 RESEARCH AND PRACTICE 

Revolving Doors: Imprisonment Among the Homeless
and Marginally Housed Population
| Margot B. Kushel, MD, Judith A. Hahn, PhD, MPH, Jennifer L. Evans, MS, David R. Bangsberg, MD, MPH, and Andrew R. Moss, PhD

People who were homeless at the time of arrest are overrepresented in prisons.1 Additionally, homeless populations include higher proportions of former prisoners compared with
the general population, 2 and inmates who are
released from prison have a high risk for
homelessness.3–5 The association between
homelessness and imprisonment is bidirectional: imprisonment may precipitate homelessness by disrupting family and community
contacts and by decreasing employment and
housing prospects.5–7 Homelessness may increase the risk for imprisonment through
shared risk factors and through increased
likelihood of arrest.
Both substance abuse and mental illness
are important risk factors for homelessness
and imprisonment.7–16 Homeless persons who
have mental health and substance abuse disorders have low rates of receipt of treatment
for their disorders.17 Mentally ill inmates are
more likely to have been homeless in the
year before their arrest than non–mentally ill
inmates,1 and inmates who had been homeless were more likely to be mentally ill than
inmates who had not been homeless immediately before their arrest.18 Homelessness also
increases the risk for recidivism among former prisoners.19
Homeless and prison populations have
high rates of communicable diseases because
of poor health, unsafe sexual practices, illicit
drug use, and close living quarters.20–23
Among homeless mentally ill persons, those
who have a history of incarceration have elevated rates of psychiatric problems and substance abuse disorders.24 It is not known
whether homeless persons who have a history of imprisonment are more likely to have
poor health status and to be more at-risk for
infectious disease than homeless persons who
do not have a history of imprisonment. We
hypothesized that homeless persons who had
a history of imprisonment would have higher
rates of substance abuse disorders, mental

Objectives. We studied a sample of homeless and marginally housed adults to
examine whether a history of imprisonment was associated with differences in
health status, drug use, and sexual behaviors among the homeless.
Methods. We interviewed 1426 community-based homeless and marginally
housed adults. We used multivariate models to analyze factors associated with
a history of imprisonment.
Results. Almost one fourth of participants (23.1%) had a history of imprisonment. Models that examined lifetime substance use showed cocaine use (odds
ratio [OR] = 1.67; 95% confidence interval [CI] = 1.04, 2.70), heroin use (OR = 1.51;
95% CI = 1.07, 2.12), mental illness (OR = 1.41; 95% CI = 1.01, 1.96), HIV infection
(OR = 1.69; 95% CI = 1.07, 2.64), and having had more than 100 sexual partners
were associated with a history of imprisonment. Models that examined recent substance use showed past-year heroin use (OR = 1.65; 95% CI = 1.14, 2.38) and
methamphetamine use (OR = 1.49; 95% CI = 1.00, 2.21) were associated with lifetime imprisonment. Currently selling drugs also was associated with lifetime imprisonment.
Conclusions. Despite high levels of health risks among all homeless and marginally housed people, the levels among homeless former prisoners were even
higher. Efforts to eradicate homelessness also must include the unmet needs of
inmates who are released from prison. (Am J Public Health. 2005;95:1747–1752.
doi:10.2105/AJPH.2005.065094)
health disorders, physical health problems,
and illegal activities than those who did not
have a history of imprisonment. We assessed
a sample of homeless and marginally housed
individuals to compare whether persons who
had a history of imprisonment differed from
persons who did not have a history of imprisonment regarding (a) lifetime health and illegal activities and (b) current health and illegal
activities.

METHODS
Sampling Design
Our sample was composed of homeless
and marginally housed adults in San Francisco, Calif. During a 12-month period beginning in April 1999, we surveyed 1426 English-speaking adults in 5 overnight shelters
(50 adults per night minimum), 6 midday
free-meal programs (100 adults at least 3
days per week minimum), and 28 low-income
residential hotels (sampled with probability
proportional to size). We included residential

October 2005, Vol 95, No. 10 | American Journal of Public Health

hotels in low-income San Francisco neighborhoods that rented rooms for less than $400
per month. We conducted systematic sampling within each sampling venue.
We invited recruits to participate in a comprehensive interview that was conducted at or
near each sampling site. Rather than record
names or other personal identifying information, we created a unique study ID code for
each respondent, which was used to eliminate
duplicate participants. Shelter and meal program recruits received a $20 cash incentive;
hotel recruits received $25. We did not find
significant gender or racial/ethnic differences
between participants and nonparticipants.

