Barriers to Healthcare of Homeless Residents of Three Honolulu Shelters

Abstract

In Honolulu, health insurance rates amongst the homeless are one of the highest in the nation, yet significant health care needs are still unmet. In a previous model, health care barriers have been divided into four domains: bureaucratic, personal, programmatic, and financial. This study aimed to determine the risk factors associated with the domains of health care barriers amongst the study's sample of 128 subjects across three Honolulu homeless shelters. Univariate models revealed health care barriers; but only the lack of health insurance was a significant financial barrier to health care in multivariate analyses (Odds ratio: 2.12; 95% Confidence Interval: 1.09–4.16). The identification of barriers should guide how health care programs approach Honolulu's homeless population to better serve their health care needs.

Introduction

In 2005, there were approximately 3,498 homeless residents in the City and County of Honolulu (44% of whom resided in a shelter).¹ At the state-level, Hawai‘i is tied for 4th in the nation with the most homeless residents per capita. Despite a high per-capita of homelessness in Hawai‘i, a previous study reported that 77% of homeless adults in Honolulu have some form of health insurance, compared to 45% in the continental United States.^1,2

A Honolulu study of the homeless reported that common health problems requiring hospitalization, among the local homeless population, are decompensated psychiatric illness, trauma, substance abuse, and infections. In addition, the study has showed that the homeless are five times more likely to be admitted to acute care hospitals compared to the general public; and 100 times more likely to be admitted to the state psychiatric hospital. The authors concluded that despite the increased likelihood of hospital admissions, the state of being homeless was probably contributory rather than causal.³

Health disparities in the homeless exist due to a perpetual, correlative triad. First, health problems cause homelessness. Second, homelessness causes health problems. Third, homelessness complicates efforts to treat health problems.⁴ We are interested in understanding the risk factors that may exist in the barriers that cause each, and working towards ending the perpetuity of the downward health spiral.

Previously, barriers to health care among American homeless were divided into four separate domains: bureaucratic, personal, programmatic, and financial.⁵

Examples of bureaucratic barriers in health care can come in the form of paperwork that is complicated, extensive, or difficult to understand due to language barriers or illiteracy. Other examples of bureaucratic barriers include long waits for insurance programs, inflexible scheduling of doctor appointments, restricted clinic hours, and the lack of transportation to and from doctor visits.

Examples of personal barriers include the lack of perceived importance or priority of health care. The homeless population are often preoccupied with survival and obtaining the basic needs (eg, food and shelter), and may not put health care at as high a priority as they should.

Examples of programmatic barriers in health care for the homeless come in the form of discontinuous/fragmented health care and negative attitudes, skepticism, or mistrust of the health care system.

Examples of financial barriers include unaffordable health care, unaffordable health insurance premiums, restrictive eligibility for insurance programs, and restrictive services in health care secondary to the lack of affordability.

Studies examining the barriers to healthcare of Honolulu's homeless population are limited. The purpose of this study is to discover the risk-factors associated with the theorized health-care barriers that may exist in Honolulu's homeless population.

Methods

In September 2009, a cross-sectional survey was developed and conducted among homeless subjects of three Honolulu homeless shelters serviced by the University of Hawai‘i John A. Burns School of Medicine. These three shelters include the Next Step, Paiolu Kaiaulu, and Onelau‘ena shelters.

Next Step Shelter is an emergency transitional night-shelter located in central Honolulu which accommodates up to 300 residents. Paiolu Kaiaulu is an emergency transitional shelter located in Waianae, Hawai‘i, that also holds up to 300 residents. Onelau‘ena Shelter is an emergency transitional shelter located in Kalaeloa, Hawai‘i, that houses up to 200 residents. All shelters receive funding through the State of Hawai‘i Department of Human Services, but are run by private organizations and charities.

This study was approved by the University of Hawai‘i Committee on Human Subjects. Three study administrators from the University of Hawai‘i John A. Burns School of Medicine and the Myron B. Thompson School of Social Work administered the surveys. The surveys were anonymous and voluntary. Verbal consent was obtained prior to participation, and we confirmed that each subject had not already completed a survey. Hygiene kits were provided as an incentive for survey completion. Surveys were available in both English and Chuukese, the two most common languages of the homeless population at these shelters.

