778 | Research and Practice | Peer Reviewed | Kushel et al. American Journal of Public Health | May 2002, Vol 92, No. 5
ï£¦ RESEARCH AND PRACTICE ï£¦
Objectives. This study examined factors associated with emergency department use
among homeless and marginally housed persons.
Methods. Interviews were conducted with 2578 homeless and marginally housed persons, and factors associated with different patterns of emergency department use were
assessed in multivariate models.
Results. Findings showed that 40.4% of respondents had 1 or more emergency department encounters in the previous year; 7.9% exhibited high rates of use (more than
3 visits) and accounted for 54.5% of all visits. Factors associated with high use rates
included less stable housing, victimization, arrests, physical and mental illness, and substance abuse. Predisposing and need factors appeared to drive emergency department
Conclusions. Efforts to reduce emergency department use among the homeless
should be targeted toward addressing underlying risk factors among those exhibiting high
rates of use. (Am J Public Health. 2002;92:778â€“784)
Emergency Department Use
Among the Homeless and Marginally Housed:
Results From a Community-Based Study
| Margot B. Kushel, MD, Sharon Perry, PhD, David Bangsberg, MD, MPH, Richard Clark, MPH, and Andrew R Moss, PhD
Although population-based studies of the
homeless have revealed high rates of emergency department use, these studies have not
examined repeated use among particular
groups of individuals. Similarly, emergency
departmentâ€“based studies have demonstrated that homelessness is associated with
repeated use but have not attempted to ascertain the proportions of homeless persons who
exhibit repeated use or the factors associated
with repeated use. We used a communitybased survey of homeless and marginally
housed persons in San Francisco to examine
patterns of emergency department use, factors associated with use, and factors associated with repeated use.
Participants and Setting
To survey a community-based population
of homeless persons, we recruited individuals
from homeless shelters and food lines. Because homeless persons are likely to spend
part of their time in substandard housing, we
also recruited individuals from low-rent, single-room-occupancy (SRO) hotels. Between
April 1996 and December 1997, we recruited a sample of 2578 English-speaking
adults. At that time, it was estimated that
there were about 5000 literally homeless persons living on the street or in homeless shelters in San Francisco and between 6000 and
8000 persons residing in low-rent hotels in
the area we studied.
The sampling design, which was based on
a design developed by Burnam and Koegel,23
involved a multistage cluster sample with
stratification into shelters, free meal programs,
and SRO hotels in San Francisco. The shelter
sample was drawn from all overnight shelters
that housed at least 50 adults per night. The
meal program sample included participants in
5 of 6 midday free-meal programs that
served a minimum of 100 adults at least 3
days per week. SRO hotels were sampled, at
a probability rate proportionate to size, from
292 low-cost residential hotels in census
tracts located in the center of the city. Hotels
eligible for inclusion were those that, according to the cityâ€™s official list, had at least 20
â€œusually occupiedâ€ residential (nontourist)
rooms licensed to rent for $400 per month or
Within each sampling site, individuals were
recruited through the use of a systematic
High rates of emergency department use create a strain on the health care system by leading to overcrowding,1,2 but they can also be
seen as a marker of systemic problems, including poor access to nonemergency health
care and the failure to prevent injuries and illnesses.3 A recent report noted increasing
emergency department use nationwide, which
contributes to overcrowding.4 Figures indicating high rates of use can reflect both a large
proportion of people using the emergency department occasionally and a small proportion
of people using it repeatedly. Among individuals without medical insurance or access to
primary medical care, the emergency department can serve as the only available source
of care.3,5,6 Population-based studies have
shown that homeless persons have high rates
of emergency department use; compared with
the general population, the homeless are 3
times more likely to use an emergency department at least once in a year.7 Emergency
departmentâ€“based studies have also shown
that homelessness is associated with repeated
emergency department use.8,9
Homeless persons are at high risk for requiring emergency department services because of their elevated rates both of unintentional injuries and of traumatic injuries from
assault10â€“12 and because of their poor health
status and high rates of morbidity.11,13â€“16
Other factors commonly associated with
homeless individualsâ€™ receipt of nonurgent
medical care in emergency departments include lack of health insurance,17 lack of transportation,18 lack of a telephone,18 poor access
to primary care,5,19,20 inner-city residence,3
minority status,20 chronic alcohol and drug
abuse,21,22 and mental illness.10 These factors
may contribute both to the high rates of
emergency department use seen in homeless
populations and to the association between
homelessness and repeated use.