Instrument
Trained interviewers conducted a structured interview (average = 45 minutes). We
assessed background characteristics, including
age, gender, racial/ethnic self-identification,
education, marital/partner status, and income
from all sources during the past 30 days. We
defined health status by asking participants to

Kushel et al. | Peer Reviewed | Research and Practice | 1747

 RESEARCH AND PRACTICE 

report their current perceived health status
with a five-point scale; we dichotomized responses into “fair or poor” health and “excellent, very good, and good” health. After
pretest counseling and receipt of informed
consent, we tested all participants for HIV
infection with enzyme-linked Immunosorbent assay (ELISA) antibody testing and
Western blot confirmation (Unilab, Tarzana,
Calif).

Imprisonment
We asked participants whether they had
ever been in prison and, if so, how much time
they had spent in prison, their last release
date, and their current probation or parole
status. We defined a lifetime history of imprisonment as having reported a prison stay in
state or federal penitentiaries (we did not include incarceration in jails).
Housing status. Participants were given a
12-month follow-back calendar, with important dates as a guide, to identify types of
places where they spent the night during the
past 12 months and the number of nights
spent in each type of place. We defined participants who spent at least 90% of nonincarcerated/nonhospitalized nights in a hotel,
apartment, or private home and who spent no
nights on the street or in a shelter as marginally housed. Anyone who had fewer than
90% of nonincarcerated nights in a hotel,
apartment, or private home and who had
spent any time staying on the streets or in
shelters was defined as homeless. We also
asked participants about their lifetime history
of homelessness.
Substance use, sexual behaviors, and mental
health. Participants were asked about their
drinking history, including whether or not
they thought that they had a drinking problem during the past year or ever. Those who
said yes were classified as having an alcohol
problem during the past year or ever.
We asked participants about their use of illicit drugs, including crack cocaine, cocaine,
heroin, methamphetamines, other opiates (for
which the participant did not have a physician’s prescription, including illicit methadone), and use of injection drugs. We also
asked participants whether they used these
drugs during their lifetime and, if so, whether
they had used illicit drugs during the past

year. We classified those who reported any of
these activities as having used illicit drugs.
Participants were asked about their sexual
behaviors, including whether they had opposite and/or same-sex partners and how many
partners they had (0–5, 6–10, 11–25,
26–50, 51–100, and > 100). We considered
those in the highest quintile of numbers of
sexual partners as having high numbers of
sexual partners.
Lastly, we asked participants whether they
had ever been admitted to an inpatient psychiatric facility and whether they had been
admitted during the past year. These answers
were used as proxies for mental illness.

Sources of Income
Participants were asked to identify all
sources of income during the past 30 days.
Details about the sampling strategy and
the interview methods have been published
elsewhere.22,25,26

Analysis
We tested for bivariate and multivariate
associations with a lifetime history of imprisonment. We analyzed the association with
lifetime history of behaviors to determine
whether homeless and marginally housed
persons who had a history of imprisonment
also had different behavior patterns from
those who did not have a history of imprisonment. We then analyzed the association with
current behaviors to determine whether any
differences persisted after release from
prison. We used the Wilcoxon rank sum test
for continuous variables and the χ2 test for
categorical variables to test for bivariate
associations. Variables for the multivariate
analyses were chosen on the basis of our hypotheses that persons who have a history of
imprisonment will have worse health and
mental health status and higher rates of drug
use and multiple sexual partners (P < .05 in
bivariate analyses). We tested for multicollinearity with Pearson correlation coefficients, and we validated final models with
the Hosmer–Lemeshow test. All analyses
were conducted with unweighted data.

RESULTS
We enrolled 1426 of 2029 respondents
(response rate = 70.3%); 1325 participants

1748 | Research and Practice | Peer Reviewed | Kushel et al.

had complete data on imprisonment and
were included in our analysis. The majority of
respondents were White (40.2%) and Black
(43.9%) men (74.9%) who were aged 30 to
39 years (24.3%) or 40 to 49 years (42.5%)
and who reported a median monthly income
of $650 (Table 1). More than one third
(43.4%) met our criteria for being marginally
housed and most (86.7%) reported that they
had been homeless at some point.
Participants had high rates of mental illness
and substance use. Twenty-five percent reported a lifetime history of psychiatric hospitalization, although less than 10% reported an
inpatient psychiatric admission during the
past year. Sixty percent of participants reported illicit drug use at least once during the
past year. Crack cocaine was the most common drug used, with 50% of participants reporting use during the past year. Most participants (84.4%) reported drug use at some
point during their life.
Almost a quarter of participants (23.9%)
reported an alcohol problem during the past
year, and almost half (44%) reported an alcohol problem during their lifetime. More than
a quarter of participants (27.7%) reported
sexual activity with same-sex partners, and
20.3% reported at least 100 sexual partners
during their lifetime. More than a third of participants (37.7%) reported that their health
was fair or poor, and 11.2% tested positive
for HIV infection.

