a b s t r a c t

Background: The health concerns that spur care-seeking in emergency departments (EDs) among homeless pop- ulations are not well described. The Veterans Affairs (VA) comprehensive healthcare system does not require health insurance and thus offers a unique window into ED service use by homeless veterans.

Objective: This study examined the top 10 diagnostic categories for ED use among homeless and non-homeless veterans classified by age, gender, and race/ethnicity.

Design: An observational study was conducted using national VA administrative data from 2016 to 2019.

Participants: Data on 260,783 homeless veterans and 2,295,704 non-homeless veterans were analyzed. Main measures: Homelessness was defined as a documented diagnostic code or use of any VA homeless program. Presenting diagnoses to the ED were grouped based on Clinical Classifications Software Refined (CCSR) catego- ries endorsed by the Agency for Healthcare Research and Quality (AHRQ).

Key results: The most common diagnostic categories for ED use among homeless veterans were, in order, muscu- loskeletal pain, Alcohol-related disorders, suicidal behaviors, low back pain, and non-specified conditions, which together accounted for 22-24% of all ED visits. Among non-homeless veterans, alcohol-related disorders, suicidal behaviors, and Depressive disorders did not number in the top 10 diagnostic categories for ED use. Some differ- ences between homeless and non-homeless veterans presenting for ED care, such as age, gender, and race/eth- nicity largely mirrored known epidemiological differences between these groups in general. But Respiratory infections and symptoms were only in the top 10 for black veterans and depressive disorder was only in the top 10 for Hispanic veterans. Conclusions: These data suggest that addressing psychosocial factors and optimiz- ing healthcare for behavioral health and pain conditions among veterans experiencing homelessness has the po- tential to reduce emergency care-seeking.

Published by Elsevier Inc.

1. Introduction

Emergency departments (EDs) serve as a major source of medical care in the U.S. and thus ED utilization can reflect health needs of a local community or population that might not otherwise be met, and thus merit examination [1]. Research has shown that homeless adults frequently seek ED care [2-6] and use EDs at rates 3-4 times greater than other adults [6-8]. Uninsurance, underinsurance, and lack of a reg- ular source of care contribute to emergency care-seeking [5,9], but they are not likely to be the main drivers. Even in the U.S. Department of

* Corresponding author at: 7411 John Smith Drive, Suite 1100, San Antonio, TX 78229, USA.

E-mail address: [email protected] (J. Tsai).

Veterans Affairs healthcare system, which provides eligible vet- erans free or low-cost comprehensive healthcare services, homeless veterans are four times more likely to use EDs than their non- homeless peers [10]. One study found that 45% of homeless VA clients qualified as Frequent ED users (i.e., had more than four ED visits in one year) compared to 1% of other veterans in the VA healthcare system [11]. A more recent analysis found that among VA patients who experi- enced any period of homelessness, 35% of them used EDs in the same year as their homeless experience. The percentage varied by age, from a low of 29% among homeless veterans in the group aged 18-29, to 38% in those aged 60-74 [12]. Compared to examinations of how much the ED is used by this population, examination of specific reasons for seeking ED care have been sparse. One older paper found heart fail- ure and schizophrenia were risk factors for ED use [13]. A more detailed,

https://doi.org/10.1016/j.ajem.2021.02.038 0735-6757/Published by Elsevier Inc.

and current, analysis of reasons for care-seeking in emergency settings could help shed light on unmet care needs and inform efforts to avert ED use when it is not necessary.

Across any population, medical and mental health needs vary by age, gender, and race/ethnicity [14-16], so it stands to reason that presenting diagnoses for ED use will vary in a similar manner. For example, males are more likely to present to EDs for substance use-related reasons than females [17]. There is some evidence that patient-reported stress and anxiety is higher among females reporting to EDs than males [18] and younger female patients may be more likely to present with vague, unspecific symptoms [19]. There have been fewer studies on the topic of race/ethnicity in ED use. Those that exist have often found race/ethnic- ity is not an independent correlate of ED use after adjusting for other fac- tors such as age, socioeconomic status, and health insurance coverage [20,21]. These studies, however, have not consistently probed reasons for presentation, which could disclose differences. On age, a number of studies have found that older age is associated with increasing ED use, length of stay, and resource intensity in EDs [22,23]. Beyond these demo- graphic profiles, some studies flag reasons for ED use that cross age, gen- der, and race/ethnicity. For example, several studies using multi-year data from the National Hospital Ambulatory Medical Care Survey have reported dramatic increases over recent time in the number of ED visits for attempted suicide or self-inflicted injury, across major demographic groups [24,25]. Whether similar patterns are evident among veterans, or to those experiencing homelessness, is not yet known.

In the current study, we used national VA administrative data on over 2 million veterans from 2016 to 2019 to examine the top 10 diag- nostic categories for ED use as documented in their medical charts. We compared homeless and non-homeless veterans on diagnostic catego- ries for ED use and further classified diagnostic categories for ED use among homeless veterans by age, gender, and race/ethnic group.