May 2002, Vol 92, No. 5 | American Journal of Public Health Kushel et al. | Peer Reviewed | Research and Practice | 779
ï£¦ RESEARCH AND PRACTICE ï£¦
sampling design. Hotels that provided special
in-house programs (e.g., health clinics or advocacy services) were excluded because of
the possibility that health use practices among
residents of such hotels are not representative. We also excluded respondents who appeared severely intoxicated or belligerent and
those who were unable to provide informed
consent. Respondents did not give their
names or other identifying information; each
respondent was assigned a unique identifier
constructed from individual data to eliminate
Interviews were conducted twice per
month in community settings near each sampling site. Individuals were reimbursed either
$10 (shelter or meal program recruits) or $15
(hotel recruits) for completing interviews. All
participants underwent a 45-minute interview
conducted by trained field staff using a standard questionnaire that involved a structured
response format. All responses were selfreports with the exception of HIV status,
which was determined through serological
testing. Health service use was not validated
with medical records.
We used the behavioral model for vulnerable populations,24 an adaptation of Andersenâ€™s behavioral model,25 as the conceptual
framework for our analysis. According to this
model, peopleâ€™s use of health care is affected
by predisposing, enabling, and need factors.
We assessed the relationships between our
outcomes of interest and (1) the predisposing
factors of age, sex, ethnicity, education, housing status, criminal history, victimization, substance abuse, and mental illness; (2) the enabling factors of income, medical insurance,
and receipt of public benefits; and (3) the
need factors of self-reported health status,
chronic illness, and HIV status. Some factors,
such as victimization, mental illness, and substance abuse, can be seen as either predisposing factors (if they contribute to an individualâ€™s overall vulnerability) or need factors (if
their presence is the proximate cause of receipt of health care).10,24
Health Service Use
Frequency and volume of emergency department encounters. The primary outcome of interest was self-reported number of emergency
department encounters in the previous 12
months. Respondents were asked to report
any emergency department visit for any reason, including visits for psychiatric reasons.
These responses were grouped into intensity
categories representing 0, 1, 2 or 3, and 4 or
more encounters. We computed the total
number of emergency department encounters
reported by all participants and estimated the
percentage of total encounters attributable to
each intensity level. We defined repeated use
as 4 or more emergency department encounters in the past year.
Use of ambulatory care services. Ambulatory
care use was defined as any health care visit
for the purpose of physical health in a nonâ€“
emergency department, non-inpatient setting
in the previous year. Number of contacts was
not ascertained. We defined exclusive use of
the emergency department in the previous
year as emergency department use but no
ambulatory care use.
Inpatient hospitalization. Participants were
asked whether they had spent the night in the
hospital for a physical problem in the previous
12 months (nights spent in a psychiatric hospital, in an emergency department, or in a hospital lobby or waiting room were excluded).
Predisposing factors. Predisposing factors included age (less than 35, 35â€“50, more than
50 years), sex, ethnicity (White, African
American, Latino, â€œotherâ€), education (less
than high school, high school or more), duration of homelessness (more than 1 year,
â€œotherâ€), housing status, mental illness and
substance abuse, housing history, and history
of victimization and arrests.
Trained interviewers used a residential
calendar to assess the housing history of all
respondents. Participants reported where
they had spent each night during the previous week and month and then, through the
use of key remembered events, estimated
where they had spent nights during the past
year. We considered participants who reported that they spent at least 90% of their
nights in a hotel and spent no nights living
on the street or in a shelter as â€œmarginally
housed.â€ All others were considered to be
Participants were asked whether they had
been arrested in the past year. They were
also asked whether they had been a victim of
property theft, assault with robbery, physical
assault, or sexual assault in the previous year.
We included any such event as an episode of
We asked respondents whether they had a
history of psychiatric hospitalization and
whether they had ever used injection drugs.
We also asked whether they had had a drug
or alcohol problem in the past year; those
who answered in the affirmative were considered to have a drug or alcohol problem.
