Journal of Community Health

, Volume 38, Issue 4, pp 685–689

The Reality of Homeless Mobility and Implications for Improving Care

Authors

    • Department of MedicineUniversity of South Carolina School of Medicine
  • Shana Dykema
    • Department of MedicineUniversity of South Carolina School of Medicine
Original Paper

DOI: 10.1007/s10900-013-9664-2

Cite this article as:
Parker, R.D. & Dykema, S. J Community Health (2013) 38: 685. doi:10.1007/s10900-013-9664-2

Abstract

Homeless persons are perceived as a highly mobile population, and have high rates of co-morbid conditions, including mental health and substance use issues. This study sought to determine the characteristics of the mobility and reported health conditions of homeless persons. The sample for this cross sectional study (n = 674) accounted for 88 % of the homeless population in a medium sized southern city in the United States. Participants were recruited from a homeless shelter operating during the winter season. Homeless persons were less mobile than the general state population (46.11 % were born in-state vs. 40.7 % of the general population) and less transient than the general state population (78 % reported an in-state zip code for the last permanent residence). 31.9 % reported a disabling condition of a serious and long term nature. These findings challenge the concept that homeless persons are primarily a mobile population. Furthermore, homeless persons in this sample were more likely to remain in the state where they lived after becoming homeless. Thus, provider perceptions that homeless persons would not benefit from referral to a regular source of outpatient care may be misinformed. As homeless persons often seek care in emergency departments for conditions that could be addressed through outpatient care, if a medical care system implemented standard practices specifically for homeless patients, this could decrease recidivism. Such interventions represent significant opportunities to reduce costs, conserve resources, and improve care through policy modification that ensures a focus on a successful, active linkage to outpatient care and programs specific to the homeless population.

Keywords

Homeless personsEmergency medicineHealth planningHealth services research

Introduction

Homelessness remains a significant social challenge in the United States. According to data from the Annual Homeless Assessment Report (AHAR) compiled by the Housing and Urban Development (HUD), 1,502,196 persons experienced homelessness in the United States in 2011 [25]. This means that homelessness is more prevalent (0.48 %) in the United States than any of leading causes of death, including heart disease, cancer, and stroke [24, 9]. Unlike these illnesses, there is no national health-based campaign against homelessness, despite its intricate ties to public health.

Homelessness is not commonly identified as a health problem, yet it is both an etiologic factor and outcome of multiple health issues, directly and indirectly [5, 7, 26]. Homeless persons are more likely to have comorbid conditions, poorer health outcomes, and decreased access to health care than other population subgroups [2, 6, 8, 19, 20]. Among homeless persons who were sheltered during the previous year (n = 1,593,794), nearly 43 % reported a disabling health condition [25]. As many homeless persons are uninsured or underinsured, any problems accessing care are further exacerbated by the fact that a relatively small number of health care systems in the United States are designed to provide consistent care for these persons. A healthcare system’s efficacy tends to decrease when attempting to provide support to large numbers of uninsured persons who also lack stable housing, especially when those patients have active mental illness issues, addiction issues, or chronic health conditions [7]. Because of many of these challenges, homeless persons are much more likely to present to the emergency department (ED) for their medical care, which can worsen ED overcrowding [2, 6, 10, 13, 15, 18, 19, 22].

Not only does homelessness have a major impact on healthcare systems, but the reverse is also true: healthcare systems can have negative impacts on homeless persons [7]. The cumulative effects of errors and delays in medical care can have a negative compounding effect on homeless persons, who may experience barriers to accessing medical care or have difficulty navigating complex systems. In this light, medical care systems, both inpatient and outpatient, present challenges for homeless persons. Active and executed referrals to outpatient care could greatly reduce emergency department (ED) use; however, the lack of outpatient care systems designed to effectively engage homeless persons contributes to the documented ED overuse and lack of continuity of medical care for homeless persons [14].

The relocation and mobility of homeless persons is not well studied, with relatively few published articles in the last decade. Older research has indicated that the homeless population is highly mobile, which could lead to problems with access to care and continuity of care [1, 3, 17]. The few more recent articles have indicated that the idea of homeless mobility may be attributable to anecdotal evidence which has created and perpetuated the stereotype of the “homeless transient” [12, 22].

Empirical research on this subject is needed to inform health care systems. This project seeks to increase the understanding in this area and to describe the mobility characteristics of a homeless population so that providers may use this knowledge to improve health care delivery to them. Since it has been shown that the homeless are more likely to use the ED for care, this research may inform policies leading to more active linkage of homeless persons to outpatient primary care, as well as increasing continuity of care over a homeless patient’s course of treatment.

Methods

This study was approved by the University of South Carolina’s Institutional Review Board (IRB). This cross sectional study recruited a convenience sample of homeless persons from a homeless registry retained from the city’s largest homeless shelter. Deidentified data from all persons who stayed in the shelter between November 1, 2009 and March 31, 2010 were included. Data were extracted from the Service Point Homeless Management Information System (HMIS) system and STATA 10 IC was used for analyses. We defined “mobility” as relocation from the state of birth, and “transience” as relocation post-homelessness.