Imprisonment
Almost a quarter of respondents (23.1%)
had been incarcerated in a prison during
their lifetime. Participants who had a history
of imprisonment had a median time of 6.4
years since last being released. They had
spent a median time of 4 years in prison;
3.8% of participants reported having been released from prison during the past year, and
4.4% reported being on parole.

Lifetime Behaviors Associated With
Lifetime Imprisonment
There was a strong bivariate association
between a history of imprisonment and lifetime history of drug use: 93.1% of all persons
who had a history of imprisonment reported
drug use during their lifetime compared with
81.7% of all persons who did not have a

American Journal of Public Health | October 2005, Vol 95, No. 10

 RESEARCH AND PRACTICE 

TABLE 1—Characteristics of the Study Sample: San Francisco, California, April 1999
Overall Sample
No. (%)
(N = 1325)

Never Imprisoned
No. (%)
(n = 1019)

Ever Imprisoned
No. (%)
(n = 306)

Pa

.05

Age, y
< 30

69 (5.3)

62 (6.2)

7 (2.3)

30–39

319 (24.3)

245 (24.3)

74 (24.2)

40–49

559 (42.5)

428 (42.5)

131 (42.8)

≥ 50

367 (27.9)

273 (27.1)

94 (30.7)

Gender
Male

990 (74.9)

728 (71.7)

262 (85.6)

Female

332 (25.1)

288 (28.4)

44 (14.4)

Non-Latino White

530 (40.2)

410 (40.4)

120 (39.3)

Black

580 (43.9)

446 (43.9)

134 (43.9)
23 (7.5)

<.001

Race/ethnicity

Latino

81 (6.1)

58 (5.7)

Asian/Pacific Islander

31 (2.4)

25 (2.5)

6 (2.0)

Other

98 (7.7)

76 (7.5)

22 (7.2)

.81

Education
< High School

368 (27.9)

265 (26.1)

103 (33.8)

High school graduate

483 (36.6)

371 (36.5)

112 (36.7)

> High school

470 (35.6)

380 (37.4)

90 (29.5)

711 (53.8)

553 (54.5)

158 (51.6)

.011

Marital status
Never married/partnered
Married/partnered

.04

93 (7.0)

60 (5.9)

33 (10.8)

Separated/divorced/widowed

517 (39.1)

402 (39.6)

115 (37.6)

Gay or bisexual

364 (27.7)

296 (29.3)

68 (22.3)

.02

Men who have sex with menb

280 (28.5)

233 (32.3)

47 (18.0)

<.001

82 (24.9)

61 (21.3)

21 (47.7)

<.001

1044 (79.7)

818 (81.4)

226 (74.1)

.01

Menb

759 (77.3)

563 (78.2)

196 (74.8)

.26

Womenc

284 (87.4)

254 (90.1)

30 (69.8)

<.001

266 (20.3)

187 (18.6)

79 (25.9)

<.01

223 (22.7)

157 (21.8)

66 (25.2)

.26

41 (12.6)

28 (9.9)

13 (30.2)

<.001

Women who have sex with womenc
No. of sex partners (lifetime)
0–100 sexual partners

> 100 sexual partners
Menb
Womenc

Current Behaviors Associated With
Lifetime Imprisonment

Housing status
Marginally housed (past year)d

575 (43.4)

449 (44.1)

126 (41.2)

.37

1148 (86.7)

879 (86.4)

269 (87.9)

.48

610 (355–780)

600 (355–770)

672 (364–780)

Fair or poor pealth

500 (37.7)

361 (35.4)

139 (45.4)

.002

HIV infection

147 (11.2)

102 (10.1)

45 (14.9)

.02

Psychiatric hospitalization (lifetime)

341 (25.8)

250 (24.6)

91 (29.7)

.07

Psychiatric hospitalization (past year)

110 (8.4)

81 (8.0)

29 (9.5)