The survey was developed de novo. It was comprised of self-reported demographic questions and 29 statements by which subjects self-reported on a Likert scale. Subjects could either strongly agree, somewhat agree, neither agree nor disagree, somewhat disagree, strongly disagree, or abstain from answering each survey statement. Subjects answered the surveys on their own, without guidance or surveillance by the study administrators. Each statement was concordant with one of the four domains of health care barriers: bureaucratic, personal, programmatic, and financial.

Data were compiled and subsequently analyzed by Statistical Analysis Software (Cary, North Carolina). Univariate and multivariate logistic regression models were used to determine the risk factors associated with health care barriers in four domains.

Results

A convenience sample of 128 total adult subjects participated in the survey. Demographic data of the sample are listed in Table 1.

Table 1.

Description of our Study Participants by Demographic Data (N=128)

	n	%
Shelter
Paiolu Kaiaulu	38	30
Onelau‘ena	47	37
Next Step Shelter	43	34
Age Group
18–30	40	35
31–50	56	50
51+	17	15
Health Insurance Status
Yes	89	70
No	38	30
Gender
Female	65	62
Male	40	38
Ethnicity/Race
Pacific Islander	66	53
Asian	2	2
Black	2	2
Caucasian	20	16
Filipino	2	2
Hispanic	4	3
Mix	28	23

Univariate analysis of individual risk factors was performed to determine their strength of association with health care barriers. Risk factors that achieved significance in the 95% confidence interval were carried into a multivariate logistic regression analysis to control for potential confounding variables. Univariate analysis of the risk factors associated with each barrier is included in table 2.

Table 2.

Univariate Analysis of Survey Data Amongst the Barriers to Healthcare (N=128)

	Bureaucratic		Personal		Programmatic		Financial
Risk Factor	ORa	95% CIb	ORa	95% CIb	ORa	95% CIb	ORa	95% CIb
Shelter
Paiolu Kaiaulu	1.00		1.00		1.00		1.00
Onelau‘ena	1.78	0.84–3.77	2.39c	1.12–5.07	1.60	0.76–3.36	1.02	0.48–2.15
Next Step Shelter	0.69	0.32–1.48	1.06	0.50–2.26	0.69	0.32–1.47	1.43	0.67–3.06
Age Group
18–30	1.00		1.00		1.00		1.00
31–50	0.51	0.25–1.05	0.99	0.49–2.01	0.83	0.41–1.67	0.70	0.34–1.42
51+	0.47	0.18–1.28	0.54	0.20–1.45	1.23	0.47–3.36	1.06	0.39–2.84
Health Insurance Status
Yes	1.00		1.00		1.00		1.00
No	0.92	0.47–1.77	1.23	0.64–2.39	1.71	0.88–3.32	2.12c	1.09–4.16
Gender
Male	1.00		1.00		1.00		1.00
Female	2.18c	1.09–4.37	2.08c	1.04–4.16	1.79	0.90–3.55	1.38	0.69–2.74
Ethnicity/Race
Pacific Islander	1.00		1.00		1.00		1.00
Asian	0.14	0.01–1.66	0.31	0.03–3.57	0.80	0.07–9.12	3.07	0.26–36.03
Black	0.04c	0.01–0.58	0.79	0.01–1.06	0.29	0.03–3.35	<0.01	0 - ∞
Caucasian	0.42	0.17–1.01	0.42	0.17–1.01	0.45	0.19–1.09	0.44	0.18–1.08
Filipino	2.43	0.21–28.09	2.45	0.21–28.29	2.52	0.22–29.06	0.68	0.06–7.89
Hispanic	0.72	1.20–4.14	0.31	0.05–1.79	0.64	0.11–3.66	0.46	0.08–2.71
Mix	0.74	0.34–1.60	0.65	0.31–1.41	0.61	0.28–1.32	0.52	0.24–1.13

^aOR = Odds Ratio;

^bCI = Confidence Interval;

^csignificant to the 95% confidence interval

In the univariate analysis for bureaucratic barriers, female gender (Odds Ratio=2.18; 95% Confidence Interval: 1.09–4.37) and black ethnicity/race (0.05; 0.01–0.58) were significant variables. In the multivariate analysis (table 3), only black ethnicity/race showed statistical significance (0.05; 0.01–0.67); however, with only 2 black subjects in the sample size of 128, it would be unreasonable to draw significant conclusions from this result.