1. Materials and methods

National VA administrative data on 260,783 homeless veterans and 2,295,704 non-homeless veterans from 2016 to 2019 were analyzed. Among the sample of homeless veterans, there were 104,904 veterans in 2016; 103,248 veterans in 2017; 100,606 veterans in 2018; and 100,606 veterans in 2019. Homelessness was defined as having a docu- mented International Classification of Diseases, Tenth Revision, Clinical Modification diagnostic code Z59.0 Homelessness or use of any VA homeless program including the following: Healthcare for Homeless Veterans (HCHV), HCHV Contract Residential Services (CRS), HCHV Low Demand Safe Haven (LDSH), Grant and Per Diem (GPD), Do- miciliary Care for Homeless Veterans (DCHV), Supportive Services for Veteran Families (SSVF), US Department of Housing and Urban Development-VA Supportive Housing (HUD-VASH), Homeless Veteran Community Employment Services (HVCES), and Compensated Work Therapy (CWT). Use of VA ED services from 2016 to 2019 was defined in the VA administrative data as all visits to clinic code 130, reflecting the Emergency Department. Presenting diagnoses to the ED were grouped based on Clinical Classifications Software Refined (CCSR) for ICD-10-CM diagnoses [26], which is one in a family of databases and software tools developed as part of the Healthcare Cost and Utilization Project (HCUP), a Federal-State-Industry partnership sponsored by the Agency for Healthcare Research and Quality (AHRQ). The CCSR for ICD-10-CM diagnoses aggregates over 70,000 Diagnosis codes into a manageable number of Clinically meaningfully categories. The catego- ries are organized across 21 body systems, which generally follow the structure of the ICD-10 diagnosis chapters. Since we sought to rank ED visits by presenting diagnosis, we utilized CCSR v2021.1 which assigns a mutually exclusive default CCSR for the first-listed diagnosis for outpa- tient encounters. Data for each year were analyzed separately; for exam- ple, veterans with evidence of homelessness in 2016 were observed in 2016, but their ED utilization was not considered in subsequent years unless there was evidence of homelessness in those years as well.

Although most categories in the CCSR are self-explanatory, some are less so. For example, CCSR category FAC025 (Other specified status) in- cludes ICD-10-CM codes Z990 Dependence on aspirator and Z992 De- pendence on renal dialysis. CCSR category FAC019 (Socioeconomic/ psychosocial factors) includes ICD-10-CM codes Z550 (Illiteracy and low-level literacy) and Z562 (Threat of job loss). CCSR category FAC012 (Other specified encounters and counseling) includes ICD-10- CM codes Z713 (Dietary counseling and surveillance) and Z7181 (Spir- itual or religious counseling).

1. Results

As presented in Table 1, the top 10 CCSR diagnostic categories for homeless veterans presenting to EDs were consistent from 2016 to 2019. Those top categories were (in order): Musculoskeletal pain, Alcohol-related disorders, suicidal ideation/attempt/intentional self- harm, Low back pain, Other specified encounters and counseling, Non- specific chest pain, Abdominal pain and other digestive/abdomen signs, Other specified status, Depressive disorders, and Skin and subcu- taneous tissue infections. The top 5 categories accounted for 22-24% of all ED visits among homeless veterans each year. Among non-homeless veterans, the top 10 categories for ED use were in some ways similar to homeless veterans. However, Alcohol-related disorders, Suicidal idea- tion/attempt/intentional self-harm, Other specified encounters and counseling, and Depressive disorders did not appear in the top 10 for this group, and four other conditions did: Respiratory signs and symp- toms, Other specified upper respiratory infections, Acute bronchitis and Urinary tract infections.

Table 2 displays the top 10 categories for ED use among homeless veterans in 2019 by age group. The top category among the younger age groups (18-29 and 30-44) was Suicidal ideation/attempt/inten- tional self-harm whereas Musculoskeletal pain was the top category among older age groups (all age groups 45 years and older). Alcohol- related disorders was the second most common category across age groups, except for the youngest (18-29) and oldest groups (75+). De- pressive disorders appeared in the top 5 only for the youngest age group, while Socioeconomic/psychosocial factors was in the top 5 only for the oldest age group. Schizophrenia spectrum and other psychotic disorders, and Headache including migraine were only in the top 10 among those aged 18-29. Chronic obstructive pulmonary disease and bronchiectasis was only in the top 10 among the two oldest age groups (60-74, 75+); and Urinary tract infection was also only in the top 10 among the oldest age group.

Among homeless veterans in 2019, the top categories for ED use were similar for both genders with a few differences (Table 3). Among male homeless veterans, the top five categories, in order, were, Muscu- loskeletal pain, Alcohol-related disorders, Suicidal ideation/attempt/ in- tentional self-harm, Other specified status, and Nonspecific chest pain. Among female homeless veterans, the top 5 were Musculoskeletal pain, Abdominal pain and other digestive/abdomen signs, Low back pain, Suicidal ideation/attempt/intentional self-harm, and Other speci- fied status. Notably, Alcohol-related disorders and Nonspecific chest pain did not make the top 5, among women.