.41

Homeless (ever)
Income (monthly)
Median
Health status

Mental health

Continued

October 2005, Vol 95, No. 10 | American Journal of Public Health

history of imprisonment (P < .001). Former
prisoners were more likely to have HIV infection than those who had never been imprisoned (14.9% versus 10.1%; P = 0.02) and
were slightly more likely to have been hospitalized in a psychiatric facility (29.7% versus
24.6%; P = 0.07). In a multivariate model
that examined factors associated with a lifetime history of imprisonment, ever having
used crack or cocaine (odds ratio [OR] = 1.67;
95% confidence interval [CI] = 1.04, 2.70)
and ever having used heroin (OR = 1.51;
95% CI = 1.07, 2.12) were associated with a
history of imprisonment. Ever having been
hospitalized in a psychiatric facility (OR =
1.41; 95% CI = 1.01, 1.96), being in fair or
poor health (OR = 1.47; 95% CI = 1.09,
1.99), and having HIV infection (OR = 1.69;
95% CI = 1.07, 2.64) also were associated
with a history of imprisonment, as were being
older, having less than a college education,
and being a man (OR = 4.28; 95% CI = 2.60,
7.05). Having a lifetime history of more than
100 sexual partners was associated with a
history of imprisonment (OR = 1.44; 95%
CI = 1.02, 2.02). The odds for a history of imprisonment were lower among men who had
sex with men (OR = 0.35; 95% CI = 0.23,
0.53), but the odds increased among women
who had sex with women (OR = 2.35; 95%
CI = 1.12, 4.91) (Table 2).

Participants who had a history of imprisonment had high rates of past-year drug use,
with more than two thirds (69.6%) reporting
illicit drug use during the previous year. Almost ten percent (9.6%) reported having sold
drugs during the previous month compared
with 2.7% who did not have a history of imprisonment. There was a strong independent
association between currently selling drugs
and a history of imprisonment (OR = 2.57;
95% CI = 1.36, 4.85). Both past-year heroin
use (OR = 1.65; 95% CI = 1.14, 2.38) and
methamphetamine use (OR = 1.49; 95%
CI = 1.00, 2.21) were associated with ever
having been imprisoned, but past-year crack
or cocaine use were not. Other significant factors included older age, lower educational attainment, fair or poor health status, HIV infection, psychiatric hospitalization, being

Kushel et al. | Peer Reviewed | Research and Practice | 1749

 RESEARCH AND PRACTICE 

TABLE 1—Continued
Substance use (lifetime)
Illicit drug use

1116 (84.4)

832 (81.7)

284 (93.1)

<.001

604 (45.8)

424 (41.8)

180 (59.4)

<.001

1024 (77.4)

757 (74.4)

267 (87.5)

<.001

Methamphetamine use

692 (52.3)

496 (48.8)

196 (64.3)

<.001

Heroin use

595 (45.0)

410 (40.3)

185 (60.7)

<.001

Drinking probleme (ever)

580 (44.0)

438 (43.1)

142 (47.0)

.23

791 (59.7)

578 (56.7)

213 (69.6)

<.001

Injection drug use
Powder or crack cocaine use

Substance use (past year)
Illegal drug use
Injection drug use

367 (27.7)

247 (24.2)

120 (39.2)

<.001

Powder or crack cocaine use

675 (51.0)

495 (48.6)

180 (59.0)

.001

Methamphetamine use

247 (18.7)

174 (17.1)

73 (23.9)

<.01

Heroin use

304 (23.0)

199 (19.6)

105 (34.4)

<.001

Alcohol probleme (past year)

315 (23.9)

246 (24.3)

69 (22.7)

0.57

56 (4.3)

27 (2.7)

29 (9.6)

<.001

Income from selling drugs (past 30 days)
a

P = comparison between never imprisoned and ever imprisoned.
Denominator = total number of men (n = 990).
c
Denominator = total number of women (n = 332).
d
Spent more than 90% of noninstitutionalized nights in a residential hotel, apartment, or private home.
e
Self–reported.
b

male, and having had more than 100 sexual
partners (Table 3). In a model that adjusted
both for lifetime history and past-year drug
use, the association with past-year drug use
was no longer significant (data not shown).
Currently selling drugs remained strongly associated with a history of imprisonment
(OR = 2.90; 95% CI = 1.56, 5.39), even after
adjustment for a lifetime history of drug use.
We did not collect data on lifetime history of
selling drugs.

