Full text of DOI:10.1111/j.1553-2712.2001.tb01114.x

ACADEMIC EMERGENCY MEDICINE • November 2001, Volume 8, Number 11
1051
The Epidemiology of the Homeless Population and Its
Impact on an Urban Emergency Department
J
ASON D’AMORE, MD, OLIVER
HUNG, MD, WILLIAM CHIANG, MD,
L
EWIS OLDFRANK, MD
G
Abstract. Objectives: To characterize the homeless 95% confidence intervals (95% CIs) are given where
adult population of an urban emergency department appropriate: mean age (ϮSD) = 42 Ϯ 10 vs 48 Ϯ 13;
(ED) and study the medical, psychiatric, and social male gender 95% vs 54% (OR = 17; 95% CI = 8 to 37);
factors that contribute to homelessness. Methods: A history of (hx) tuberculosis 49% vs 15% (OR = 2.5;
prospective, case–control survey of all homeless adult 95% CI = 1.2 to 3); hx HIV infection 35% vs 13% (OR
patients presenting to an urban, tertiary care ED and = 3.8; 95% CI = 1.8 to 8); hx penetrating trauma 62%
a random set of non-homeless controls over an eight- vs 16% (OR = 8.62; 95% CI = 4.4 to 17.1); hx depres-
week period during summer 1999. Research assis- sion 70% vs 15% (OR = 13.4; 95% CI = 6.7 to 27); hx
tants administered a 50-item questionnaire and were schizophrenia 27% vs 7% (OR = 5.1; 95% CI = 2.0 to
trained in assessing dentition and triceps skin-fold 14); hx alcoholism 81% vs 15% (OR = 24; 95% CI =
thickness. Inclusion criteria: all homeless adults who 12 to 49); significant tooth loss (>3) 43% vs 18% (OR
consented to participate. Homelessness was defined = 3.3; 95% CI = 1.8 to 6.4); percentage of body fat
as being present for any person not residing at a pri- 16.5% vs 19.7%; hx social isolation (no weekly social
vate address, group home, or drug treatment pro- contacts) 81% vs 11% (OR = 33.3; 95% CI = 14 to 100);
gram. Randomly selected controls were concurrently mean number of ED visits/year 6.0 vs 1.6. Conclu-
enrolled with a 3:1 homeless:control rate. Exclusion sions: In the study population homelessness was as-
criteria: critically ill, injured, or incapacitated pa- sociated with a history of significantly higher rates of
tients, or patients <21 years of age. Univariate anal- infectious disease, ethanol and substance use, psy-
ysis with appropriate statistical tests was used. The chiatric illness, social isolation, and rates of ED uti-
Mantel-Haenszel test was used to adjust for popula- lization. Key words: homeless; public health; epi-
tion differences. Results: Two hundred fifty-two demiology; urban medicne; emergency department.
homeless subjects and 88 controls were enrolled. ACADEMIC EMERGENCY MEDICINE 2001; 8:
Data are presented for homeless vs control patients, 1051–1055
and all p-values were <0.01. Odds ratios (ORs) with
OMELESSNESS remains a social disease of of medical, psychiatric, and social problems such
epidemic proportions. In the United States, as tuberculosis, HIV infection, hepatitis, alcohol-
H
the homeless population is estimated to be be-
tween 4 and 13.5 million persons, and 7.5% of the
general population will become homeless at some
point in their lifetimes.1 Homelessness affects both
genders and all ethnicities with a predilection for
lower socioeconomic classes and urban environ-
ments. In New York City, 0.6% of the total popu-
lation may be homeless, a cohort of 42,500 per-
sons.2 This population is predisposed to a number
ism and substance use, skin and foot disease,
schizophrenia and related psychoses, malnutrition,
and trauma.3–7 The homeless are also less likely
to be integrated into any type of primary care sys-
tem and routinely utilize emergency departments
(EDs) for a majority of their health care needs.8
These factors result in extremely high age-ad-
justed mortality. In a cohort of homeless men in
Philadelphia, the life expectancy was 44 years.9 In
addition to higher mortality, the hospitalization
costs of caring for homeless patients are also sub-
stantially higher. In one study, homelessness was
associated with four additional hospital days per
admission at a cost of $2,000-$4,000 per day.10
In the current study, homeless persons account
for approximately 20–30% of all adult ED visits
comprising 20,000–30,000 visits annually.11 This
high volume is partially explained by the hospital’s
close proximity to a large (>1,000 bed) homeless
men’s shelter. This provides our institution with a
unique perspective on the homeless issue as well
From New York University–Bellevue Department of Emer-
gency Medicine (JDA, WC, LG), New York, NY; and Morristown
Memorial Hospital Department of Emergency Medicine (OH),
Morristown, NJ. Dr. D’Amore is currently in the Department
of Emergency Medicine, North Shore University Hospital,
Manhasset, NY.