There were some notable similarities and differences when ED use in 2019 was broken down by race/ethnic categories, as shown in Table 4. Suicidal ideation/attempt/intentional self-harm, Musculoskeletal pain, and Skin and subcutaneous tissue infections were in the top 4 categories across race/ethnic groups. Alcohol-related disorders were also in the top 5 among across race/ethnic groups with the exception of non-Hispanic black veterans. Socioeconomic/psychosocial factors were in the top 10 across race/ethnic groups except for non-Hispanic black veterans. Among categories that were more specific to certain race/ethnic groups, Skin and subcutaneous tissue infections were only in the top 5 for non- Hispanic white veterans; Other specified upper respiratory infections and Respiratory signs and symptoms were only in the top 10 for

Table 1

Top 10 Clinical Classifications Software Refined (CCSR) for ICD-10 categories for VA emergency department visits among homeless and non-homeless veterans, 2016-2019.

	Homeless veterans
	2016		2017		2018			2019
	CCSR category	#visits (%)	CCSR category	#visits (%)	CCSR category	#visits (%)		CCSR category	#visits (%)
	MUS010 Musculoskeletal pain	17,774 (5.6%)	MUS010	18,497 (5.6%)	MUS010	18,723 (5.8%)		MUS010	19,070 (5.9%)
	MBD017 Alcohol-related disorders	17,594 (5.5%)	MBD017	17,335 (5.2%)	MBD017	16,543 (5.1%)		MBD017	17,312 (5.4%)
	MBD012 Suicidal ideation/attempt/intentional self-harm	12,177 (3.8%)	FAC025	16,360 (5.0%)	FAC025	15,333 (4.7%)		MBD012	15,890 (4.9%)
	MUS038 Low back pain	12,032 (3.8%)	MBD012	13,992 (4.2%)	MBD012	15,009 (4.6%)		FAC025	11,980 (3.7%)
	FAC012 Other specified encounters and counseling	10,664 (3.3%)	FAC012	11,674 (3.5%)	FAC012	10,996 (3.4%)		FAC012	10,517 (3.3%)
	CIR012 Nonspecific chest pain	10,384 (3.3%)	MUS038	11,088 (3.4%)	MUS038	10,333 (3.2%)		CIR012	10,487 (3.3%)
	SYM006 Abdominal pain and other digestive/abdomen signs	9435 (3.0%)	CIR012	10,180 (3.1%)	CIR012	9998 (3.1%)		MUS038	10,033 (3.1%)
	FAC025 Other specified status	9225 (2.9%)	SYM006	9246 (2.8%)	SYM006	9363 (2.9%)		SYM006	9650 (3.0%)
	MBD002 Depressive disorders	9085 (2.8%)	MBD002	8877 (2.7%)	SKN001	8478 (2.6%)		SKN001	8299 (2.6%)
	SKN001 Skin and subcutaneous tissue infections	8307 (2.6%)	SKN001	8374 (2.5%)	FAC019	8047 (2.5%)		FAC019	8091 (2.5%)
	TOTAL VISITS	319,008		330,288		325,050			321,022
	Non-Homeless Veterans
	2016		2017		2018			2019
	CCSR category	# visits (%)	ICD code	# visits (%)	ICD code	# visits (%)		ICD code	# visits (%)
	MUS010 Musculoskeletal pain	116,318 (6.0%)	MUS010	127,471 (6.4%)	MUS010	136,818 (6.6%)		MUS010	140,508 (6.9%)
	MUS038 Low back pain	75,974 (4.0%)	MUS038	76,181 (3.8%)	SYM006	78,852 (3.8%)		SYM006	80,324 (3.9%)
	CIR012 Nonspecific chest pain	72,184 (3.8%)	SYM006	74,392 (3.7%)	MUS038	76,306 (3.7%)		CIR012	76,558 (3.7%)
	SYM006 Abdominal pain and other digestive/abdomen signs	71,752 (3.7%)	CIR012	72,704 (3.6%)	CIR012	75,422 (3.7%)		MUS038	76,557 (3.7%)
	SKN001 Skin and subcutaneous tissue infections	62,765 (3.3%)	SKN001	63,272 (3.2%)	SYM013	64,984 (3.2%)		SYM013	66,091 (3.2%)
	SYM013 Respiratory signs and symptoms	55,800 (2.9%)	SYM013	61,635 (3.1%)	SKN001	64,593 (3.1%)		SKN001	63,371 (3.1%)
	FAC012 Other specified encounters and counseling	50,524 (2.6%)	FAC012	57,118 (2.8%)	FAC012	54,430 (2.6%)		RSP006	54,391 (2.7%)
	RSP006 Other specified upper respiratory infections	47,553 (2.5%)	RSP006	53,087 (2.6%)	RSP006	53,587 (2.6%)		FAC012	52,755 (2.6%)
	RSP005 Acute bronchitis	44,541 (2.3%)	RSP005	45,713 (2.3%)	FAC025	43,507 (2.1%)		GEN004	42,443 (2.1%)
	GEN004 Urinary tract infections	39,096 (2.0%)	FAC025	44,623 (2.2%)	RSP005	41,719 (2.0%)		MUS011	40,321 (2.0%)
	TOTAL VISITS	1,918,565		2,005,622		2,062,055			2,049,053

Note: FAC019 = Socioeconomic/psychosocial factors.

non-Hispanic black veterans; and Depressive disorder was only in the top 10 for Hispanic veterans.