Enabling factors. We classified income as
total monthly income from all benefits, earnings (legal and illegal), and panhandling or
donations. Respondents were grouped according to whether they reported receiving
Supplemental Security Income or Social Security Disability Insurance and according to
their medical insurance status (uninsured,
Medicaid or Medicare, veteransâ€™ insurance,
Need factors. Need factors included health
status, chronic illnesses, and HIV status. Respondents were asked whether they considered their health to be excellent, very good,
good, fair, or poor. They were also asked
whether a physician had ever told them that
they had emphysema or chronic bronchitis,
asthma, hypertension, diabetes, heart disease,
or stroke. Those who indicated that they had
been diagnosed with at least 1 of these conditions were categorized as having a chronic
medical condition. As mentioned earlier, HIV
status was determined via serological testing.
We examined factors associated with any
emergency department encounter and with 4
or more encounters in the previous 12
months, defining the dependent variable dichotomously in each case. We also examined
factors associated with the occurrence of at
least 1 ambulatory care visit in the past year.
Univariate associations were assessed with
Student t tests or Wilcoxon rank sum tests
(for continuous or ordinal terms) and with Ï‡2
tests (for categorical terms). In separate models, we used stepwise logistic regression to
characterize adjusted odd ratios for any emergency department encounter and for 4 or
780 | Research and Practice | Peer Reviewed | Kushel et al. American Journal of Public Health | May 2002, Vol 92, No. 5
ï£¦ RESEARCH AND PRACTICE ï£¦
TABLE 1â€”Characteristics of Participant Groups: San Francisco, 1996â€“1997
All Participants Any ED Use â‰¥4 ED Encounters
Characteristic (n = 2532) (n = 1022) (n = 199)
Age, y, mean (SD) 43.3 (10.8) 42.0 (10.0)*** 41.7 (9.2)**
Age, y, no. (%)
< 35 498 (19.7) 225 (22.0) 41 (20.6)
35â€“50 1479 (58.4) 609 (59.6) 126 (63.3)
> 50 555 (21.9) 188 (18.4) 32 (16.1)
Male, no. (%) 1967 (77.7) 751 (73.5)*** 138 (69.4)***
Ethnicity, no. (%)
White 988 (39.0) 454 (44.4)*** 87 (43.7)*
African American 1122 (44.3) 426 (41.7)** 77 (38.7)*
Latino 136 (5.4) 41 (4.0) 9 (4.5)
Other 286 (11.3) 101 (9.9) 26 (13.1)
Did not complete high school, no. (%) 667 (26.3) 268 (26.2) 56 (28.1)
Veteran 600 (23.7) 226 (22.1)* 41 (20.6)
Homeless more than 1 year, no. (%) 1008 (39.8) 433 (42.4)** 104 (52.3)***
Nights on street, mean (median) 88 (12) 96 (30)*** 107 (46)***
Marginally housed, no. (%) 589 (23.3) 171 (16.7)*** 21 (10.6)***
All sources, mean $ (median) 631 (500) 631 (545) 624 (576)
SSI or SSDI, no. (%) 710 (28.0) 315 (30.8)** 73 (36.7)***
Health insurance, no. (%)
Medicaid or Medicare 891 (35.2) 398 (38.9)*** 96 (48.2)***
Veteransâ€™ insurance 337 (13.3) 123 (12.0) 24 (12.1)
Other 163 (6.4) 71 (7.0) 8 (4.0)
Uninsured 1322 (52.2) 516 (50.5)* 87 (43.7)**
Crime/victimization, no. (%)
Arrested in past year 735 (29.0) 373 (36.5)*** 86 (43.2)***
Crime victim in past year 1447 (57.2) 729 (71.3)*** 159 (79.9)***
Mental health/substance abuse, no. (%)
Mental health inpatient (lifetime) 564 (22.3) 282 (27.6)*** 75 (37.7)***
Drug or alcohol problem in past year 1218 (48.1) 549 (53.7)*** 125 (62.8)***
Need factors, no. (%)
Excellent/very good 886 (35.0) 291 (28.5) 39 (19.6)
Good 766 (30.3) 277 (27.1) 45 (22.6)
Fair/poor 880 (34.8) 454 (44.4)*** 115 (57.8)***
HIV positive 217 (8.6) 91 (8.9) 15 (7.5)
Medical comorbidity 703 (27.8) 376 (36.8)*** 103 (51.8)***
Note. See text for descriptions of variables. ED = emergency department; SSI = Social Security Income; SSDI = Social Security
*.05 < P < .20; **.01 < P < .05; ***P < .01 (vs other respondents in bivariate logistic regression).
more encounters; candidate covariates were
factors with a significance level below .25 as
determined in univariate analyses (Table 1).
