. Fazel S, Khosla V, Doll H, Geddes J. The prevalence of mental disorders among the homeless in Western countries: systematic review and meta-regression analysis. PLoS Med

p u b l i c h e a l t h 1 7 8 ( 2 0 2 0 ) 8 2 e8 9

10.

11.

12.

13.

14.

15.

16.

17.

2008;5(12):1670e81. https://doi.org/10.1371/ journal.pmed.0050225. Lam CN, Arora S, Menchine M. Increased 30-day emergency department revisits among homeless patients with mental health conditions. West J Emerg Med 2016;17(5):607e12. https://doi.org/10.5811/westjem.2016.6.30690. Moore DT, Rosenheck RA. Factors affecting emergency department use by a chronically homeless population. Psychiatr Serv 2016;(December). https://doi.org/10.1176/ appi.ps.201500526. appips201500526. Lin WC, Bharel M, Zhang J, O'Connell E, Clark RE. Frequent emergency department visits and hospitalizations among homeless people with medicaid: implications for medicaid expansion. Am J Public Health 2015;105:S716e22. https:// doi.org/10.2105/AJPH.2015.302693. Baggett TP, Chang Y, Singer DE, et al. Tobacco-, alcohol-, and drug-attributable deaths and their contribution to mortality disparities in a cohort of homeless adults in Boston. Am J Public Health 2015;105(6):1189e97. https://doi.org/10.2105/ AJPH.2014.302248. Holtyn AF, Jarvis BP, Subramaniam S, et al. An intensive assessment of alcohol use and emergency department utilization in homeless alcohol-dependent adults. Drug Alcohol Depend 2017;178(June):28e31. https://doi.org/10.1016/ j.drugalcdep.2017.04.025. Ku BS, Scott KC, Kertesz SG, Pitts SR. Factors associated with use of urban emergency departments by the U.S. homeless population. Public Heal Reports 2010;125(3):398e405. Washington, DC 1974, http://content.ebscohost.com/ ContentServer.asp?T¼P&P¼AN&K¼20433034& S¼R&D¼mnh&EbscoContent¼dGJyMNLe80 SeprQ4zdnyOLCmr0mep7RSr6y4SLeWxWXS&Content Customer¼dGJyMPGrtE%2Bwp7dMuePfgeyx44Dt6fIA% 5Cnhttp://search.ebscohost.com/login.aspx? direct¼true&db¼mnh&AN¼20433. Oates Gary, Tadros Allison, Davis Stephen M. A comparison of national emergency department use by homeless versus nonhomeless people in the United States. J Health Care Poor Underserved 2009;20(3):840e5. https://doi.org/10.1353/ hpu.0.0192. Hammig B, Jozkowski K, Jones C. Injury-related visits and comorbid conditions among homeless persons presenting to

18.

19.

20.

21.

22.

23.

24.

25.

26.

89

emergency departments. Acad Emerg Med 2014;21(4):449e55. https://doi.org/10.1111/acem.12343. Ayangbayi T, Okunade A, Karakus M, Nianogo T. Characteristics of hospital emergency room visits for mental and substance use disorders. Psychiatr Serv 2016;(April). https://doi.org/10.1176/appi.ps.201600125. appi.ps.2016001. Albert M, McCaig LF. Emergency department visits related to schizophrenia among adults aged 18-64: United States, 20092011. NCHS Data Brief 2015;(215):1e8. Centers for Disease Control and Prevention. NAMCS/ NHAMCS - questionnaires, datasets, and related documentation. https://www.cdc.gov/nchs/ahcd/ahcd_ questionnaires.htm. Accessed December 11, 2017. Tangherlini N, Pletcher MJ, Covec MA, Brown JF. Frequent use of emergency medical services by the elderly: a case-control study using paramedic records. Prehospital Disaster Med 2010;25(3):258e64. http://www.ncbi.nlm.nih.gov/pubmed/ 20586020. [Accessed 11 December 2017]. Kramer CB, Gibran NS, Heimbach DM, Rivara FP, Klein MB. Assault and substance abuse characterize burn injuries in homeless patients. J Burn Care Res 2008;29(3):461e7. https:// doi.org/10.1097/BCR.0b013e31817112b0. National Center for Health Statistics TC for DC and P. NAMCS/ NHAMCS - ambulatory health care data homepage. https:// www.cdc.gov/nchs/ahcd/index.htm. Accessed December 11, 2017. Tadros A, Layman SM, Brewer MP, Davis SM. A 5-year comparison of ED visits by homeless and non-homeless patients. Am J Emerg Med 2016. https://doi.org/10.1016/ j.ajem.2016.01.012. Hammig B, Jozkowski K, Jones C. Injury-related visits and comorbid conditions among homeless persons presenting to emergency departments. In: Cunningham RM, editor. Acad Emerg Med, 21; 2014. p. 449e55. https://doi.org/10.1111/ acem.12343. Trudnak T, Kelley D, Zerzan J, Griffith K, Jiang HJ, Fairbrother GL. Medicaid admissions and readmissions: understanding the prevalence, payment, and most common diagnoses. Health Aff 2014;33(8):1337e44. https://doi.org/ 10.1377/hlthaff.2013.0632.