DISCUSSION
In our study, participants reported high
rates of lifetime imprisonment: approximately
one quarter had been imprisoned in a state
or federal prison at some point during their
lifetime. Those who had been imprisoned
were more likely to have a history of psychiatric hospitalizations, drug use, multiple sexual partners ( > 100), and HIV infection than
those who had not been imprisoned. The proportion who had ever been imprisoned was
slightly higher than the proportion found in a
national sample of homeless persons2 and
was significantly higher than the proportion
found in the general US population.27 The association between homelessness and impris-

onment is complex because of shared risk factors and causal pathways in both directions.
Studies have shown that prisoners are at
high risk for becoming homeless at the time
of their release. Exiting prisoners face important challenges to successfully reestablishing
community life, including difficulties with securing housing and employment.28–32 They
also have difficulty obtaining medical, mental
health, and substance abuse treatment after
their release.33 A report released in 1998
stated that 10% of parolees in California were
homeless; in San Francisco and Los Angeles,
the estimates were 30% to 50%.34 The fact
that former prisoners remained in the homeless and marginally housed community more
than 6 years after their release is the result of
(1) the persistence of risk factors common to
imprisonment and homelessness and (2) the
difficulties ex-prisoners experience when they
reintegrate into community life.
We found that having a history of psychiatric hospitalization was independently associated with a history of imprisonment. Mental
illness is a risk factor for both homelessness
and imprisonment.11 People who have mental
illnesses have higher rates of imprisonment
than the general population: an estimated 5%
of the overall population has a serious mental

1750 | Research and Practice | Peer Reviewed | Kushel et al.

TABLE 2—Characteristics and
Multivariate Lifetime Behaviors
Associated With Ever Being Imprisoned
(N = 1325): San Francisco, April 1999
Adjusted
Odds Ratio
(95% Confidence
Interval)
Characteristics
Age, y
< 30 years
1.00
30–39 years
2.23 (0.94, 5.31)
40–49 years
2.39 (1.02, 5.59)
≥ 50 years
3.04 (1.27, 7.26)
Gender
Female
1.00
Male
4.28 (2.60, 7.05)
Race/ethnicity
Non–Latino White
1.00
Black
1.02 (0.72, 1.43)
Other
1.15 (0.75, 1.75)
Education
< High school
1.00
High school graduate
0.78 (0.55, 1.11)
> High school
0.56 (0.38, 0.81)
Health status
Good, very good or excellent
1.00
Fair or poor health
1.47 (1.09, 1.99)
No HIV infection
1.00
HIV infection
1.69 (1.07, 2.64)
Mental health
No psychiatric hospitalization 1.00
Psychiatric hospitalization in
1.41 (1.01, 1.96)
lifetime
Lifetime behaviorsa associated with
ever being imprisonedb
Substance use (lifetime)
Never used crack/cocaine
1.00
Crack/cocaine use
1.67 (1.04, 2.70)
Never used methamphetamines 1.00
Methamphetamines use
1.33 (0.92, 1.93)
Never used heroin <
Heroin use
1.51 (1.07, 2.12)
Sexual behaviors
Men who have sex with women 1.00
Men who have sex with men
0.35 (0.23, 0.53)
Women who have sex with men 1.00
Women who have sex with women 2.35 (1.12, 4.91)
0–100 sexual partners
1.00
> 100 sexual partners
1.44 (1.02, 2.02)
a

Models adjusted for all variables listed.
Includes lifetime substance use variables.

b

American Journal of Public Health | October 2005, Vol 95, No. 10

 RESEARCH AND PRACTICE 

TABLE 3—Characteristics and
Multivariate Past-Year Behaviors
Associated With Ever Being Imprisoned
(N = 1325): San Francisco, April 1999
Adjusted
Odds Ratio
(95% Confidence
Interval)
Characteristics
Age, y
< 30
1.00
30–39
2.64 (1.09, 6.39)
40–49
3.14 (1.32, 7.51)
≥ 50
3.69 (1.51, 9.01)
Gender
Female
1.00
Male
4.68 (2.83, 7.72)
Race/ethnicity
Non-Latino White
1.00
Black
1.05 (0.75, 1.47)
Other
1.12 (0.74, 1.71)
Education
< High school
1.00
High school graduate
0.76 (0.53, 1.07)
> High school
0.54 (0.37, 0.78)
Income
Other sources
1.00
Selling drugs (past 30 days)
2.57 (1.36, 4.85)
Health status
Good, very good or excellent
1.00
Fair or poor health
1.44 (1.07, 1.96)
No HIV infection
1.00
HIV infection
1.65 (1.04, 2.62)
Mental health
No psychiatric hospitalization 1.00
Psychiatric hospitalization
1.40 (1.00, 1.95)
(lifetime)
Past-year behaviorsa associated with
ever being imprisonedb
Substance use (past year)
Never used crack/cocaine
1.00
Crack/cocaine use
1.07 (0.78, 1.48)
Never used methamphetamines 1.00
Methamphetamines use
1.49 (1.00, 2.21)
Never used heroin
1.00
Heroin use
1.65 (1.14, 2.38)
Sexual behaviors
Men who have sex with women 1.00
Men who have sex with men
0.34 (0.22, 0.52)
Women who have sex with men 1.00
Women who have sex with women 2.81 (1.33, 5.90)
0–100 sexual partners
1.00
> 100 sexual partners
1.46 (1.04, 2.06)
a