1. Discussion

In a comprehensive healthcare system that does not rely on insur- ance payments, the top presenting diagnostic categories in EDs among

homeless veterans were related to physical pain, suicidal behaviors, de- pressive disorders, alcohol misuse, and non-specified factors, which we presume did not reflect easily specified medical diagnoses. Pain diagno- ses featured prominently among the ED visits of non-homeless vet- erans, as well. However non-homeless veterans did not present so frequently with suicidal behaviors, depressive disorders, and alcohol misuse. The high frequency of ED visits related to suicidal behaviors

Table 2

Top 10 Clinical Classifications Software Refined (CCSR) for ICD-10 categories for VA emergency department visits among homeless veterans by age in 2019.

	18-29 years old			30-44 years			45-49 years			60-74 years			75+ years
	CCSR category	#visits		CCSR category	#visits		CCSR category	#visits		CCSR category	#visits		CCSR category	#visits
	MBD012 Suicidal ideation/attempt/intentional self-harm	991		MBD012	3814		MUS010	6980		MUS010	7628		MUS010	699
		(7.7%)			(6.5%)			(6.5%)			(5.9%)			(5.5%)
	MUS010 Musculoskeletal pain	658		MBD017	3811		MBD017	6949		MBD017	5979		FAC025	557
		(5.1%)			(6.5%)			(6.4%)			(4.6%)			(4.4%)
	FAC025 Other specified status	526		MUS010	3104		MBD012	6009		FAC025	5032		FAC019	506
		(4.1%)			(5.3%)			(5.6%)			(3.9%)			(4.0%)
	MBD002 Depressive disorders	490		FAC025	2230		MUS038	3698		CIR012	4949		SYM013	471
		(3.8%)			(3.8%)			(3.4%)			(3.8%)			(3.7%)
	SYM006 Abdominal pain and other digestive/abdomen signs	458		MUS038	1992		CIR012	3689		MBD012	4923		CIR012	412
		(3.5%)			(3.4%)			(3.4%)			(3.8%)			(3.2%)
	MBD017 Alcohol-related disorders	416		SYM006	1844		FAC025	3635		FAC012	4601		FAC012	384
		(3.2%)			(3.2%)			(3.4%)			(3.6%)			(3.0%)
	RSP006 Other specified upper respiratory infections	392		MBD002	1838		FAC012	3469		SYM006	3916		SYM016	372
		(3.0%)			(3.2%)			(3.2%)			(3.0%)			(2.9%)
	MUS038 Low back pain	391		FAC012	1725		SYM006	3077		FAC019	3781		RSP008	359
		(3.0%)			(3.0%)			(2.9%)			(2.9%)			(2.8%)
	MBD001 Schizophrenia spectrum and other psychotic disorders	377		SKN001	1612		SKN001	2897		RSP008	3690		SYM006	355
		(2.9%)			(2.8%)			(2.7%)			(2.9%)			(2.8%)
	NVS010 Headache, including migraine	347		RSP006	1455		MBD002	2596		MUS038	3678		GEN004	320
		(2.7%)			(2.5%)			(2.4%)			(2.8%)			(2.5%)
	TOTAL VISITS	12,902			58,335			107,927			129,174			12,684

Note: CIR012 = Nonspecific chest pain; FAC012 = Other specified encounters and counseling; FAC019 = Socioeconomic/psychosocial factors; GEN004 = Urinary tract infections; MBD002 = Depressive disorders; RSP008 = Chronic obstructive pulmonary disease and bronchiectasis; SKN001 = Skin and subcutaneous tissue infections; SYM013 = Respiratory signs and symptoms; SYM016 = Other general signs and symptoms.

Table 3

Top 10 Clinical Classifications Software Refined (CCSR) for ICD-10 categories for VA emergency department visits among homeless veterans by gender in 2019.

	Male			Female
	CCSR category	# visits		CCSR category	#visits
	MUS010 Musculoskeletal pain	16,828		MUS010 Musculoskeletal pain	2242
		(5.9%)			(6.5%)
	MBD017 Alcohol-related disorders	16,501		SYM006 Abdominal pain and other digestive/abdomen signs	1610
		(5.8%)			(4.6%)
	MBD012 Suicidal ideation/attempt/ intentional self-harm	14,618		MUS038 Low back pain	1274
		(5.1%)			(3.7%)
	FAC025 Other specified status	10,783		MBD012 Suicidal ideation/attempt/ intentional self-harm	1272
		(3.8%)			(3.7%)
	CIR012 Nonspecific chest pain	9491		FAC025 Other specified status	1197
		(3.3%)			(3.4%)
	FAC012 Other specified encounters and counseling	9344		NVS010 Headache, including migraine	1178
		(3.3%)			(3.4%)
	MUS038 Low back pain	8759		FAC012 Other specified encounters and counseling	1173
		(3.1%)			(3.4%)
	SYM006 Abdominal pain and other digestive/abdomen signs	8040		RSP006 Other specified upper respiratory infections	1009
		(2.8%)			(2.9%)
	SKN001 Skin and subcutaneous tissue infections	7617		CIR012 Nonspecific chest pain	996
		(2.7%)			(2.9%)
	FAC019 Socioeconomic/psychosocial factors	7472		GEN004 Urinary tract infections	855
		(2.6%)			(2.5%)
	TOTAL VISITS	286,318			34,704

among homeless veterans is troubling particularly because suicide rep- resents VA’s top clinical priority [27] and previous studies have found homeless veterans are already at elevated risk for suicide [28,29].