Received February 28, 2001; accepted May 22, 2001. Presented
at the SAEM annual meeting, San Francisco, CA, May 2000.
Address for correspondence and reprints: Jason D’Amore, MD,
Department of Emergency Medicine, North Shore University
Hospital, 300 Community Drive, Manhasset, NY 11030. Fax:
212–263–6826; e-mail: jzd29@hotmail.com

1052
HOMELESS
D’Amore et al. • EPIDEMIOLOGY OF HOMELESS POPULATION
Study Setting and Population. This study was
conducted in the adult emergency services of an
inner-city, Level I ED with an annual census of
65,000 adults, of whom approximately 20–30% are
homeless. The survey was administered wherever
these adults were evaluated in the ED.
Survey Contents and Administration. A 50-item
questionnaire consisting of yes/no and single-word-
answer questions including age, sex, ethnicity, ed-
ucational background, duration of residence in
New York City, medical and psychiatric history,
utilization of health care resources (clinics, ED),
eating habits, nutritional status, dental care, to-
bacco, alcohol, and other substance use, and sleep
habits. There were no ‘‘free-form’’ or ‘‘short-an-
swer’’ questions. An assessment of dentition and a
measurement of triceps skin-fold thickness were
also obtained. The discharge diagnosis and final
disposition for each patient were recorded. The
questionnaire was administered in English and
Spanish. Unique patient identifiers were used to
prevent multiple enrollments for the same subject.
A cohort of trained medical student research as-
sociates collected the survey data. The research as-
sociates trained with the survey in the ED for one
week prior to the commencement of data collection.
They were also trained in the measurement of tri-
ceps skin-fold thickness and assessment of denti-
tion. They were available in the department 24
hours a day.
Figure 1. Physical characteristics (%) of the homeless
population and the control patients.
as one of the largest concentrations of homeless
adults in the country. The objective of this study
was to compare the medical, psychiatric, social,
and economic characteristics of an urban, inner-
city ED homeless population with those of a non-
homeless control group. The utilization and effect
of existing public assistance, shelter, and social
programs on the homeless population are also dis-
cussed. Previous studies have addressed solitary
issues, such as the economics of homelessness or
the prevalence of certain disease entities, and have
largely focused on outpatient settings such as clin-
ics or shelters.3–6,8 A careful analysis of the epi-
demiologic characteristics of these patients drawn
from the ED may complement the previously ob-
tained information, thereby facilitating a more
complete assessment of this important public
health issue.
Data Analysis. SPSS (SPSS Inc., Chicago, IL) and
Epi-Info (USD Inc., Snellville, GA) software were
used to analyze the data. Univariate analysis with
appropriate statistical tests for continuous and cat-
egorical data was used. A Mantel-Haenszel test
METHODS
Study Design. This was a prospective, survey-
based, case–control study of homeless adults (>21
years of age) presenting to an inner-city ED en-
rolled concurrently with a random set of un-
matched control patients. All homeless adults who
gave consent to participate were eligible for inclu-
sion. Homelessness was defined as being present
for any person not residing at a private address,
group home, or drug treatment program. Ran-
domly selected controls were concurrently enrolled
with a 3:1 homeless:control rate. Critically ill or
injured patients, patients with an altered senso-
rium (that did not permit response to questions),
or patients whose physicians thought that partic-
ipation would interfere with the delivery of patient
care were excluded. The study was conducted dur-
ing an eight-week period in June, July, and August
1999. The IRB approved the study.
Figure 2. Prevalence of medical and psychiatric condi-
tions (%) among the homeless population and the control
patients. TB Rx = tuberclosis therapy.

ACADEMIC EMERGENCY MEDICINE • November 2001, Volume 8, Number 11
1053
was used to adjust for population differences (age,
gender).
RESULTS
Over an eight-week period from June to August
1999, 252 homeless patients and 88 non-homeless
controls were enrolled in the study.
