Characteristics of Emergency Department Visits by Older Versus Younger Homeless Adults in the United States

Abstract

Objectives. We compared the characteristics of emergency department (ED) visits of older versus younger homeless adults.

Methods. We analyzed 2005–2009 data from the National Hospital Ambulatory Medical Care Survey, a nationally representative survey of visits to hospitals and EDs, and used sampling weights, strata, and clustering variables to obtain nationally representative estimates.

Results. The ED visits of homeless adults aged 50 years and older accounted for 36% of annual visits by homeless patients. Although demographic characteristics of ED visits were similar in older and younger homeless adults, clinical and health services characteristics differed. Older homeless adults had fewer discharge diagnoses related to psychiatric conditions (10% vs 20%; P = .002) and drug abuse (7% vs 15%; P = .003) but more diagnoses related to alcohol abuse (31% vs 23%; P = .03) and were more likely to arrive by ambulance (48% vs 36%; P = .02) and to be admitted to the hospital (20% vs 11%; P = .003).

Conclusions. Older homeless adults’ patterns of ED care differ from those of younger homeless adults. Health care systems need to account for these differences to meet the needs of the aging homeless population.

The average age of the US homeless population is increasing. Whereas 11% of the homeless population was aged 50 years or older in 1990, this percentage had increased to 32% by 2003¹ and has continued to rise since then.² The median age of single homeless adults has increased from 37 years in 1990¹ to between 49 and 50 years today.^2,3 This trend is thought to be because of the aging of individuals born in the second half of the baby boom generation (those born 1954–1965), who have a higher risk of homelessness than do other age cohorts.³ Most homeless adults aged 50 years and older are aged between 50 and 64 years, with adults aged 65 years and older making up less than 5% of the total homeless population.^1,3

In the general population, adults aged 50 to 64 years are considered middle aged and have lower rates of chronic conditions than do those considered elderly, adults aged 65 years and older.^4,5 However, homeless adults aged 50 years and older have rates of chronic illnesses and geriatric conditions similar to or higher than those of housed adults 15 to 20 years older, including conditions often thought to be limited to the elderly, such as falls and memory loss.^6,7 Because middle-aged homeless adults face the same geriatric problems as do elderly housed adults, experts consider them to be elderly when aged 50 years, despite their relatively younger age.^6,8 Similar patterns of premature aging have been found in other vulnerable populations, including prisoners⁹ and patients with developmental disabilities.¹⁰

Despite the growth of the older homeless population, relatively little is known about use of health services among older homeless adults. Homeless health services and research have focused on problems that are common among younger homeless adults, including infectious disease,¹¹ substance use,¹² and mental illness.¹³ The few studies that have focused on older homeless adults found that they have medical problems that differ from those of younger homeless adults, including higher rates of chronic illnesses^6,14 and geriatric syndromes⁶ and lower rates of substance use.¹⁵ New frameworks for providing care to the vulnerable and growing older population are needed but cannot be developed until more is known about their use of health services.

Homeless adults aged 50 years and older use the emergency department (ED) frequently and at rates nearly 4 times those of the general population.^16–18 Knowledge about ED care that older homeless adults receive may allow researchers and clinicians to design interventions to reduce use of the ED and improve ED care for this vulnerable older population. Therefore, we have identified the demographic, clinical, and health services characteristics of ED visits in older versus younger homeless adults, using a nationally representative survey of US ED visits.

METHODS

We analyzed data from the ED component of the National Hospital Ambulatory Medical Care Survey (NHAMCS). The National Center for Health Statistics of the Centers for Disease Control and Prevention designed this nationally representative sample of ED visits. NHAMCS employs a 4-stage probability sample and includes EDs based in noninstitutional general short-stay hospitals and excludes EDs in federal, military, and Veterans Administration Hospitals. Hospital staff that US census field representatives have trained collect data from a systematic random sample of patient visits during a randomly assigned 4-week period. To make the results of the survey nationally representative, each patient visit is weighted on the basis of survey sampling probabilities. Details of the survey methodology are available from the National Center for Health Statistics.¹⁹

Study Sample of Homeless Adults

Our study sample included all ED visits of homeless patients aged 18 years and older in the 2005–2009 NHAMCS survey years. We defined homelessness using the NHAMCS patient residence variable. In 2005, NHAMCS added a residence category for homeless in addition to private residence, nursing home, other institution, and other residence. ED staff recorded the residence as homeless if “the patient has no home (e.g., lives on the street) or patient’s current place of residence is a homeless shelter.”²⁰ The sensitivity and specificity of the NHAMCS residence variable for detecting homelessness have not been described. However, other studies that used similar residence variables to identify homelessness validated reported homelessness by performing chart reviews and matching residential addresses to local homeless shelters.^21,22

Although these studies did not report sensitivity and specificity, chart reviews confirmed homelessness in patients reported as such; yet matching addresses to shelters identified additional homeless persons not captured by the residence variable. These findings suggest that residence variables may have high specificity but lower sensitivity for identifying homelessness.