Our findings are consistent with some previous studies that have found that frequent ED users often have significant psychiatric and so- cial comorbidities [30]. The presenting diagnoses for ED use we found were also similar to previous studies in settings where there is no com- prehensive healthcare coverage. For example, one study of over 1000 homeless adults in New York City shelters found that traumatic injury, mental health, and Substance use disorders were among the Most frequently cited reasons for ED visits [31]. Another study, based on the National Electronic Injury surveillance System (NEISS), found that substance use was more common among Homeless patients who presented to NEISS EDs than non-homeless control patients. The most common injuries among homeless patients in that study

occurred in the lower extremities; sprains/strains, contusions/abra- sions, and burns [32].

Homelessness itself likely contributes directly to some of the medical needs captured in these data, while offering an incomplete explanation. For example, it is plausible that unstable housing and exposure to weather and other harsh living environments increase risk for injury, skin and subcutaneous tissue infections, poor physical health, and even behavioral health problems, such as feelings of suicidality. At the same time, a good number of these conditions may have predated the home- less experience, raising the traditional and not fully resolvable matter of how much homelessness “causes poor health” versus health adversity contributing to homelessness [33]. Regardless of how causality is under- stood, emergency department presentations often hint at failures of care or social intervention that could have, at least in some instances, averted the need for emergency care. Successful outpatient care for addiction or

Table 4

Top Clinical Classifications Software Refined (CCSR) for ICD-10 categories for VA emergency department visits among homeless veterans by racial/ethnic group in 2019.

	Non-Hispanic white			Non-Hispanic black			Hispanic			Multi/Other
	CCSR category	#visits		CCSR category	#visits		CCSR category	#visits		CCSR category	#visits
	MBD017 Alcohol-related disorders	11,355		MUS010	8971		MUS010	1398		MBD017	793
		(7.9%)			(6.7%)			(6.2%)			(7.5%)
	MBD012 Suicidal ideation/attempt/intentional self-harm	7714		MBD012	6188		MBD017	1185		MUS010	622
		(5.3%)			(4.6%)			(5.2%)			(5.9%)
	MUS010 Musculoskeletal pain	7572		FAC025	5441		FAC025	1137		MBD012	497
		(5.2%)			(4.1%)			(5.0%)			(4.7%)
	SKN001 Skin and subcutaneous tissue infections	4580		CIR012	4896		MBD012	1038		FAC025	433
		(3.2%)			(3.7%)			(4.6%)			(4.1%)
	FAC025 Other specified status	4554		FAC012	4812		MUS038	815		MUS038	339
		(3.2%)			(3.6%)			(3.6%)			(3.2%)
	FAC012 Other specified encounters and counseling	4530		MUS038	4732		SYM006	782		SYM006	308
		(3.1%)			(3.5%)			(3.4%)			(2.9%)
	SYM006 Abdominal pain and other digestive/abdomen signs	4415		SYM006	3892		CIR012	685		CIR012	303
		(3.1%)			(2.9%)			(3.0%)			(2.9%)
	CIR012 Nonspecific chest pain	4310		MBD017	3431		FAC019	651		FAC019	290
		(3.0%)			(2.6%)			(2.9%)			(2.8%)
	MUS038 Low back pain	3862		SYM013	3193		MBD002	612		FAC012	279
		2.7%)			(2.4%)			(2.7%)			(2.6%)
	FAC019 Socioeconomic/psychosocial factors	3849		RSP006	2999		SKN001	593		SKN001	264
		(2.7%)			(2.2%)			(2.6%)			(2.5%)
	TOTAL VISITSa	144,530			133,455			22,707			10,535

Note: MBD002 = Depressive disorders; RSP006 Other specified upper respiratory infections; SYM013 Respiratory signs and symptoms.

^a The column totals do not sum to 321,022 due to case missing race/ethnicity data.

pain are appropriate goals for outpatient care, even with highly vulnera- ble populations. On whole the present findings underscore the impor- tance of stable housing and need to attend to pain-related conditions as well as mental and substance abuse treatment needs.

The age, gender, and race/ethnicity differences we observed for ED use often mirrored epidemiological differences in the general population. For example, alcohol-related disorders were in the top 5 categories for ED use among homeless male veterans, but not female veterans. This mirrors differences in the prevalence of substance use disorders in the general population [34,35]. Conversely, abdominal pain was in the top 5 among female homeless veterans but not homeless male veterans, mirroring population-based prevalence studies among women in general [36-38].

There were some categories for ED use that were more specific to certain race/ethnic groups and which our data cannot explain, e.g., skin infections for white veterans and respiratory infections for black veterans. Surprisingly, alcohol-related disorders, and socioeco- nomic and psychosocial factors did not number in the top categories for ED use among homeless black veterans. There is some data to sug- gest that the prevalence of Alcohol use disorder is lower among blacks than whites in the general population [39], but it is not clear how such data would translate directly into a population defined by need for safety net assistance, such as veterans who are homeless. Additionally, annual needs assessments of homeless veterans have found that white veterans report greater unmet needs in housing, healthcare, and basic needs than black veterans [40]. It is possible that socioeconomic factors may not drive ED use as much for black homeless veterans, or simply that there are other medical and mental health factors that are larger drivers of their ED use. Further research is needed to understand the un- derlying causes of these racial differences.