Models adjusted for all variables listed.
Includes past year substance use variables.

b

illness compared with 10% to 20% of the imprisoned population.11,35 However, within
prison and following release, there are limited
resources for receiving mental health care.36
Community-based mental health care facilities
may be unable to offer care to certain exoffenders, including those who have a history
of dangerous behavior.37 Among homeless
persons, this tendency to not receive mental
health care may be exacerbated.17
We found that illicit drug use was associated with imprisonment. More than 70% of
federal inmates and 80% of state and local
inmates reported a lifetime history of substance abuse38; however, only a small proportion received substance abuse services while
incarcerated.39 Imprisonment for drug offenses increased 16-fold between the early
1980s and the late 1990s40 and accounted
for much of the rise in prison populations.
Currently selling drugs remained highly associated with a history of imprisonment, even
after we controlled for drug use. Selling drugs
puts an individual at higher risk for involvement with the criminal justice system; after
prison release, persons who have a history of
imprisonment may find it particularly difficult
to gain employment in the legitimate labor
market.28
HIV infection remained independently associated with a history of imprisonment. Former prisoners had higher rates of HIV infection, had high numbers of sexual partners,
and had higher rates of active drug use compared with the homeless population at large.
Studies have estimated that 2.3% of imprisoned persons are known to be HIV positive,7
although these rates may underestimate the
true prevalence. We found rates 10 times that
high among homeless and marginally housed
persons who had been imprisoned. Homeless
persons who had a history of imprisonment
also had higher rates of HIV infection and
were in fair or poor health, even after we
controlled for drug use, injection drug use,
sexual preference, and number of sexual partners. HIV, tuberculosis, and hepatitis C are
common in both the homeless and prison
populations.23,41,42 Both homelessness and imprisonment may foster environments in which
communicable diseases are easily spread by
placing high-risk persons in close proximity to
one another.

October 2005, Vol 95, No. 10 | American Journal of Public Health

We did not find an association between
being Black and imprisonment among the
homeless population. Black Americans are
more likely than White Americans to be imprisoned and are more likely to be homeless.2,27 We believe our not finding a difference between rates of imprisonment on the
basis of race/ethnicity may be the result of
the differential effect of race/ethnicity on
homelessness and the effect of imprisonment
as a causal factor for homelessness. Within
the homeless population, the differences
among the general population in rates of imprisonment no longer hold.

Limitations
Our study has several limitations that affected our ability to draw conclusions. Because the study is cross-sectional, we were unable to draw causal conclusions about the
association between homelessness and imprisonment. We did not know whether imprisonment preceded or followed episodes of homelessness. All results, except for HIV status,
were self-reported; estimates of imprisonment
and reported participation in illegal activities
may have been underreported. We did not
have diagnostic information on mental illness;
rather, we used psychiatric hospitalization as
a proxy for mental illness, which likely underestimated the true rate of mental illness. We
used a 1-question assessment of drinking status; however, the use of 2 positive responses
to the CAGE questionnaire or the use of 5 or
more drinks daily did not change our results.
Our study excluded non-English-speaking
homeless people; we do not know if nonEnglish speakers are at higher or lower risk
for imprisonment.

Conclusion
High rates of imprisonment among homeless populations may be the end result of a
system that does not provide access to timely
services, including access to housing, health
care, mental health care, and substance abuse
treatment, and systems that have obstacles
preventing receipt of these services by people
exiting prison. High rates of HIV infection
among homeless ex-prisoners and high rates
of continued risky behavior provide motivation for targeting risk reduction efforts at
persons exiting prison. The intersection of

Kushel et al. | Peer Reviewed | Research and Practice | 1751

 RESEARCH AND PRACTICE 

substance abuse, unemployment, imprisonment, and homelessness is potent and lasting.
Efforts to eradicate homelessness also must
include the many unmet needs of persons
exiting prison.