We adjusted results of stepwise models for
demographic characteristics (age, sex, and
ethnicity) and other factors to display a common vector of covariates. We validated final
models with the Hosmerâ€“Lemeshow test.
A total of 2578 individuals completed the
questionnaire (two thirds of those approached
agreed to participate). Forty-six responses
were excluded on the basis of missing data;
2532 (98.2%) questionnaires were included
in the analysis. There were no significant differences between respondents and nonrespondents in terms of sex, race, or ethnicity.
More than three quarters of the respondents were men; the mean reported age was
43 years (range: 15â€“77 years). Respondents
were predominantly White (39.0%) and African American (44.3%; Table 1).
Housing status. Forty percent of the respondents reported having been homeless for
more than a year. Respondents had spent a
mean number of 88 days living on the street
or in a shelter in the previous year. One fifth
(20.5%) reported spending most of the past
year living on the street or in a shelter. In addition, 589 respondents (23.3%) reported
spending at least 90% of the days in the past
year living in a hotel and spending no nights
on the street or in a shelter; thus, they were
classified as marginally housed.
Crime and victimization. Almost a quarter
of the respondents (22.6%) reported ever
having spent time in prison. Twenty-nine percent had been arrested in the previous year,
and more than half (57.2%) had been a victim of crime in that period.
Mental illness and substance abuse. Almost a
quarter of the respondents reported a history
of psychiatric hospitalization, and 41.7% reported ever having used injection drugs. Almost half (48.1%) considered themselves to
have had a drug or alcohol problem in the
Overall mean monthly income was $631.
Almost 85% of respondents reported a formal source of monthly income, such as state
general assistance (39.4%), Social Security Income or Social Security Disability Insurance
(28.0%), a job (19.9%), or veteransâ€™ benefits
May 2002, Vol 92, No. 5 | American Journal of Public Health Kushel et al. | Peer Reviewed | Research and Practice | 781
ï£¦ RESEARCH AND PRACTICE ï£¦
FIGURE 1â€”Frequency and volume of emergency department (ED) encounters in a 12-month period among homeless and marginally housed
individuals (n=2532): San Francisco, 1996â€“1997.
TABLE 2â€”Health Service Encounters
in the Previous 12 Months Among
Homeless and Marginally Housed
Individuals (n=2532): San Francisco,
Encounter Sample, No. (%)
Any ED use 1022 (40.4%)
No. of ED visits
0 1510 (59.6)
1 469 (18.5)
2â€“3 354 (14.0)
â‰¥4 199 (7.9)
Any ambulatory 1171 (46.3)
Inpatient hospitalization 367 (14.5)
for physical illness
Note. ED = emergency department.
(2.3%). Of respondents with formal incomes,
41.4% also reported casual sources, including
selling bottles and cans, help from family or
friends, selling drugs, and sex work; 392
(15.0%) respondents had only casual sources
of income. In all, 1322 respondents (52.2%)
were medically uninsured; 35.2% had Medicaid or Medicare insurance, and 13.3% had
More than a quarter of the respondents
(27.8%) reported having at least 1 of the 5
chronic health problems assessed (heart disease or stroke, high blood pressure, asthma,
diabetes, or chronic bronchitis/emphysema).
More than a third (34.8%) reported their
health status as fair or poor. According to
serological testing, 8.6% of the respondents
were HIV positive.
Health Care Use
Emergency department use. Among the respondents, 1022 (40.4%) reported that they
had received care in an emergency department in the previous year (Table 2). Almost a
fifth (18.5%) reported having had 1 emergency department visit in the past year,
14.0% reported 2 or 3 visits, and 7.9% reported 4 or more visits (Figure 1). In terms of
exclusive use, 18.4% of all respondents reported receiving outpatient care only in an
emergency department in the past year;
45.6% of all emergency department users
were exclusive users.
In a multivariate analysis of all respondents, factors associated with any use of an
emergency department in the past year included the following: younger age, female
sex, White ethnicity, less stable housing
(being homeless as opposed to marginally
housed), worse health status (having medical
comorbidities or being in fair or poor health),
Medicaid or Medicare insurance (as compared with no insurance), and involvement
with crime (as either a victim or a perpetrator). There was a trend toward an association
between a history of psychiatric hospitalization and emergency department use, but this
relationship was not significant. Neither substance abuse nor HIV status was associated
with emergency department use (Table 3).