Measures of Visit Characteristics

We compared the demographic characteristics, clinical characteristics, and health services use of ED visits of older and younger homeless adults. Demographic characteristics included gender, race/ethnicity, ED region, and insurance status. ED staff entered race/ethnicity according to their hospital’s usual practice or on the basis of the medical record or the staff member’s knowledge of the patient. We collapsed insurance status to the following categories: no insurance (self-pay or no charge), Medicaid, Medicare (including patients with both Medicaid and Medicare), and other insurance (private, workers’ compensation, or other).

Clinical characteristics included triage level and services and medications received in the ED, complaint, and discharge diagnosis. We collapsed triage level from a 5-level to a 3-level scale for analysis: low (nonurgent), medium (urgent or semiurgent), or high (immediate or emergent). ED staff recorded up to 3 complaints, symptoms, or other reasons for the visit in patients’ own words. NHAMCS central staff abstracted these free text complaints and coded them by using the standardized Reason for Visit Classification for Ambulatory Care.²³ To capture all patient complaints, we used all (i.e., up to 3) recorded codes per patient.

To identify ED discharge diagnoses, we used codes from the International Classification of Diseases, Ninth Revision (ICD-9).²⁴ ED staff recorded up to 3 provider diagnoses as free text, which NHAMCS coded centrally using the ICD-9. To identify diagnoses related to alcohol or drug intoxication, withdrawal, abuse, or dependence, we used ICD-9 codes 291–292, 303–305, 790.3, 962, 965, 967–971, 977, 980, V11.3, and V79.1.²⁵ To identify psychiatric diagnoses, we used codes 290, 293–302, and 306–319. ICD-9 codes 800–959 and 990–999 defined injuries. NHAMCS includes an additional injury variable that asks ED staff to record whether the visit was related to an “injury, poisoning, or adverse effect.” Because this item is not derived from clinician discharge diagnosis and does not distinguish injuries from poisonings and adverse effects of treatment, we defined injuries using discharge diagnoses.

Health services included mode of arrival, length of ED visit, discharge disposition, and length of hospital admission stay.

Statistical Analysis

To obtain nationally representative estimates, we used sampling weights, strata, and clustering variables that NHAMCS provided. We used descriptive statistics to present ED visit characteristics of homeless adults, dichotomized by age (18–49 years vs 50 years and older), using means and SDs for continuous variables and frequencies and proportions for categorical variables. We have presented all results using weighted estimates. We have presented individual results as weighted percentages without the corresponding unweighted N; unweighted summary Ns are shown in the Table 1 footnotes.

TABLE 1—

Characteristics of US Emergency Department Visits by Homeless Adults by Age: 2005–2009

	Weighted ED Visits, %
Characteristics	Aged 18–49 Yearsa	Aged 50–95 Yearsb	P
Demographic
Male	74	80	.17
Race/ethnicity			.72
White	53	57
Black	24	24
Latino	18	15
Other	5	4c
Geography			.54
Northeast	19	17
Midwest	13	16
South	26	23
West	43	45
Insurance			.005
None	42	37
Medicaid	25	20
Medicare	9	20
Other	11	12
Clinical
Triage level			.44
High	16	21
Medium	59	53
Low	9	9
Diagnostic or screening test in ED	68	73	.59
Procedure performed in ED	39	46	.15
Medications received in ED or prescribed at discharge	65	68	.33
Complaintd
Psychiatric	23	15	.01
Injury	21	28	.04
Musculoskeletal	19	16	.37
Alcohol or other drug abuse	12	8c	.1
Request for substance abuse detoxification	40	4c	< .001
Neurologic	10	11	.64
Gastrointestinal	11	12	.73
Respiratory	8	12	.09
Cardiovascular	5	11	.02
Discharge diagnosisd
Psychiatric	20	10	.002
Injury	14	19	.16
Musculoskeletal	10	9c	.7
Substance abuse, overall	35	35	.94
Substance abuse, alcohol only	23	31	.03
Substance abuse, drugs only	15	7c	.003
Neurologic	5	2c	.08
Gastrointestinal	2c	2c	.15
Respiratory	6	4c	.86
Cardiovascular	4	8c	.02
Health services
Arrived by ambulance	36	48	.02
Length of ED visit, mean hr	2	3	.41
Admitted to hospital	11	20	.003
Length of admission of patients admitted to the hospital, mean d	7	8	.22

Note. ED = emergency department. Percentages may not add to 100% because of rounding.

