The effect of the global financial crisis on preventable hospitalizations among the homeless in New York State

Abstract

Objective

Periods of economic instability may increase preventable hospitalizations because of increased barriers to accessing primary care. For underserved populations such as the homeless, these barriers may be more pronounced due to limited resources in the health care safety net. This study examined the impact of the global financial crisis of 2007–2008 on access to care for the homeless in New York State.

Methods

Hospitalizations for ambulatory care sensitive conditions (ACSCs) were used as a proxy measure for primary care access. Admissions for ACSCs were identified in the New York State Inpatient Database from 2006 to 2012. Hospitalization rates for ACSCs were calculated for the homeless and nonhomeless. Multivariable linear regression was used to investigate the impact of the financial crisis on hospitalization rates for ACSCs.

Results

The findings indicate that during the financial crisis, homeless adults had significantly higher preventable hospitalizations than nonhomeless adults, and the uninsured homeless had significantly higher preventable hospitalizations when compared to other homeless subgroups. After the financial crisis, preventable hospitalizations for the homeless stabilized but remained at higher rates than those for the nonhomeless.

Conclusions

These findings are important to developing health policies designed to provide effective care for underserved population such as the homeless.

Keywords

access to care homeless insurance status primary care

Introduction

The global financial crisis of 2007–2008 severely affected the United States (US). It resulted in more than seven million jobs lost by mid-2009, and the unemployment rate doubled to a rate of nearly 10%.¹ The housing market collapsed and trillions of dollars were lost on the stock market. The crisis left many families in financial turmoil.

With the increase in unemployment, the proportion of adults with employer-sponsored health insurance fell from 65.2% in 2004 to 60.4% in 2009,² and the proportion of those without insurance rose from 16.9% to 18.9% during the same period.² Out-of-pocket health care costs increased as more Americans did not have sufficient coverage,³ which may have resulted in more individuals delaying or foregoing health care to meet more basic needs.⁴ For vulnerable populations, this may have been especially salient as generally they face additional barriers to accessing care.^5,6

The volatility of the US economy had a significant influence on the health care system, including the utilization and the provision of health care services. For vulnerable populations such as the homeless and persons who are marginally housed, this instability was especially challenging as poor Americans are more likely to have difficulties accessing health care services compared to those from high-income households.⁷ Estimates of homeless, defined as using an emergency shelter or a transitional housing program, were 1,588,595 in 2007 and 1,558,917 in 2009 (the six-month homeless estimate was approximately 1,150,000 in 2006; standardized data collection for one-year homeless estimates did not begin until 2007).⁸ Homelessness can result from a number of interrelated factors such as unemployment, debt, drug or alcohol abuse, domestic violence, mental health problems, institutionalization and limited social support.⁹ Most homeless individuals lack sustained employment, health insurance, a usual source of health care, and they experience substantial barriers to accessing quality health care.¹⁰ Lack of access to primary care for homeless and other underserved populations has been associated with higher levels of costly and preventable hospital admissions.^11–13

A small number of studies have investigated the crisis’ impact on other vulnerable populations, such as recent immigrants and Medicaid beneficiaries. One study found that recent uninsured immigrants (non-US citizens <5 years) were less likely to engage in health care spending during the financial crisis compared to US citizens and non-US citizens ≥5 years using 2005–2006 and 2008–2009 Medical Expenditure Panel data.⁴ One other study found that from 2008 to 2010, Medicaid beneficiaries were significantly more likely to be admitted through the emergency department compared to other payer groups (such as the uninsured, Medicare and private insurance).¹⁴ There remains a dearth of evidence on the impact of the crisis on the homeless compared to nonhomeless population groups.

This study aims to address this gap. It examined the impact of the global financial crisis from 2007 to 2008 on homeless people using preventable hospitalizations as an indicator to understand the added burden that may have been caused through reduced access to primary care services in the state of New York. We focused on the state of New York because it had significantly higher rates of homelessness compared to other US states during the financial crisis.^8,15 Preventable hospitalizations are defined as admissions for acute or chronic conditions that are considered potentially avoidable if managed appropriately in primary care.¹⁶ High rates of preventable hospitalizations for any population group can be seen as indicators of poor access to and poor quality primary care, and they are of concern to health care providers and policy makers because of their significant impacts on health care costs.^17,18

Methods

Data were obtained from the Agency for Healthcare Research and Quality Healthcare Cost and Utilization Project State Inpatient Database (HCUP-SID).¹⁹ The NY HCUP-SID database includes de-identified longitudinal discharge data from all acute care licensed hospitals, as well as rehabilitation, psychiatric and substance abuse facilities in New York State. It also provides de-identified annual inpatient discharges with clinical and patient demographic data (e.g. homeless status, insurance status). The homeless variable in the HCUP-SID was dichotomous and derived from reporting by hospital staff. This study covered the period 2006 to 2012.