Repeated use. The 199 respondents (7.9%)
who reported 4 or more emergency department visits in the previous year accounted for
55% of all visits reported (Figure 1). In a
multivariate model comparing respondents
782 | Research and Practice | Peer Reviewed | Kushel et al. American Journal of Public Health | May 2002, Vol 92, No. 5
ï£¦ RESEARCH AND PRACTICE ï£¦
TABLE 3â€”Multivariate Factors Associated With Emergency Department (ED) Use in the Previous
12 Months Among Homeless and Marginally Housed Individuals: San Francisco, 1996â€“1997
Any ED visits (n = 1022) 4 or More ED Visits (n = 199)a
Adjusted 95% Confidence Adjusted 95% Confidence
Characteristic Odds Ratio Interval P Odds Ratio Interval P
Age (10-year increase) 0.84 0.77, 0.92 < .01 0.86 0.73, 1.01 .06
Male 0.76 0.62, 0.94 < .01 0.77 0.55, 1.09 .14
Ethnicity < .01 .33
White . . . . . .
African American 0.76 0.63, 0.92 < .01 0.81 0.58, 1.14 .23
Latino 0.47 0.31, 0.72 < .01 0.62 0.29, 1.33 .22
Other 0.68 0.51, 0.92 .01 1.12 0.69, 1.84 .64
Marginally housed 0.61 0.49, 0.76 < .01 0.42 0.26, 0.69 < .01
Medicaid/Medicare insurance (reference: uninsured) 1.24 1.02, 1.51 .03 1.49 1.07, 2.07 .02
Medical Comorbidity 1.94 1.59, 2.37 < .01 2.57 1.86, 3.55 < .01
Health status fair/poor (reference: good to excellent) 1.74 1.44, 2.1 < .01 2.01 1.45, 2.78 < .01
Alcohol/drug problem in past 12 months 1.10 0.92, 1.32 .28 1.41 1.02, 1.94 .04
History of psychiatric hospitalization 1.20 0.97, 1.48 .09 1.53 1.10, 2.14 .01
Arrested in previous 12 months 1.53 1.27, 1.86 < .01 1.65 1.19, 2.28 < .01
Crime victim 2.24 1.87, 2.68 < .01 2.26 1.55, 3.28 < .01
Reference is 3 or fewer ED visits.
with 4 or more visits and those with 3 or
fewer visits in the previous year (Table 3), the
following factors were associated with repeated use: younger age, female sex, less stable housing (homeless vs marginally housed),
Medicaid or Medicare insurance (vs no insurance), poorer health status (comorbid illness
or fair or poor health), involvement in crime
(as either perpetrator or victim), mental illness
(history of psychiatric hospitalization), and
substance abuse (drug or alcohol problem).
Other encounters. Almost half (46.3%) of
the respondents reported at least 1 ambulatory care visit in the previous year. Among
people who did not use the emergency department, fewer than half (40.7%) reported
an ambulatory care visit. Among those with
any emergency department use, more than
half (54.4%) reported such a visit. Finally,
among those exhibiting a high rate of use,
59.8% reported an ambulatory care visit. In a
multivariate model examining factors associated with making at least 1 ambulatory care
visit in the past year (n=1171; model not
shown), older persons, those with higher incomes, crime victims, those in fair or poor
health, and those with medical comorbidities
were more likely to have made at least 1 such
visit. African Americans were less likely to
have made an ambulatory care visit.
Whereas insurance status was associated
with ambulatory episodes in the univariate
analysis, there was no such association in the
multivariate model. More than a third
(34.6%) of the overall sample had had no
contact with a physician (emergency department, ambulatory care, or inpatient care) in
the past year. Overall, 361 (14%) respondents
reported at least 1 inpatient stay for physical
illness in the previous 12 months. Among the
199 respondents with 4 or more emergency
department visits, 56.4% (n=114) had been
hospitalized at least once.
In this community-based sample of homeless and marginally housed persons, we found
that 40% had used an emergency department at least once in the previous year, a rate
3 times the US norm.4 However, it was the
persons classified as repeated users (i.e., 4 or
more emergency department encounters in
the past year)â€”less than 8% of the total sampleâ€”who accounted for the majority of the
total emergency department use. Concerns
about emergency department overcrowding
have led to a focus on reduction of use
among the homeless.9 Our study suggests that
such efforts should be targeted specifically toward homeless individuals who exhibit repeated emergency department use, given that
these individuals account for a disproportionate amount of emergency department use.