Age-related differences worth noting. While suicidal behavior was the top category for ED use among younger homeless veterans, pain conditions was in the top category among older homeless veterans. Cer- tainly, pain conditions become more common with age [41], which may explain their prominence for older homeless veterans. Age-related sui- cide risk, by contrast, has fluctuated. A 2020 VA report found that vet- erans aged 75 years and older were at particularly high risk for suicide, until recent years (2016-2018), when their risk fell below that of younger veterans [27]. Because suicide rates are higher in later life in the general population [42], we suggest the trend in age-related risk should be monitored closely as it may change once more.

Our tabulation of diagnostic categories coded as part of ED care should not overshadow a complementary line of study, which asks why the ED serves as the site of care for some people, under some con- ditions, but not all. Some factors driving ED use are not readily discerned in large database analyses. A qualitative study of 100 Medicaid-enrolled frequent ED users identified three potential drivers: negative personal experiences with the healthcare system, challenges associated with so- cioeconomic status, and significant chronic mental and physical disease burden [43]. Another study found that many homeless adults present to EDs with basic needs for food, shelter, and safety. Such ED use might abate if those needs were met [44]. These findings about psychosocial factors are consistent with studies of homeless populations in Canada, who have universal health coverage, but who report barriers to obtaining health care and have higher rates of ED use than the general population [45,46]. It may that even for patients with healthcare cover- age, ED use is affected more by their fears regarding accessibility of care than the acuity of their condition [47]. The prominence of “Other” cate- gories (i.e., “other specified encounters and counseling”, “other speci- fied status”) hint that the concerns driving ED presentation often fit poorly within a medical or psychiatric diagnostic framework and may reflect more complex or subtle psychosocial concerns, including a need for human contact.

A number of interventions have been proposed to address frequent ED use in homeless populations. Several older studies in the U.S. found that some unique Case management-like models have been promising. For example, one randomized controlled trial found that “compassionate

care” led to reduced ED use compared to treatment-as-usual, presum- ably because of increased satisfaction with treatment [48]. Another ran- domized controlled trial found that patients who were enrolled in a transitional housing and case management program had greater reduc- tions in ED visits and hospitalization days than treatment-as-usual [49]. There have also been pilot programs targeting specific subgroups of homeless frequent ED users, such as those with alcohol use disorders [50]. However, these interventions have been resource-intensive and have not been widely implemented. It should also be stipulated, some unknowable percentage of these ED visits represented situations of un- avoidable medical necessity. More recently, several international rigor- ous trials have been conducted of the popular Housing First model, which offers subsidized housing and case management with no prereq- uisites for treatment or sobriety. While a randomized controlled trial in Vancouver found that “scattered site” Housing First programs showed reduced ED visits compared to usual care [51], another trial of four cities in France found Housing First programs did not reduce ED visits more than usual care [52]. A synthesis of the literature on the Housing First model has characterized the evidence its ability to reduce ED use as moderate [53]. Together, these findings indicate there have been vari- ous programs developed to address frequent ED use in this population, but there are many unanswered questions and opportunities for targeted intervention, especially in the top reasons for ED use such as those we found.

Several study limitations deserve mention. First, large database anal- yses cannot show if emergency care would have been considered justi- fiable based on acuity of illness or the “reasonable person” standard historically used by insurers. Second, the CSSR for ICD-10-CM categories aggregate diagnostic categories and are perhaps more revelatory than ICD code listings but are not as precise. Third, the study population- vet- erans who have experienced homelessness- is heavily weighted toward native-born U.S. men with high school education or equivalent and can- not be considered wholly representative of all persons experiencing homelessness. These limitations are counterbalanced by the strengths of the study including the large national sample across multiple years, use of administrative records versus self-report data, and examination of the issue broken down by age, gender, and race/ethnicity. Together, the study findings contribute to the research literature and have clinical implications for provider about ED use in this population.

Despite these limitations, these data on diagnostic reasons for ED strongly hint that some portion of ED visits could be mitigated through buttressing of primary care and social services. We would argue EDs are not ideal service delivery sits for many conditions. For now, there re- mains a lack of scalable cost-Effective interventions to address their use across different healthcare systems, including the VA, but options such as compassionate care, tailored outpatient settings and robust social interventions continue to hold promise.

Funding

This work was supported by internal funds from the VA National Center on Homelessness among Veterans.

Declaration of Competing Interest

None of the authors report any conflicts of interest. Dr. Kertesz holds stock in CVS Caremark, Thermo Fisher, and Zimmer Biomet, not exceed- ing 5% of his assets. Dr. Kertesz reports his spouse holds equity in Merck, Abbot, Thermo Fisher, and Johnson and Johnson, in her private assets, not exceeding 10% of her assets. Dr. Kertesz also receives income from UpToDate, Inc.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influ- ence the work reported in this paper.