The primary outcome measure was preventable hospitalizations using ambulatory care sensitive conditions (ACSCs), defined as ‘conditions for which good outpatient care can potentially prevent the need for hospitalization, or for which early intervention can prevent complications or more severe disease’.²⁰ ACSCs used in this study were angina without procedure; asthma; bacterial pneumonia; cellulitis; chronic obstructive pulmonary disease; congestive heart failure; general symptoms of convulsions, grand mal status and other epileptic convulsions; dehydration/volume depletion; dental conditions; diabetes mellitus; gastroenteritis; hypertension; hypoglycemia; immunization-related and preventable conditions (e.g. pertussis, tetanus, acute poliomyelitis and rheumatic fever); kidney/urinary tract infection; severe ear, nose and throat infections; and skin grafts with cellulitis.^20,21 Clinical diagnoses of ACSCs were identified using the International Classification of Diseases, 9th Revision, Clinical Modification codes, using the primary discharge diagnosis. Patients younger than 18 years and/or with neonatal and/or maternal diagnoses and procedures were excluded from the analyses.

Quarterly ACSC hospitalization rates were calculated per 10,000 population subgroup (homeless and nonhomeless) for demographic and insurance coverage groups. Rates were adjusted for annual population differences from the US Census Bureau and the Continuum of Care Homeless Assistance Program.^8,22 The denominator for ACSC hospitalization rates was calculated from the annual point-in-time estimates of the Continuum of Care Homeless Assistance Program for sheltered and unsheltered homeless persons by state (collected annually in January each year).⁸ We used descriptive statistics for demographic variables. Multivariable linear regression models were used to examine the effect of the financial crisis on ACSC rates for demographic groups, using SAS Software (Version 9.3, Cary, NC). The key predictor variable was an interaction term that signified the ACSC hospitalization rate for the homeless during the crisis. The study was submitted to the Medical University of South Carolina (MUSC) Institutional Review Board (IRB); the IRB deemed it not to be human subject research because the HCUP data were de-identified prior to analysis.

Results

Throughout the study period (2006–2012), the percentages of ACSC admissions among the homeless were consistently higher compared to the nonhomeless (Table 1). The highest percentage of ACSC admissions among the homeless was 11.7% in 2007; among the nonhomeless, the highest percentage was 8.9% in 2009. Homeless patients were also younger. In 2009, the mean age for admitted homeless patients was 55.8 years (SD ± 18.6) and 62.5 years (SD ± 19.6) for nonhomeless patients. Minority homeless patients made up over 75% of ACSC admissions from 2006 to 2012; however, the proportion of admissions among minority nonhomeless patients was less than 50%.

Table 1.

Characteristics among the homeless and nonhomeless discharged for a primary ACSC condition, New York State Inpatient Databases 2006–2012.

	2006		2007		2008		2009		2010		2011		2012
	H	NH	H	NH	H	NH	H	NH	H	NH	H	NH	H	NH
No. ACSC Admissions (% total admissions)	10,551 (11.60%)	162,043 (8.63%)	10,206 (11.65%)	159,946 (8.65%)	9,498 (10.66%)	161,184 (8.62%)	9,881 (10.70%)	163,587 (8.63%)	9,305 (10.09%)	158,829 (8.51%)	9,185 (9.97%)	158,882 (8.62%)	8,819 (9.98%)	155,410 (8.61%)
18–64 years (adult), %	68.30	49.75	67.67	49.69	63.11	47.77	64.17	48.60	63.91	48.10	64.98	48.25	65.96	49.10
Age, mean (SD)	55.20 (±18.2)	62.41 (±19.7)	55.69 (±18.1)	62.42 (±19.7)	56.44 (±18.4)	62.73 (±19.6)	55.82 (±18.6)	62.46 (±19.6)	56.36 (±18.2)	62.80 (±19.5)	56.49 (±18.4)	63.09 (±19.5)	56.15 (±18.0)	62.66 (±19.5)
Male, %	42.56	40.48	42.52	40.77	40.61	40.13	41.38	40.26	42.40	40.33	41.61	40.36	44.39	40.78
Minority race, %	89.24	40.72	89.87	43.55	89.05	41.93	89.06	41.76	89.60	42.25	89.34	42.33	90.11	45.93

Source: Authors’ analysis of data from the NY HCUP-SID from 2006 to 2012.