About the Authors
Margot B. Kushel is with the Division of General Internal
Medicine, University of California at San Francisco at San
Francisco General Hospital, San Francisco, Calif. Judith A.
Hahn is with the Epi-Center, Department of Medicine,
University of California at San Francisco at San Francisco
General Hospital. Jennifer L. Evans and Andrew R. Moss
are with the Department of Epidemiology and Biostatistics,
University of California, San Francisco. David R. Bangsberg is with the Division of Infectious Diseases and the
Positive Health Program, University of California at San
Francisco at San Francisco General Hospital.
Requests for reprints should be sent to Margot B.
Kushel, MD, UCSF at SFGH, Box 1364, San Francisco,
CA 94143 (e-mail: kushel@itsa.ucsf.edu).
This article was accepted April 18, 2005.

Contributors
M. B. Kushel originated the study, led the writing, and
synthesized the analysis. J. A. Hahn and J. L. Evans conducted the analysis. All the authors originated ideas
and designed the study. J. A. Hahn, D. R. Bangsberg, and
A. R. Moss interpreted findings; contributed to writing;
obtained funding; and originated, designed, and supervised data collection and analysis.

Acknowledgments
This project received funding from the National Institute of Mental Health (grant R0154907). M. B. Kushel
received funding from the Agency for Healthcare Research and Quality (grant 1K08 HS 11415) and from
the Hellman Family Award for Junior Faculty. D. R.
Bangsberg received funding from the Doris Duke Charitable Foundation.
We thank Clifford Wilson for his help with the article.

Human Participant Protection
The committee on human research at the University of
California, San Francisco, approved this study.

References
1. Ditton P. Mental Health and Treatment of Inmates
and Probationers. Washington, DC: US Dept of Justice,
Bureau of Justice Statistics; 1999.
2. Burt M, Aran L, Douglas T, Valente J, Lee E, Iwen
B. Homelessness: Programs and the People They Serve:
Findings from the National Survey of Homeless Assistance
Providers and Clients, Technical Report. Washington, DC:
Urban Institute; 1999.
3. Desai RA, Lam J, Rosenheck RA. Childhood risk
factors for criminal justice involvement in a sample of
homeless people with serious mental illness. J Nerv
Ment Dis. 2000;188:324–332.
4. Martell DA, Rosner R, Harmon RB. Base-rate estimates of criminal behavior by homeless mentally ill persons in New York City. Psychiatr Serv. 1995;46:596–601.
5. Center for Poverty Solutions. Barriers to Stability:
Homelessness and Incarceration’s Revolving Door in Bal-

timore City. Baltimore, Md: Center for Poverty Solutions; 2003.
6. Solomon P, Draine J. Using clinical and criminal
involvement factors to explain homelessness among
clients of a psychiatric probation and parole service.
Psychiatr Q. 1999;70:75–87.
7. Freudenberg N. Jails, prisons, and the health of
urban populations: a review of the impact of the correctional system on community health. J Urban Health.
2001;78:214–235.
8. Belcher JR. Are jails replacing the mental health
system for the homeless mentally ill? Community Ment
Health J. 1988;24:185–195.
9. Gelberg L, Linn LS, Leake BD. Mental health, alcohol and drug use, and criminal history among homeless adults. Am J Psychiatry. 1988;145:191–196.
10. Greene JM, Ennett ST, Ringwalt CL. Prevalence
and correlates of survival sex among runaway and
homeless youth. Am J Public Health. 1999;89:
1406–1409.
11. Lamb HR, Weinberger LE. Persons with severe
mental illness in jails and prisons: a review. Psychiatr
Serv. 1998;49:483–492.
12. Wenzel SL, Gelberg L, Bakhtiar L, et al. Indicators
of chronic homelessness among veterans. Hosp Community Psychiatry. 1993;44:1172–1176.
13. Linn LS, Gelberg L, Leake B. Substance abuse
and mental health status of homeless and domiciled
low-income users of a medical clinic. Hosp Community
Psychiatry. 1990;41:306–310.
14. Fischer PJ, Breakey WR. The epidemiology of alcohol, drug, and mental disorders among homeless persons. Am Psychol. 1991;46:1115–1128.
15. Breakey WR, Fischer PJ, Kramer M, et al. Health
and mental health problems of homeless men and
women in Baltimore. JAMA. 1989;262:1352–1357.
16. Rock M. Emerging issues with mentally ill offenders: causes and social consequences. Adm Policy Ment
Health. 2001;28:165–180.
17. Koegel P, Sullivan G, Burnam A, Morton SC,
Wenzel S. Utilization of mental health and substance
abuse services among homeless adults in Los Angeles.
Med Care. 1999;37:306–317.