References

1. Weiss AJ, Wiler LM, Stocks C, Blanchard J. Overview of emergency department visits in the United States, 2011. Statistical brief #174, healthcare cost and utilization pro- ject (HCUP) statistical briefs. Rockville, MD: Agency for Healthcare Research and Quality; 2014.
2. Kushel MB, Perry S, Bangsberg D, Clark R, Moss AR. Emergency department use among the homeless and marginally housed: results from a community-based study. Am J Public Health. 2002;92(5):778-84.
3. Kushel MB, Vittinghoff E, Haas JS. Factors associated with the health care utilization of homeless persons. JAMA. 2001;285(2):200-6.
4. Oates G, Tadros A, Davis SM. A comparison of national emergency department use by homeless versus non-homeless people in the United States. J Health Care Poor Underserved. 2009;20(3):840-5.
5. Ku BS, Scott KC, Kertesz SG, Pitts SR. Factors associated with use of urban emergency departments by the U.S. homeless population. Public Health Rep. 2010;125(3): 398-405.
6. Mandelberg JH, Kuhn RE, Kohn MA. Epidemiologic analysis of an urban, public emer- gency department’s frequent users. Acad Emerg Med. 2000;7(6):637-46.
7. D’Amore J, Hung O, Chiang W, Goldfrank L. The epidemiology of the homeless pop- ulation and its impact on an urban emergency department. Acad Emerg Med. 2001; 8(11):1051-5.
8. Doran KM, Raven MC, Rosenheck RA. What drives frequent emergency department use in an integrated health system? National data from the Veterans Health Admin- istration. Ann Emerg Med. 2013;62(2):151-9.
9. Chwastiak LA, Tsai J, Rosenheck RA. Health insurance, not serious Mental illness, de- termines whether chronically homeless individuals engage in primary care. Am J Public Health. 2012;102(12):e83-9.
10. Tsai J, Doran KM, Rosenheck RA. When health insurance is not a factor: national comparison of homeless and nonhomeless US veterans who use veterans affairs emergency departments. Am J Public Health. 2013;103(Suppl2):S225-31.
11. Tsai J, Rosenheck RA. Risk factors for ED use among homeless veterans. Am J Emerg Med. 2013;31(5):855-8.
12. Tsai J, Pietrzak RH, Szymkowiak D. The problem of veteran homelessness: An update for the new decade. Am J Prev Med. 2021. https://doi.org/10.1016/j.amepre.2020.12. 012.
13. Doran KM, Raven MC, Rosenheck RA. What drives frequent emergency department use in an integrated health system? National data from the veterans health admin- istration. Ann Emerg Med. 2013;62(2):151-9.
14. Hatch SL, Dohrenwend BP. Distribution of traumatic and other stressful life events by race/ethnicity, gender, SES and age: a review of the research. Am J Community Psychol. 2007;40(3-4):313-32.
15. Stewart WF, Lipton RB, Celentano DD, Reed ML. Prevalence of migraine headache in the United States: relation to age, income, race, and other sociodemographic factors. JAMA. 1992;267(1):64-9.
16. Fillingim RB, King CD, Ribeiro-Dasilva MC, Rahim-Williams B, Riley III JL. Sex, gender, and pain: a review of recent clinical and experimental findings. J Pain. 2009;10(5): 447-85.
17. Cannon RD, Beauchamp GA, Roth P, Stephens J, Burmeister DB, Richardson DM, et al. sex differences in prevalence of emergency department patient substance use. Clin Ther. 2018;40(2):197-203.
18. Patel R, Biros MH, Moore J, Miner JR. gender differences in patient-described pain, stress, and anxiety among patients undergoing treatment for painful conditions in the emergency department. Acad Emerg Med. 2014;21(12):1478-84.
19. Anson O, Carmel S, Levin M. Gender differences in the utilization of emergency de- partment services. Women Health. 1991;17(2):91-104.
20. Baker DW, Stevens CD, Brook RH. Determinants of emergency department use: are race and ethnicity important? Ann Emerg Med. 1996;28(6):677-82.
21. Hong R, Baumann BM, Boudreaux ED. The emergency department for routine healthcare: race/ethnicity, socioeconomic status, and perceptual factors. J Emerg Med. 2007;32(2):149-58.
22. Latham LP, Ackroyd-Stolarz S. Emergency department utilization by older adults: a descriptive study. Can Geriat J. 2014;17(4):118-25.
23. Shankar KN, Liu SW, Ganz DA. Trends and characteristics of emergency department visits for fall-related injuries in older adults, 2003-2010. West J Emerg Med. 2017;18 (5):785-93.
24. Ting SA, Sullivan AF, Boudreaux ED, Miller I, Camargo Jr CA. Trends in US emergency department visits for attempted suicide and self-inflicted injury, 1993-2008. Gen Hosp Psychiatry. 2012;34(5):557-65.
25. Larkin GL, Smith RP, Beautrais AL. Trends in US emergency department visits for sui-

cide attempts, 1992-2001. Crisis. 2008;29(2):73-80.