^aUnweighted n = 826.

^bUnweighted n = 351.

^cUnweighted n = 2–27.

^dPatients could have up to 3 complaints or discharge diagnoses.

To compare characteristics of visits by older versus younger adults, we used the t-test and the Rao-Scott χ² test, a design-adjusted version of the Pearson χ² test. According to National Center for Health Statistics statistical guidelines, we considered an estimate to be reliable if it was derived from 30 or more records and had a relative SE of 30% or less.¹⁹ We considered P values < .05 statistically significant.

We conducted analyses using the survey package provided by SAS version 9.2 (SAS Institute, Cary, NC).

RESULTS

From 2005 to 2009, adults in the United States made 468 million ED visits (95% confidence interval [CI] = 428 million, 508 million), or an average of 94 million visits annually. Homeless adults made 0.6% (95% CI = 0.5%, 0.7%) of these visits, or 560 510 ED visits annually. Homeless adults aged 50 years and older accounted for 36% (95% CI = 32%, 40%) of all homeless adults’ ED visits. Adults aged 50 to 64 years made the majority of visits by older homeless patients, and accounted for 32% (95% CI = 28%, 36%) of total visits by homeless persons; homeless adults aged 65 years and older made just 4% of visits (95% CI = 2%, 6%; Figure 1).

FIGURE 1—

Age distribution of US emergency department visits by homeless adults: 2005–2009.

^aResults weighted to generate nationally representative estimates. Total unweighted n = 1177.

^b< 20 includes those aged 18–19 years.

Table 1 shows characteristics of homeless adults’ ED visits, dichotomized by age (18–49 years vs ≥ 50 years). Demographics were similar in older and younger homeless adults, including gender, race/ethnicity, and geographic distribution. Older homeless adults were more likely to have Medicare than were their younger counterparts (20% vs 9%; P = .001).

Certain clinical characteristics were similar in older and younger homeless adults, including triage level and services and medications received in the ED. However, complaints and ED discharge diagnoses differed by age. Older homeless adults were less likely to have psychiatric complaints than were younger homeless adults (15% vs 23%; P = .01) but were more likely to seek care for injuries (28% vs 21%; P = .04) and cardiovascular complaints (11% vs 5%; P = .02). Other types of complaints did not differ by age.

Most discharge diagnoses were psychiatric, substance abuse related, or injury related, accounting for 64% of diagnoses among older homeless adults and 70% among younger ones. Older homeless adults were less likely to have a psychiatric discharge diagnosis than were their younger counterparts (10% vs 20%; P = .002), the same pattern found for psychiatric complaints. Diagnoses at discharge (which the treating clinician recorded) did not always match the patient’s complaint, however. Only 8% of older and 12% of younger homeless adults had a complaint of substance use, but 35% of adults in each age group received a clinician diagnosis of alcohol or drug use at ED discharge. Of older homeless patients with a chief complaint of substance use, only 4% requested alcohol or drug detoxification, compared with 40% of younger homeless patients (P < .001).

Although the overall rate of discharge diagnoses related to substance abuse did not differ by age, older homeless adults were more likely to have alcohol-related diagnoses (31% vs 23%; P = .03) and less likely to have drug-related diagnoses (7% vs 15%; P = .003). Nearly 30% of older homeless patients had a complaint of injury, but only 19% were diagnosed with an injury; reported and diagnosed injuries among younger homeless adults showed a similar but less pronounced discrepancy (21% vs 14%).

Health service use differed by age. Older homeless adults arrived by ambulance more often than did younger homeless adults (48% vs 36%; P = .02), despite similar triage levels. Of older homeless patients who arrived by ambulance, 13% had cardiovascular complaints, compared with 3% of younger homeless adults (P = .002). Rates of other chief complaints among homeless patients arriving by ambulance did not differ by age (data not shown). Older homeless adults were also admitted to the hospital from the ED at higher rates than were younger homeless adults (20% vs 11%; P = .003). Of older homeless patients who were admitted to the hospital, 22% had cardiovascular ED discharge diagnoses, compared with 5% of younger homeless adults (P = .01), but rates of admission for other ED discharge diagnoses did not differ by age (data not shown). Length of stay for patients with hospital admissions did not differ by age.