24. McGuire JF, Rosenheck RA. Criminal history as a
prognostic indicator in the treatment of homeless people with severe mental illness. Psychiatr Serv. 2004;55:
42–48.
25. Kushel MB, Evans JL, Perry S, Robertson MJ,
Moss AR. No door to lock: victimization among homeless and marginally housed persons. Arch Intern Med.
2003;163:2492–2499.
26. Robertson MJ, Clark RA, Charlebois ED, et al.
HIV seroprevalence among homeless and marginally
housed adults in San Francisco. Am J Public Health.
2004;94:1207–1217.
27. Bonczar TP. Prevalence of Imprisonment in the US
Population, 1974–2001. Washington, DC: Bureau of
Justice Statistics; 2003 August. Report No. NCJ
197976.
28. Petersilia J. When prisoners return to communities. Federal Probation. 2001;65:3–8.
29. Bradley KH, Oliver RBM, Richardson NC, Slayter
EM. No Place Like Home: Housing and the Ex-prisoner.
Boston, Mass: Community Resources for Justice, Inc.;
2001.
30. Pager D. The mark of a criminal record. Am J Sociol. 2003;108:937–975.
31. Davies S, Tanner J. The long arm of the law: effects of labeling on employment. Sociological Q. 2003;
44:385–404.
32. Visher CA, Travis J. Transitions from prison to
community: understanding individual pathways. Ann
Rev Sociol. 2003;29:89–113.
33. Visher CA, Naser RL, Baer D, Jannetta J. In Need
of Help: Experiences of Seriously Ill Prisoners Returning
to Cincinnati. Washington DC: Urban Institute; 2005.
34. Beyond Bars: Correctional Reforms to Lower Prison
Costs and Reduce Crimes. Sacramento, Calif: Little
Hoover Commission; 1998.
35. American Psychiatric Association. Psychiatric Services in Jails. 2nd edition. Washington DC: American
Psychiatric Association; 2000.
36. Ill-equipped: us prisons and offenders with mental
illness. Available at: http://www.hrw.org/reports/
2003/usa1003/index.htm. Accessed on January 21,
2005.

18. Michaels D, Zoloth SR, Alcabes P, Braslow CA,
Safyer S. Homelessness and indicators of mental illness
among inmates in New York City’s correctional system.
Hosp Community Psychiatry. 1992;43:150–155.

37. Lamb HR, Weinberger LE, Gross BH. Mentally ill
persons in the criminal justice system: some perspectives. Psychiatr Q. 2004;75:107–126.

19. Metraux S, Culhane DP. Homeless shelter use and
reincarceration following prison release: assessing the
risk. Criminol Public Policy. 2004;3:201–222.

38. Mumola C. Substance Abuse and Treatment, State
and Federal Prisoners 1997. Washington, DC: Bureau of
Justice Statistics; 1999. Report No. NCJ 172871.

20. Cheung RC, Hanson AK, Maganti K, Keeffe EB,
Matsui SM. Viral hepatitis and other infectious diseases
in a homeless population. J Clin Gastroenterol. 2002;
34:476–480.

39. Byrne C, Faley J, Flaim L. Drug Treatment in the
Criminal Justice System. Washington DC: Executive Office of the President, Office of National Drug Control
Policy; 1998. Report No. NCJ 1700012.

21. Hammett TM, Gaiter JL, Crawford C. Reaching seriously at-risk populations: health interventions in criminal justice settings. Health Educ Behav. 1998;25:
99–120.

40. Iguchi MY, London JA, Forge NG, Hickman L,
Fain T, Riehman K. Elements of well-being affected by
criminalizing the drug user. Public Health Rep. 2002;
117 (suppl 1):S146–S150.

22. Kushel MB, Perry S, Bangsberg D, Clark R, Moss
AR. Emergency department use among the homeless
and marginally housed: results from a communitybased study. Am J Public Health. 2002;92:778–784.

41. Torres RA, Mani S, Altholz J, Brickner PW.
Human immunodeficiency virus infection among
homeless men in a New York City shelter. Association
with mycobacterium tuberculosis infection. Arch Intern
Med. 1990;150:2030–2036.

23. Zolopa AR, Hahn JA, Gorter R, et al. HIV and tuberculosis infection in San Francisco’s homeless adults.
Prevalence and risk factors in a representative sample.
JAMA. 1994;272:455–461.

1752 | Research and Practice | Peer Reviewed | Kushel et al.

42. Glaser JB, Greifinger RB. Correctional health care:
a public health opportunity. Ann Intern Med. 1993;118:
139–145.

American Journal of Public Health | October 2005, Vol 95, No. 10