Physical Characteristics of the Study Popula-
tion (Fig. 1). Mean ages of the patient populations
are presented in years. Tooth counts were per-
formed by direct visualization of the subject’s den-
tition. Teeth were rated as none, 1, 2, 3, or >3 miss-
ing teeth. Percentage of body fat was calculated by
measuring triceps skin-fold thickness with a stan-
dard conversion formula. The difference between
the two populations was statistically significant in
each study category, and all p-values were <0.01.
Figure 3. Prevalence of substance use (%) among the
homeless population and the control patients.
Prevalence of Medical and Psychiatric Condi-
tions (Fig. 2). The prevalence (%) of each condi-
tion is given for both study groups. Answers were
simple yes/no answers [e.g.: Are you HIV-positive?
Have you ever been treated for tuberculosis (TB)?].
The data presented are therefore based on subjec-
tive data reporting. The difference between the two
populations was statistically significant in each
study category, and all p-values were <0.01.
Prevalence of Substance Use (Fig. 3). The prev-
alence (%) of use of each substance is given for the
two populations. Alcoholism is defined as having
been given the diagnosis of alcoholism by a health
care professional or having been through at least
one detoxification program in the past five years;
CAGE questions were not used. Cocaine, heroin,
and marijuana use was defined as a positive re-
sponse to ‘‘habitual’’ or regular use. Methadone use
was determined by asking whether the subject was
in a methadone maintenance program and what
the daily dose of methadone was. Tobacco users
were identified as all subjects who considered
themselves ‘‘smokers,’’ generally more than a half
pack of cigarettes/day. The difference between the
two populations was statistically significant in all
cases, and all p-values were <0.01.
Figure 4. Social and economic factors (%) among the
homeless population and the control patients.
as full-time, regular employment. All p-values are
<0.01.
Utilization of Health Care Resources (Table 1).
The percentage of each group with access to a pri-
mary care physician or clinic is shown. The median
number, 25th percentile, and 75th percentile of
yearly ED visits are provided for each group. The
mean number of annual inpatient hospitalizations
as well as the percentage of each group who re-
ported more than three hospitalizations per year
is also shown. All p-values are <0.01.
Social and Economic Factors (Fig. 4). Data are
presented in percentages. Social isolation is de-
fined as having no meaningful social contacts with
non-homeless persons on a weekly basis. Assault
victims are those who reported being physically as-
saulted in the last ten years. Recipients of public
assistance are those who reported weekly or
monthly receipt of government subsidies or en- Relative Risks Associated with Homelessness
titlements (welfare, disability, Social Security in- (Table 2). The relative risk associated with home-
surance, unemployment). Employment is defined lessness for various medical, psychiatric, and so-

1054
HOMELESS
D’Amore et al. • EPIDEMIOLOGY OF HOMELESS POPULATION
T
ABLE 1. Utilization of Health Care Resources by the
gration into society. In pilot studies, such ap-
proaches have been very successful, resulting in
decreased ED utilization, improved access to pri-
mary care, and an overall decrease in homeless-
ness by as much as 57%.12,13 Even without a com-
plex case management system, it has been shown
that by creating ‘‘compassionate,’’ goal-directed
care for the homeless, reliance on the ED can be
reduced.14 Given the fact that a majority of EDs do
not even have 24-hour in-house social work staff,
this approach would require a significant reallo-
cation of resources.
Homeless Population and the Control Patients
Health Care Resource
Homeless
Control
p-value
Primary physician (%)
Regular clinic (%)
7.5
28
82
83
<0.01
<0.01
n/a
Shelter avoidance (%)
Median no. emergency
department visits/year
(25th, 75th percentiles)
Mean no. admissions/year
(% with >3/year)
30
n/a
6.1 (3, 9)
1.0 (1, 4)
<0.01
<0.01
2.4 (25%) 0.25 (5.5%)
In extreme cases where individuals place them-
selves or others in clear and present danger (i.e.:
chronic debilitating alcoholism), it may be neces-
sary to institute mandatory commitment under
9.39 of the New York State mental health code, not
for the sole purpose of treating any dangerous
medical condition, but for intensive social rehabil-
itation. This unique approach has been imple-
mented several times in our institution for se-
verely debilitated alcoholic individuals whose
frequency of ED presentation averaged once per
day and who presented a clear and present danger
to themselves and others secondary to their alco-
holism. The standard approach of a ‘‘social admis-
sion’’ was an inadequate gesture because most in-
patient teams were unable to address the patient’s
myriad needs and because the patients often
T
ABLE 2. Relative Risks Associated with Homelessness
Disease Process
Relative Risk
95% CI
Tuberculosis
HIV-positive
Depression
2.5
3.8
1.2, 3
1.8, 8
13.4
24
5.1
6.7, 27
12, 49
2, 14
Alcoholism
Schizophrenia
Social isolation
Assault victim
33.3
8.62
14, 100
4.4, 17
cial conditions is provided along with 95% confi-
dence intervals (95% CIs).