Predisposing and need factorsâ€”less stable
housing, chronic medical illness, and victimizationâ€”predominated in our models of emergency department use. The majority of our respondents exhibited high levels of housing
instability, spending, on average, 3 months a
year on the street or in shelters. Respondents
who were marginally housed, spending almost
all of their nights in SRO hotels and none in
the street, were significantly less likely to use
the emergency department or to be repeated
users. Previous research has linked housing
instability with more use of ambulatory care
and less use of acute care services.7,26,27 This
study adds support to such findings. The effects of lack of housing, which include exposure to violence, problems in managing
chronic medical conditions, and difficulty in
planning for health care, may increase emergency department use.
May 2002, Vol 92, No. 5 | American Journal of Public Health Kushel et al. | Peer Reviewed | Research and Practice | 783
ï£¦ RESEARCH AND PRACTICE ï£¦
Much has been written about the overrepresentation of homeless persons among users
of emergency departments.9,28 Our study suggests that the homeless do access emergency
departments in large numbers but that they
may not have their medical needs met in
other forums. Almost half of those who used
an emergency department used it as their
only source of health care, and half of those
who received care in nonemergency ambulatory care settings also used the emergency department. This suggests that the features of,
and services offered by, emergency departments (e.g., accessibility at all hours of the
day, availability of care without an appointment, treatment of acute injuries and severe
illnesses) may encourage greater use. The
predominance of need factors in our models
supports previous research suggesting that
homeless people use emergency departments
according to medical need.10
Acute injuries are an important predictor of
emergency department use by the homeless.10
In the present study, victimization was highly
associated with exclusive emergency department use, any use, and repeated use. Victimization is ordinarily considered a predisposing
factor; if it is the proximate cause of emergency department use, however, it can also
be considered a need factor.10 Injuries caused
by victimization may not be amenable to
treatment in the primary care setting, in that
they demand urgent attention, may occur
when primary care is not available, and may
require services not available at nonemergency ambulatory care sites.
An important finding of our study was that
public health insurance was associated with
higher rates of emergency department use.
Contrary to findings in an earlier national
study of homeless persons,7 insurance was
not independently associated with ambulatory care in the present study. The reason
may be that San Francisco has an extensive
system of health care for the uninsured and
homeless. The cityâ€™s network of federally
funded community health centers and integrated system of public health care services
include 13 clinics funded by the department
of public health, a public nursing home, and a
public hospital. There are also a variety of
clinics and outreach services specifically targeted to homeless persons.
In our sample, in which there was a
broader penetration of public insurance
among disabled individuals, insurance may
have been a marker for higher levels of physical and mental disability. Further work is
needed to more fully explore the complex interaction of health insurance status, emergency department use, and ambulatory care
use patterns in this population.
Homeless individuals with repeated emergency department use represent an extreme
example of the complications of homelessness. Many of the same factors that are associated with any emergency department use
are associated with repeated use: poorer
health, less stable housing, and involvement
in crime. In the present study, however, individuals exhibiting high use rates were more
likely than the total population of homeless
ED users to have substance abuse and mental
health problems. Because psychiatric emergency department visits were included in our
measure of use, it is possible that mental illness and substance abuse represent need factors and lead directly to use of emergency departments.
This study involved several limitations. All
responses were self-reports without medical
record validation. Previous studies have
shown that homeless persons are not significantly less accurate in reporting health care
use than the general population, although
they may be less accurate in reporting frequency of use.27,29 We did not have information on frequency of ambulatory care visits or
on whether these visits signaled the presence
of a regular provider. It is possible that ambulatory care, as assessed in the present study,
does not represent health care received from
a primary care provider with whom a patient
has a continuous relationship.
In addition, we had no way of assessing
whether the emergency department visits that
were reported represented appropriate use or
whether problems could have been addressed
in nonemergency settings. Because of its extensive range of available health care services
for the uninsured and homeless, San Francisco may not be representative of other
urban centers. Finally, this study was cross
sectional; we do not know whether the associations we found are causal. The same factors that lead people to be more stably
housed may enable them to access ambulatory care and decrease their use of emergency departments.