DISCUSSION

Using a nationally representative sample of ED visits, we found that older homeless patients accounted for more than a third of all visits by homeless adults and that visits by older and younger homeless patients differed in several key aspects. In terms of clinical characteristics, older homeless adults were less likely than were younger homeless adults to have psychiatric complaints or to receive a psychiatric discharge diagnosis. Alcohol-related diagnoses were more common among older homeless adults, whereas drug-related diagnoses were less common. In the category of health services, older homeless adults were more likely than were their younger counterparts to arrive by ambulance and to be admitted to the hospital following a visit to the ED. These differences have important implications for service delivery.

Visits by older homeless patients accounted for more than a third of visits by homeless adults. This proportion is consistent with their reported age distribution in the underlying population. Homeless adults aged 50 years and older made up 32% of the homeless population in 2003, a percentage that has likely increased over the past 9 years.^1–3 Because the proportion of older homeless adults’ ED visits reflects their proportion in the population, if the older homeless population continues to increase as predicted,³ we will see a corresponding increase in the proportion of older homeless adults’ ED visits over the coming decade.

Consistent with previous studies, homeless patients in this study had high rates of ED discharge diagnoses related to mental illness and substance use compared with the general population.^26–28 When compared with their younger counterparts, however, older homeless adults were less likely to have psychiatric chief complaints or discharge diagnoses. Because NHAMCS only includes up to 3 chief complaints or discharge diagnoses, it is possible that older and younger homeless adults had equally high rates of psychiatric chief complaints and discharge diagnoses but that a higher rate of medical illnesses among older homeless adults replaced these complaints and diagnoses. However, whereas older homeless adults did have a significantly higher rate of cardiovascular complaints than did younger homeless adults, rates of neurologic, gastrointestinal, and respiratory complaints did not differ by age. Moreover, the lower rate of psychiatric visits among older homeless adults is consistent with findings in housed adults.²⁹

In the general population, lower rates of ED visits for psychiatric diagnoses²⁹ have been attributed to lower rates of mental illness among older adults³⁰ and to less perceived need for mental health treatment among older adults with mental illness.³¹ Homeless adults may also have these characteristics. Alternatively, older homeless adults with mental illness may be underrepresented among homeless persons visiting the ED because of death or institutionalization. The lower rate of psychiatric visits among older homeless adults suggests that as the homeless population ages, there may be less need for ED-based psychiatric care for homeless adults.

Older and younger homeless adults had similar rates of chief complaints related to substance use, but older adults complaining of substance use were significantly less likely to request alcohol or drug detoxification, supporting previous research.¹⁵ Although older homeless adults were unlikely to request detoxification, they may experience more severe effects of substance use as a consequence of aging.³² For this reason, it may be especially important to screen older homeless patients in the ED for substance use problems.

More than a third of discharge diagnoses for both older and younger homeless adults were related to alcohol or drug intoxication. The similar rate of substance-related diagnoses in older and younger homeless adults contrasts with the general older population, in which the number of ED visits related to substance use decreases with age.³³ Of note, the percentage of discharge diagnoses related to substance use (35% in both age groups) was much higher than was the percentage of patient complaints related to substance use (8% among older homeless adults and 12% among younger). This discrepancy likely reflects a tendency to underreport substance use.^33,34

Although the overall rate of substance use–related discharge diagnoses did not differ by age, older homeless adults had significantly more alcohol-related and fewer drug-related diagnoses than did younger homeless adults. The lower rate of drug-related diagnoses among older homeless adults is consistent with research showing that they are less likely to use illegal drugs than are their younger counterparts.⁶ However, although older homeless adults have been found to be less likely to drink heavily,^1,6 they actually had a higher rate of alcohol-related ED discharge diagnoses than did younger homeless adults. This apparent discrepancy may reflect a tendency for older adults to experience more severe effects of alcohol intoxication,³² leading to more ED visits and more alcohol-related discharge diagnoses.

The high proportion of ED diagnoses related to substance use in older adults underscores the potential to use existing interventions for this vulnerable older population. Like younger homeless adults, older homeless adults use the ED frequently and at rates nearly 4 times higher than the general population.^16–18 Clinicians, researchers, and policymakers have sought to decrease high rates of public services use among homeless adults with substance use problems.³⁵ Two recent trials that provided housing to homeless adults with active substance use found that use of health services and costs of health care decreased after housing was provided.^36,37 Similar programs, adapted for older homeless adults, have the potential to decrease the number of ED visits in the growing older homeless population.