Sociodemographic data included sex, age, race, ethnicity, veteran status, income, income source, non-cash income, employment status, highest level of education, domestic violence status, total monthly income, disability status, years of residence in city/state, and insurance status. Homeless information included prior housing situation, length of stay, current housing status, extent of homelessness (frequency and duration), chronic homeless status, primary and secondary reasons for homelessness, and zip code of last permanent residence. Chronic homelessness was defined by Housing and Urban Development (HUD) as four or more occurrences of homelessness in the last 3 years or 1 year or more of homelessness among an unaccompanied adult with a disabling condition. Since participants and study staff may not have understood this definition, this variable was operationalized during data collection.

For univariate analyses, Chi square tests were used to analyze differences among categorical variables and t-tests were used for numeric data. If the cell sizes were small, then the nonparametric equivalent was used to increase statistical reliability. Logistic regression was conducted in multivariable analyses with −2 log likelihood ratio tests to compare models ensuring adherence to the rule of parsimony.

Results

Demographics are presented in Table 1. Participants (n = 674) were primarily (79.53 %) male. 112 persons (16.62 %) reported military service. A majority (62.76 %) identified their primary race as Black American. Less than half (36.80 %) were high school graduates or equivalent, and nearly one-third (31.90 %) reported a disabling condition. The median age was 45.37 years. Criminal domestic survivorship was reported among almost one-third of women (n = 40, 31.25 %), versus 3.72 % (n = 20) of men.
Table 1

Basic demographics (n = 674)

Measure

N

Percent

Median age: 45.37 years

Sex

 Male

536

79.79

 Female

132

19.61

 Missing/omitted

6

0.60

Veteran status

 Non-veteran

542

80.42

 Veteran

112

16.62

 Missing/omitted

20

2.96

Primary race

 Black American

426

62.76

 White American

213

31.60

 Other

38

5.64

Hispanic/Latino ethnicity

 No

633

93.92

 Yes

18

2.67

 Missing/omitted

23

3.41

Survivor of criminal domestic violence

 No

593

87.98

 Yes

60

8.90

 Missing/omitted

21

3.12

Report of a disability

 No

449

66.62

 Yes

215

31.90

 Missing/omitted

10

1.48

Born in-state

 No

366

54.30

 Yes

308

45.70

Homeless markers are reported in Table 2. Over one-third (n = 260, 38.58 %) reported “first time homeless” and one quarter (n = 171, 25.37 %) reported “one or two episodes of homelessness.” Chronic homelessness was reported by 28.93 % of participants. Almost half of participants (n = 308, 45.70 %) were born in-state, while the general adult population reports 40.7 % of persons were born in-state [16]. The majority of persons surveyed (n = 519, 77.02 %) also lived in South Carolina when they became homeless. The most commonly reported prior living situations were outdoors (n = 141, 20.92 %), a shelter (n = 129, 19.14 %), staying with family (n = 96, 14.39 %) and staying with friends (n = 76, 11.38 %).
Table 2

Overview of homelessness (n = 674)

Measure

N

Percent

Zip code of residence at homeless onset

 In-state

519

77.00

 Out-of-state

149

22.11

 Missing/omitted

6

0.89

Chronically homelessa

 No

475

70.47

 Yes

195

28.93

 Missing/omitted

4

0.59

Extent of homelessness

 First time

260

38.58

 1–3 times in past

173

25.67

 4 times or more in past 3 years

97

14.39

 1 year or more

123

18.25

 Missing/omitted

21

3.12

Prior living situation

 Outdoors

141

20.92

 Emergency shelter

129

19.14

 Staying with family/friends

172

33.53

 Renting apartment

47

6.97

 Owned their own home

40

4.94

 Jail/prison

29

4.30

 Otherb

116

17.21

Primary reason for homelessness

 Underemployed or low income

199

29.53

 Loss of job

93

13.80

 No affordable housing

61

9.13

 Medical condition

35

5.24

 Otherc

286

42.43

a“Chronically homeless”: unaccompanied individual with a disabling condition who has been continuously homeless for a year or more OR has had at least four episodes of homelessness in the past 3 years (HUD definition)

b“Other” includes: car, care home, doubled up, foster care, hospital, hotel/motel, place not for habitation, psychiatric hospital, refused, substandard structure, subsidized housing, substance abuse treatment facility, transitional housing, and missing

c“Other” includes: criminal activity, domestic violence, health/safety, loss of child care, loss of public assistance, loss of transportation, mental health issues, foreclosure, substandard housing, substance abuse, released from institution, eviction, and missing

Common reasons for homelessness included un/underemployed/low income (n = 199, 29.53 %) and loss of job (n = 93, 13.80 %) along with lack of affordable housing (9.13 %). A chronic medical condition was reported among 5.24 % (n = 35) of participants. Men were more likely to report military service (χ2(1) = 18.94, p < 0.01) and in state birth (χ2(1) = 6.74, p = 0.03) than women.