One of the strengths of our study was that
respondents were drawn from a community
sample, and we were able to gather information on both those with and those without access to the health care system. In addition, detailed information about residential history
allowed us both to include and to differentiate between those who used residential hotels
intermittently and those who lived stably in
This study raises questions about the limits
of medical interventions (e.g., provision of insurance or ambulatory care) designed to decrease emergency department use among
homeless persons. Addressing the needs of
homeless individuals who exhibit repeated
emergency department use represents a particular challenge but may lead to the greatest
reductions in use. The high rates of hospitalization and comparatively extensive use of
ambulatory care among those demonstrating
repeated emergency department use support
the hypothesis that need for services drives
homeless individualsâ€™ use of emergency departments.
Interventions providing health insurance or
ambulatory care alternatives may not, on
their own, be able to decrease emergency department use. The finding that those who live
stably in residential hotels use fewer emergency department services suggests that provision of housing, particularly to the small
proportion of homeless individuals who exhibit repeated use, may help decrease reliance on emergency departments and improve
health care outcomes. This policy option merits further exploration.
About the Authors
Margot B. Kushel is with the Division of General Internal
Medicine, University of California at San Francisco/San
Francisco General Hospital, San Francisco, Calif. Sharon
Perry, David Bangsberg, Richard Clark, and Andrew R.
Moss are with the Department of Epidemiology and Biostatistics, University of California at San Francisco. David
Bangsberg is also with the Division of Infectious Diseases,
Department of Medicine, University of California at San
Francisco/San Francisco General Hospital.
Requests for reprints should be sent to Margot B.
Kushel, MD, Division of General Internal Medicine,
UCSF/San Francisco General Hospital, Box 1364, San
Francisco, CA 94143 (e-mail: [email protected]).
This article was accepted November 26, 2001.
784 | Research and Practice | Peer Reviewed | Kushel et al. American Journal of Public Health | May 2002, Vol 92, No. 5
ï£¦ RESEARCH AND PRACTICE ï£¦
M.B. Kushel was responsible for the study design; she
participated in the analysis and was responsible for the
interpretation of the data and the drafting and final revision of the manuscript. S. Perry was responsible for
data management, conducted and assisted in the interpretation of the analysis, and participated in the revision of the manuscript. D. Bangsberg assisted with the
study design, interpretation, and revision of the manuscript. R. Clark participated in the study design, in data
collection, and in revision of the manuscript. A.R. Moss
conceived of the study; supervised the collection, analysis, and interpretation of data; and participated in revision of manuscript.
This work was supported by National Heart, Lung, and
Blood Institute grant 55729 and National Institute of
Mental Health grant 54907. David Bangsbergâ€™s work
was supported in part by the Doris Duke Charitable Foundation. Margot B. Kushelâ€™s work was supported
in part by a faculty development grant (5f D08 HP
50109) from the US Department of Health and Human
Services. The University of California, San Francisco,
Committee on Human Research approved the study.
We thank Marjorie Robertson, PhD, for acting as
study director during the data collection period and for
supervising work on study design and implementation,
instrument development, field work implementation,
and initial data management.
1. Richards JR, Navarro ML, Derlet RW. Survey of
directors of emergency departments in California on
overcrowding. West J Med. 2000;172:385â€“388.
2. Derlet R, Richards J, Kravitz R. Frequent overcrowding in US emergency departments. Acad Emerg
3. Tyrance PH Jr, Himmelstein DU, Woolhandler S.
US emergency department costs: no emergency. Am J
Public Health. 1996;86:1527â€“1531.
4. National Hospital Ambulatory Medical Care Survey:
1999 Emergency Department Summary. Hyattsville, Md:
National Center for Health Statistics; 2001. Report 320.
5. Grumbach K, Keane D, Bindman A. Primary care
and public emergency department overcrowding. Am J
Public Health. 1993;83:372â€“378.
6. The Medicaid Access Study Group. Access of
Medicaid recipients to outpatient care. N Engl J Med.
7. Kushel MB, Vittinghoff E, Haas JS. Factors associated with the health care utilization of homeless persons. JAMA. 2001;285:200â€“206.