Table 3.

Multivariate Analysis of Bureaucratic Barriers

Risk Factor	ORa	95% CIb
Gender
Male	1.00
Female	0.51	0.95–4.11
Ethnicity/Race
Pacific Islander	1.00
Asian	0.16	0.01–1.90
Black	0.05c	0.01–0.67
Caucasian	0.62	0.24–1.62
Filipino	3.45	0.29–40.89
Hispanic	1.04	0.17–6.33
Mix	1.05	0.45–2.43

^aOR = Odds Ratio;

^bCI = Confidence Interval;

^csignificant to the 95% confidence interval

In the univariate analysis for personal barriers, subjects at the Onelau‘ena shelter (2.39; CI: 1.12–5.07) and female gender (2.08; 1.04–4.16) were significant variables. These risk factors were not significant in the multivariate analysis (table 4).

Table 4.

Multivariate Analysis of Personal Barriers

Risk Factor	ORa	95% CIb
Gender
Male	1.00
Female	1.96	0.96–4.01
Shelter
Paiolu Kaiaulu	1.00
Onelau‘ena	1.39	0.61–3.17
Next Step Shelter	0.95	0.41–2.17

^aOR = Odds Ratio;

^bCI = Confidence Interval;

^csignificant to the 95% confidence interval

There were no significant risk factors associated with programmatic barriers in the univariate analysis, and thus, no multivariate analysis was performed.

Subjects without health insurance (2.12; 1.09–4.16) were the only statistically significant risk factor in the univariate analysis for personal barriers, and thus, no multivariate analysis was performed.

Discussion

This study revealed specific barriers to health care in our homeless population. As we expected, financial barriers exist amongst homeless subjects without health insurance. Numerous studies have concluded the same.^6–9 Not surprisingly, the cost of health care becomes an issue, especially when health insurance is not available to absorb some of the expenses. However, subjects without insurance appear to be unaffected by bureaucratic, programmatic, and personal barriers to health care; which may indicate that these barriers associated with health insurance enrollment were not a factor in accessing health care. Conversely, it could mean that the homeless subjects were not exposed to the barriers that may exist when being insured. Nonetheless, this study revealed that only financial barriers were perceived, given the lack of health insurance.

Limitations exist in this study. Length of shelter residency was not asked in the survey. A study has shown that “shelterization,” or the undesirable task of independent living, is especially high among people with increased dependency on others.¹⁰ We predict that a longer length of shelter residency may predict complacency, which would translate into a favorable opinion of healthcare.

“Pacific Islander” could have been too broad as a racial category in this study. Study administrators noted that the cultures of Micronesia and native Hawaiian people are different, and thus could have hidden underlying risk factors in either racial group.

Another limitation is that we only examined health care barriers in sheltered homeless subjects. We did not examine unsheltered homeless adults. We predict that results of this study underestimate the barriers for unsheltered homeless adults.

Despite limitations, strengths do exist in the study. Our multiethnic homeless population has seldom been studied. The Pacific Islander population has been notoriously underserved, and studies on Pacific Islanders have been scant. Another strength is that we examined a homeless population where 70% of subjects had health insurance, compared to 45% in the continental United States.¹ Surveying subjects with health insurance gave us an opportunity to discover what barriers exist among the insured; and can allow us to secondarily look at why, despite the abundance of local affordable health insurance programs, homeless subjects still go without health insurance.

The results of this study reveal a prominent deficiency in accessing health care for the homeless, ie, health insurance. Despite the wide availability of government-assisted health insurance plans in Hawai‘i, a subset of homeless individuals still go without insurance. It would be important for future studies to focus on what is preventing this subset of individuals to live without health insurance. Additionally, we must ascertain their cognizance of the financial benefits associated with having health insurance. Furthermore, we must discover the barriers to health insurance now that we have revealed the barriers to health care among Honolulu's homeless population. Once better elucidated, specific recommendations for the rectification of financial barriers associated with the lack of health insurance can be made.

Future studies should also examine health care barriers throughout all socio-economic groups. Such analysis would allow a comparison between lower, middle, and upper class subjects to reveal what health care barriers are truly unique to the homeless. In addition, children should be included in future studies to determine health care barriers in the pediatric population.