DISCUSSION
In our study population homelessness was associ- signed out against medical advice shortly after ad-
ated with extremely high rates of medical illness mission. When these patients were committed to a
(35% HIV-positive, 49% with TB), psychiatric ill- locked psychiatric unit, they were able to receive
ness (27% schizophrenia, 70% depression), alco- needed medical and psychiatric care and could be
holism (81%), substance use (36% use heroin, 35% placed in outpatient programs with close follow-up
use cocaine), and social isolation (81%). The home- under court mandate. This approach was success-
less population is not integrated into a primary ful in re-uniting at least one individual with his
care system, which results in overutilization of the family, although it has not been proven in any type
ED and high rates of hospitalization. Although of prospective study.
66% of the homeless population (in this study) uses
A successful example of this type of program is
existing public assistance programs, these primar- the directly observed therapy (DOT) program for
ily financial subsidies do not address the pervasive the treatment of TB. This program involves com-
medical and social concerns of this population. Ad- munity-based and clinic-based DOT workers who
ditionally, placement into long-term shelters does administer TB medication under direct observa-
not solve the complex medical and psychosocial tion. Patients are served with a co-DOT (health
problems facing the homeless. Approximately 60– commissioner’s order for DOT), and in cases where
70% of our homeless population resides in a shel- the patients are noncompliant with outpatient
ter, a facility adjacent to our hospital in which DOT, they are subject to detention in a locked fa-
many of the men have resided for several years. cility for the duration of their therapy. Of all pa-
However, other than food, shelter, and rudimen- tients served with a co-DOT between 1993 and
tary social services, there are essentially no organ- 1996, only 8% required detention for noncompli-
ized on-site programs that focus on reintegrating ance. In the same time period, there was a 34%
these people into society. Shelter-based programs drop in the overall number of TB cases and a re-
also fail to reach the 30% of our homeless popula- duction of multiple-drug-resistant TB by 75%.15
tion who avoid the shelter system entirely.
The application of this type of intensive program
One possible approach would be to implement expanded to include medical, psychiatric, and so-
an intensive case management approach to home- cial interventions under a single aegis could dras-
lessness wherein each homeless individual would tically alter the size and acuity of the homeless
be assigned to a case manager to facilitate reinte- population.

ACADEMIC EMERGENCY MEDICINE • November 2001, Volume 8, Number 11
1055
An integrated two-tiered system would be the and substance abuse along with their attendant
most efficacious. Homeless persons presenting to risks to themselves and the rest of society will ul-
timately determine the future for this large popu-
lation of patients. These patients exist in our cities
wandering from ED to ED and hospital to hospital
without social, medical, or psychiatric contact.
Only when a critical incident or problem inter-
venes are they brought to the attention of health
care providers, usually for focused, short-term in-
terventions with subsequent discharge to their
previous environment. This approach leads to a vi-
cious circle of ED recidivism, continued homeless-
ness, and excessive morbidity and mortality.
There is little question that the homeless pop-
ulation places an incredible financial and social
burden on society. The real challenge lies not in
quantifying the problem, but in organizing and al-
locating the essential public resources that will be
necessary to combat this critical epidemic. Given
the immense social and financial costs of untreated
homelessness, the investment would be more than
cost-effective.
the ED would be assigned dedicated case mana-
gers who would be responsible for coordinating all
aspects of their care: housing, drug treatment, pri-
mary care, entitlements, and employment. Those
individuals refractory to this approach who repre-
sent a clear and present danger to themselves or
others (due to infectious/psychiatric disease, or
profoundly maladaptive behavior) would be man-
dated to undergo a period of intense inpatient hos-
pitalization focusing on detoxification, treatment,
and rehabilitation. These individuals would rep-
resent the most debilitated of the homeless popu-
lation who, in such an impaired condition due to
alcohol, substance use, or untreated psychiatric ill-
ness, arguably lack clear decisional capacity. In
this manner those individuals who continually
‘‘fall through the cracks’’ could be assisted. This ap-
proach would need to be supported by a body of
case law, and ultimately new legislature.