1. Agency for Healthcare Research and Quality. Clinical Classifications Software Refined (CCSR) for ICD-10-CM Diagnoses. https://www.hcup-us.ahrq.gov/toolssoftware/ ccsr/dxccsr.jsp;; 2020. [accessed December 1 2020].
2. U.S. Department of Veterans Affairs. 2020 National Veteran suicide prevention Re- port. Washington, D. C.: Office of Mental Health and Suicide Prevention, U.S. Depart- ment of Veterans Affairs; 2020
3. Tsai J, Cao X. Association between suicide attempts and homelessness in a population-based sample of US veterans and non-veterans. J Epidemiol Commun Health. 2019;73(4):346-52. https://doi.org/10.1136/jech-2018-211065.
4. Hoffberg AS, Spitzer E, Mackelprang JL, Farro SA, Brenner LA. Suicidal self-directed violence among homeless US veterans: a systematic review. Suicide Life Threat Behav. 2018;48(4):481-98.
5. LaCalle EJ, Rabin EJ, Genes NG. High-frequency users of emergency department care. J Emerg Med. 2013;44(6):1167-73.
6. Padgett DK, Struening EL. Influence of substance abuse and Mental disorders on emergency room use by homeless adults. Psychiatr Serv. 1991;42(8):834-8.
7. Mackelprang JL, Graves JM, Rivara FP. Homeless in America: injuries treated in US emergency departments, 2007-2011. Int J Inj Control Saf Promot. 2014;21(3): 289-97.
8. Hwang SW. Is homelessness hazardous to your health? Can J Public Health. 2002;93

(6):407-10.

1. McHugh RK, Votaw VR, Sugarman DE, Greenfield SF. Sex and gender differences in substance use disorders. Clin Psychol Rev. 2018;66:12-23.
2. Lev-Ran S, Le Strat Y, Imtiaz S, Rehm J, Le Foll B. Gender differences in prevalence of substance use disorders among individuals with lifetime exposure to substances: re- sults from a large Representative sample. Am J Addict. 2013;22(1):7-13.
3. Beard RW, Reginald PW, Wadsworth J. Clinical features of women with chronic lower abdominal pain and pelvic congestion. BJOG. 1988;95(2):153-61.
4. Stones RW, Price C. Health services for women with chronic pelvic pain. J R Soc Med.

2002;95(11):531-5.

1. Ahangari A. Prevalence of chronic pelvic pain among women: an updated review. Pain Physician. 2014;17(2):E141-7.
2. Grant BF, Goldstein RB, Saha TD, Chou SP, Jung J, Zhang H, et al. Epidemiology of DSM-5 alcohol use disorder: results from the National Epidemiologic Survey on al- cohol and related conditions III. JAMA Psychiat. 2015;72(8):757-66.
3. Tsai J, Mitchell L, Nakashima J, Blue-Howells J. Unmet needs of homeless U.S. vet- erans by gender and race/ethnicity: data from five annual surveys. Psychol Serv [in press].
4. Jones MR, Ehrhardt KP, Ripoll JG, Sharma B, Padnos IW, Kaye RJ, et al. Pain in the el- derly. Curr Pain Headache Rep. 2016;20(4):23.
5. Conwell Y, Duberstein PR, Caine ED. Risk factors for suicide in later life. Biol Psychi- atry. 2002;52(3):193-204.
6. Capp R, Kelley L, Ellis P, Carmona J, Lofton A, Cobbs-Lomax D, et al. Reasons for fre- quent emergency department use by Medicaid enrollees: a qualitative study. Acad Emerg Med. 2016;23(4):476-81.
7. Rodriguez RM, Fortman J, Chee C, Ng V, Poon D. Food, shelter and safety needs mo- tivating homeless persons’ visits to an urban emergency department. Ann Emerg Med. 2009;53(5):598-602.
8. Hwang SW, Ueng JJM, Chiu S, Kiss A, Tolomiczenko G, Cowan L, et al. Universal health insurance and health care access for homeless persons. Am J Public Health. 2010;100(8):1454-61.
9. Hwang SW, Chambers C, Chiu S, Katic M, Kiss A, Redelmeier DA, et al. A comprehen- sive assessment of health care utilization among homeless adults under a system of universal health insurance. Am J Public Health. 2013;103(S2):S294-301.
10. Capp R, Rooks SP, Wiler JL, Zane RD, Ginde AA. National study of health insurance type and reasons for emergency department use. J Gen Intern Med. 2014;29(4): 621-7.
11. Redelmeier DA, Molin JP, Tibshirani RJ. A randomised trial of compassionate care for the homeless in an emergency department. Lancet. 1995;345(8958):1131-4.
12. Sadowski LS, Kee RA, VanderWeele TJ, Buchanan D. Effect of a housing and case management program on emergency department visits and hospitalizations among chronically ill homeless adults: a randomized trial. JAMA. 2009;301(17): 1771-8.
13. McCormack RP, Hoffman LF, Wall SP, Goldfrank LR. Resource-limited, collaborative pilot intervention for chronically homeless, alcohol-dependent frequent emergency department users. Am J Public Health. 2013;103(S2):S221-4.
14. Russolillo A, Patterson M, McCandless L, Moniruzzaman A, Somers J. Emergency de- partment utilisation among formerly homeless adults with mental disorders after one year of housing first interventions: a randomised controlled trial. Int J Hous Pol- icy. 2014;14(1):79-97.
15. Tinland A, Loubiere S, Boucekine M, Boyer L, Fond G, Girard V, et al. Effectiveness of a housing support team intervention with a recovery-oriented approach on hospital and emergency department use by homeless people with severe mental illness: a randomised controlled trial. Epidemiol Psychiatr Sci. 2020;29:E169. https://doi. org/10.1017/S2045796020000785.
16. Tsai J. Is the housing first model effective? Different evidence for different outcomes. Am J Public Health. 2020;110(9):1376-7.