8. Okin RL, Boccellari A, Azocar F, et al. The effects
of clinical case management on hospital service use
among ED frequent users. Am J Emerg Med. 2000;18:
9. Mandelberg JH, Kuhn RE, Kohn MA. Epidemiologic analysis of an urban, public emergency departmentâ€™s frequent users. Acad Emerg Med. 2000;7:
10. Padgett DK, Struening EL, Andrews H, Pittman J.
Predictors of emergency room use by homeless adults
in New York City: the influence of predisposing, enabling and need factors. Soc Sci Med. 1995;41:
11. Brickner PW, Scanlan BC, Conanan B, et al.
Homeless persons and health care. Ann Intern Med.
12. Padgett DK, Struening EL. Victimization and traumatic injuries among the homeless: associations with
alcohol, drug, and mental problems. Am J Orthopsychiatry. 1992;62:525â€“534.
13. Hwang SW, Lebow JM, Bierer MF, Oâ€™Connell JJ,
Orav EJ, Brennan TA. Risk factors for death in homeless adults in Boston. Arch Intern Med. 1998;158:
14. Hwang SW. Mortality among men using homeless
shelters in Toronto, Ontario. JAMA. 2000;283:
15. Hwang SW, Orav EJ, Oâ€™Connell JJ, Lebow JM,
Brennan TA. Causes of death in homeless adults in
Boston. Ann Intern Med. 1997;126:625â€“628.
16. Hibbs JR, Benner L, Klugman L, et al. Mortality in
a cohort of homeless adults in Philadelphia. N Engl J
17. Baker DW, Stevens CD, Brook RH. Determinants
of emergency department use: are race and ethnicity
important? Ann Emerg Med. 1996;28:677â€“682.
18. Rask KJ, Williams MV, Parker RM, McNagny SE.
Obstacles predicting lack of a regular provider and delays in seeking care for patients at an urban public hospital. JAMA. 1994;271:1931â€“1933.
19. Gill JM, Mainous AG III, Nsereko M. The effect of
continuity of care on emergency department use. Arch
Fam Med. 2000;9:333â€“338.
20. Baker DW, Stevens CD, Brook RH. Regular
source of ambulatory care and medical care utilization
by patients presenting to a public hospital emergency
department. JAMA. 1994;271:1909â€“1912.
21. McGeary KA, French MT. Illicit drug use and
emergency room utilization. Health Serv Res. 2000;35:
22. Cherpitel CJ. Drinking patterns and problems,
drug use and health services utilization: a comparison
of two regions in the US general population. Drug Alcohol Depend. 1999;53:231â€“237.
23. Burnam MA, Koegel P. Methodology for obtaining
a representative sample of homeless persons: the Los
Angeles Skid Row Study. Eval Rev. 1988;12:117â€“152.
24. Gelberg L, Andersen RM, Leake BD. The behavioral model for vulnerable populations: application to
medical care use and outcomes for homeless people.
Health Serv Res. 2000;34:1273â€“1302.
25. Andersen RM. Revisiting the behavioral model
and access to medical care: does it matter? J Health Soc
26. Duchon LM, Weitzman BC, Shinn M. The relationship of residential instability to medical care utilization among poor mothers in New York City. Med Care.
27. Oâ€™Toole TP, Gibbon JL, Hanusa BH, Fine MJ. Utilization of health care services among subgroups of
urban homeless and housed poor. J Health Polit Policy
28. Padgett DK, Brodsky B. Psychosocial factors influencing non-urgent use of the emergency room: a review of the literature and recommendations for research and improved service delivery. Soc Sci Med.
29. Gelberg L, Siecke N. Accuracy of homeless adultsâ€™
self-reports. Med Care. 1997;35:287â€“290.
American Public Health Association
E-mail: [email protected]
Tel: (301) 893-1894
FAX: (301) 843-0159
This book describes the varied spectrum of work done at the local public
health level, and how practitioners take
the lead in social justice today. The wide
array of public health department approaches, such as budgeting, staffing, services, involvement in personal health
services, and their relationships with
states is disclosed.
This book is an incredible resource
for: local public health officers, administrators, and state and local health planners for use in their own local public
2000 âš 281 pages âš softcover
$24.50 APHA Members
plus shipping and handling
Local Public Health
Trends & Models
By Glen P. Mays, PhD, MPH;
C. Arden Miller, MD; and
Paul K. Halverson, DrPH, MHSA
778 | Research and Practice | Peer Reviewed | Kushel et al. American Journal of Public Health | May 2002, Vol 92, No. 5