LIMITATIONS AND FUTURE QUESTIONS
References
1. Link BG, Susser E, Steuve A. Lifetime and five year prev-
alence of homelessness in the United States [abstract]. Am J
Public Health. 1994; 84:1907.
2. Homelessness in New York State. Albany, NY: New York
State Department of Social Services, 1984; vol I:6–10.
3. Gelberg L, Linn L, Smith MH. Health, homelessness, and
poverty: a study of clinic users. Arch Intern Med. 1990; 150:
The primary limitation of this study is the subjec-
tive nature of data reporting. No attempt was
made to objectively validate the answers to survey
questions. Therefore, there may be some over/un-
derstatements of certain types of information.
However, previous research has determined that 2325–30.
4. Arch S, Leake B, Knowles L, Gelberg L. Tuberculosis in
self-reporting by homeless persons on medical is-
sues is relatively accurate.16,17 The close proximity
(three city blocks) of our center to a large homeless
men’s shelter of approximately 1,000 beds acts as
a significant referral bias and may explain why
there were so few female homeless subjects. There
may also be significant external validity issues
since this is only a single center study with a sig-
nificantly higher volume of homeless patients than
may be seen at other centers. The homeless pop-
ulation in this study may also be significantly more
compromised than homeless populations found
elsewhere. A national, multicenter study would be
a logical next step. Finally, at our institution all
patients with a primary psychiatric complaint are
triaged to a psychiatric assessment area (annual
volume 6,000 to 8,000). The fact that these pa-
tients were not included in the survey may serve
to significantly underestimate the psychiatric con-
ditions of both homeless and control populations.
homeless patients: potential for case finding in public emer-
gency departments. Ann Emerg Med. 1998; 32:144–7.
5. Susser E, Valencia E, Conover C. Prevalence of HIV infec-
tion among psychiatric patients in a New York City men’s shel-
ter. Am J Public Health. 1993; 83:568–70.
6. Wrenn S. Immersion foot: a problem of the homeless in the
1990’s. Arch Intern Med. 1991; 151:785–8.
7. Padgett DK, Struening EL, Andrews H. Predictors of emer-
gency use by homeless adults in New York City: the influence
of predisposing, enabling, and need factors. Soc Sci Med. 1995;
41:547–56.
8. Gallagher TC, Anderson RM, Koegel P. Determinants of reg-
ular sources of care among homeless adults in Los Angeles.
Med Care. 1997; 35:814–30.
9. Hibbs JR, Brenner L, Klugman L. Mortality of homeless
adults in Philadelphia. N Engl J Med. 1994; 331:305–9.
10. Salit SA, Kuhn EM, Hartz AJ. Hospitalization costs asso-
ciated with homelessness in New York City. N Engl J Med.
1998; 338:1734–40.
11. Goldfrank L. Caring for homeless patients: challenge for
the 90’s [abstract]. Hosp Physician. 1991; 4:13.
12. Okin RL, Boccelar A, Azocar F. The effects of clinical case
management on hospital service use among frequent ED users.
Am J Emerg Med. 2000; 18:603–8.
13. Plumb JD, McManus P, Carson L. A collaborative com-
munity approach to homeless care. Prim Care Clin. 1996; 23(1):
17–25.
14. Redelmeier DA, Molin JP, Tibshirani RJ. A randomized
trial of compassionate care for the homeless in an emergency
department. J Accid Emerg Med. 1996; 13:415–7.
15. Fujiwara P, Larkin C, Frieden TR. Directly observed ther-
apy in New York City: history, implementation, results, and
challenges. Clin Chest Med. 1997; 18(1):135–48.
16. Gelberg L, Siecke N. Accuracy of homeless adults self-re-
ports. Med Care. 1997; 35:287–90.
C
ONCLUSIONS
The resolution of the battle between the civil lib-
erty to be homeless and the societal obligation to
care for men and women who suffer from un-
17. Anderson R, Kasper J, Frankel MR. Total Survey Error.
treated psychosis, TB, HIV infection, alcoholism, San Francisco, CA: Jossey-Bass Publishers, 1979.