Although older homeless adults were less likely to report than to be diagnosed with substance use, the opposite pattern was found for injuries. Nearly 30% of older homeless patients reported injuries, but only 19% were diagnosed with injuries. Higher reports of injuries relative to diagnoses may reflect differing perceived needs of homeless patients versus those of treating clinicians. For example, a patient who sustained an injury while under the influence of alcohol or drugs may consider the injury to be the reason for the visit, whereas the treating clinician may consider substance use to be the discharge diagnosis.

Nearly one half of older homeless adults arrived by ambulance compared with about one third of younger homeless adults, despite similar triage acuity. Of homeless patients who arrived by ambulance, older adults were more likely to have cardiovascular complaints than were younger adults. Homeless adults’ high rates of ambulance use are thought to be because of physical barriers, including lack of transportation.^26,38 Consistent with these findings, 90% of homeless adults transported to the ED by ambulance are transported nonurgently.²⁶ Higher rates of ambulance use by older versus younger homeless adults may result from higher rates of acute medical conditions (e.g., myocardial infarction) and higher rates of physical barriers, such as functional and mobility impairment.^6,7 For nonurgent ED visits, providing homeless adults with access to alternative transportation may decrease use of costly ambulance services.

Although not unexpected, the finding that older homeless adults had a higher hospital admission rate following an ED visit than did younger homeless adults is important given the expected increase in the proportion of the homeless population aged 50 years and older. Whereas the proportion of older homeless adults’ ED visits reflects their population distribution (approximately one third), their rate of admission was nearly double that of younger homeless adults, accounting for about two thirds of admissions among homeless persons. These data support research showing that older homeless adults had rates of ED use similar to those of younger homeless adults but a trend toward higher rates of hospitalization.¹⁶ Because older homeless adults who were admitted to the hospital had higher rates of cardiovascular ED discharge diagnoses than did younger homeless adults, developing interventions to meet the needs of older homeless adults with cardiovascular disease has the potential to decrease use of the ED among these patients and, in turn, to decrease costs related to hospital admission.

Limitations

This study has several limitations. Because NHAMCS defines homelessness as self-reported patient residence, the number of homeless patients’ visits is likely underestimated. Some EDs may not systematically obtain housing status. Moreover, patients may underreport homelessness because of feelings of shame or distrust and instead report the address of a friend or shelter. Determining the degree to which homeless adults’ visits are underestimated is challenging, as few studies have examined the proportion of all homeless persons’ ED visits; those studies that have were typically based at single urban safety net hospitals and used methods to determine patient residence similar to those NHAMCS used. Reported proportions range from 0.4% (for an urban area with a relatively small homeless population) to 19.5% (at a safety net hospital in San Francisco, CA).^26,39–41

Although differential underreporting of homelessness by older versus younger homeless adults is possible, we are not aware of data that support this concern. Because NHAMCS samples visits rather than patients, we were unable to assess characteristics of visits made by “frequent users,” patients who use the ED multiple times per year and account for a disproportionately high number of ED visits.⁴² Although veterans are overrepresented in the homeless population and especially among homeless men aged 45 to 54 years,⁴³ NHAMCS does not include Veterans Affairs hospitals. Therefore, the results of this study may not be generalizable to homeless veterans. Diagnosis codes may underestimate psychiatric and substance use diagnoses if clinicians do not screen for these conditions.

An additional limitation is the testing of multiple visit characteristics for differences by age. Because of this multiple hypothesis testing, the results should be confirmed in other data sets. Furthermore, because of limited power owing to small sample sizes for some visit characteristics, we were unable to determine if small differences between characteristics of ED visits made by older versus younger homeless adults were statistically significant. However, small differences in visit characteristics are unlikely to be important for clinical practice or health policy. Finally, small sample sizes for homeless adults aged 65 years and older precluded meaningful analyses for this relatively small but important population.

Conclusions

The average age of the homeless population is expected to continue to increase. This aging trend will lead to a corresponding increase in the proportion of ED visits by older homeless patients. We have shown that older homeless adults receive different types of ED care than do their younger counterparts. Given the higher use of ambulances and higher admission rate of older homeless adults, the aging of the homeless population is likely to pose an increased financial burden on health care systems. Health care systems need to account for differences in use of the ED by older homeless adults to meet the needs of the aging homeless population.