Logistic regressions were modeled for in state birth as well as further comparisons between in-state born participants and out-of-state born participants. The odds of being born in South Carolina for men were 1.68 times that of the odds for in-state birth among women [95 % CI (1.13, 2.49)]. The logistic regression used to explore differences between participants born in state and out of state included the covariates: chronic homelessness, gender, and primary race. The findings are provided in Table 3. While controlling for the presence of each variable, statistically significant findings indicated that persons born in state were 1.52 times more likely to be chronically homeless [95 % CI (1.06, 2.18)], male [OR = 1.75, 95 % CI (1.14, 2.66)], and Black American [2.87, 95 % CI (2.01, 4.09)].
Table 3

Final reduced logistic regression comparing in-state birth to out-of-state birth

Variable

Odds ratio

Standard error

p value

95 % CI

Chronically homelessa

1.52

0.28

0.02

(1.06, 2.18)

Male sexb

1.75

0.38

0.01

(1.14, 2.66)

Black American racec

2.87

0.51

0.00

(2.01, 4.09)

aReference group: non-chronically homeless

bReference group: female sex

cReference group: non-Black American race

Discussion

This study found that homeless persons were actually less mobile and less transient than the general state population, with 45.70 % of the homeless born in-state and 78 % reporting their last permanent residence before becoming homeless as in-state. These findings challenge the popular stereotype of a highly mobile homeless population. These findings may help dispel the notion among health care providers that as a result of their mobility and transience, homeless persons are unlikely to follow up on their medical care or outside referrals.

Homeless persons who were born in-state were more likely to be chronically homeless, male, and Black American. The chronically homeless have been found to have fewer financial resources, poorer physical and mental health outcomes, and less family support [2]. Their lower levels of social support and socioeconomic status may increase retention in the state of birth, as these persons may lack the necessary resources to relocate. Additionally, the lack of financial stability may affect their choice of where to seek medical care, as the chronically homeless are more likely to utilize the ED [6, 18].

If the perception among clinicians, especially in the ED, is that homeless persons would not benefit from referral to a regular outpatient source of primary care, one intervention to combat ED overuse by the homeless is provider education. Standard practice in many EDs is to advise the patient to return in case the symptoms that brought them to the ED persist [4, 11, 21]. However, if patients are presenting for non-emergent care issues and are encouraged to return because providers do not believe they will follow up with their care elsewhere, this could create an unending dependence on ED use and exacerbate overcrowding issues. Thus, an effective intervention for health care delivery systems could be an intentional effort to actively refer these patients to outpatient care providers and retain those patients there once referred, ensuring continuity of care. Active referral to non-ED sources of care could also result in significant cost savings, both for the organization and the health care system as a whole [23, 18]. In the US, the Federally Qualified Health Center (FQHC) System is comprised of publicly funded centers which see patients on an income based sliding fee scale. In many communities, these clinics also receive funding to provide health care for homeless individuals, and are therefore primed to be involved in such interventions, as well as future research on health care provision for the homeless.

Considering the individual barriers to care among the homeless such as cost, transportation, and social stigma, any intervention aimed at increasing homeless patients’ involvement in medical care while funneling them to more appropriate sources of said care must be patient-centered and provide real time referrals. Since nurses are the primary providers responsible for discharge planning in inpatient and outpatient settings, an intervention should be designed to also be clinician focused. Any such intervention to increase outpatient primary care for the homeless would require a significant emphasis on and commitment to communication, integration and sharing of resources and responsibilities.

There are limitations to this study based on the study design, including the convenience sampling method. The cross-sectional methodology means that we were unable to establish causation. Convenience sampling increases the potential for bias versus random sampling; however, the sample to population percentage of this project (88 %) should mitigate bias in the population within the city. Additionally, given the high percentage of sample to population, generalizability to other homeless populations in similar cities may be valid, though any extrapolation should be done with caution. Another limitation was the ability of the multivariable logistic regression model to fit the data. While the associations were strong, these data only account for 5 % of the variability in the data to explain whether or not a person is born in state. This indicates that there are other influencing factors not explored in this project which would more strongly account for the reasons that a homeless person remains in his/her state of origin.

Future research should further evaluate concepts of active engagement and direct intervention by shifting treatment for non-acute and chronic care to outpatient care providers. Research could include a prospective cohort of homeless persons measured on multiple markers to include health, service access, mobility and other key factors that could improve care.

Acknowledgments

Research was funded by the City of Columbia, SC.

Conflict of interest

The authors report no conflict of interest.

Copyright information

© Springer Science+Business Media New York 2013