H: homeless; NH: nonhomeless; SD: standard deviation.

Data not sufficient to report.

The overall mean ACSC rate during the study period was 373.0 per 10,000 homeless persons and 20.6 per 10,000 nonhomeless persons (p < 0.0001) (Table 2). The ACSC rates were highest for homeless minorities, at 333.7 per 10,000 homeless persons compared to 8.8 per 10,000 nonhomeless persons (p < 0.0001), followed by those for the homeless insured, at 33.6 compared to 19.3 for the nonhomeless (p < 0.0001). The ACSC rate for the uninsured homeless was the second lowest, at 39.1 compared to 0.9 for the nonhomeless (p < 0.0001).

Table 2.

Mean ACSC hospitalization rates for the homeless and nonhomeless in New York, overall and by subgroups.

	Homeless	Nonhomeless	p value
Overall	373.0 ± 36.49	20.55 ± 1.16	<0.0001
Adults	244.10 ± 25.05	15.65 ± 0.84	<0.0001
Seniors	128.9 ± 15.31	77.78 ± 5.97	<0.0001
Females	215.80 ± 24.59	12.24 ± 0.73	<0.0001
Males	157.20 ± 13.23	8.31 ± 0.44	<0.0001
Minority race	333.70 ± 32.90	8.76 ± 0.46	<0.0001
White race	39.04 ± 5.01	11.62 ± 0.83	<0.0001
Insured	333.60 ± 28.90	19.34 ± 1.10	<0.0001
Uninsured	39.10 ± 25.47	0.90 ± 0.13	<0.0001

Source: Authors’ analysis of data from the NY HCUP-SID from 2006 to 2012.

ACSC rates for the homeless uninsured decreased rapidly starting in quarter 4 of 2010 (Figure 1), falling from 67.4 per 10,000 homeless persons from October to December of 2010 to 26.3 from January to March of 2011. During the same time period, ACSC rates for the homeless insured increased from 294.0 to 369.6 per 10,000 homeless persons, and this rate did not decrease until 2012.

Figure 1.

Quarterly ACSC hospitalization rates for the uninsured and insured homeless per 10,000 homeless persons in New York, 2006–2012. ACSC: ambulatory care sensitive condition; NY HCUP-SID: New York Healthcare Cost and Utilization Project State Inpatient Database.

Table 3 reports a summary of the factors found to be associated with ACSC hospitalization rates. Overall, the homeless population had a significantly higher ACSC rate during the financial crisis compared to the nonhomeless rate (β = 37.72; p < 0.01) (Table 3). Higher ACSC rates during the crisis were observed for most homeless demographic groups, including the uninsured homeless (β = 27.65; p < 0.01), seniors (β = 15.61; p < 0.01), females (β = 27.94; p < 0.01), minorities (β = 30.13; p < 0.05) and homeless people of white race (β = 4.19; p < 0.05). Overall, ACSC rates decreased significantly from 2006 to 2012 for all demographic groups (homeless and nonhomeless), except for the insured. Homeless and nonhomeless minorities experienced the largest decrease (β = −1.17; p < 0.001), followed by adults (β = −1.15; p < 0.0001), females (β = −0.82; p < 0.01) and the uninsured (β = −0.17; p < 0.01). In all models, ACSC rates for the homeless were significantly higher compared to the nonhomeless rates.

Table 3.

Factors associated with ACSC hospitalization rates for population subgroups in New York State.

Independent variables	Subgroup
Independent variables	Overall (R²= 0.99)	Adults (R²= 0.98)	Seniors (R²= 0.89)	Females (R²= 0.98)	Males (R²= 0.99)	Minorities (R²= 0.99)	Whites (R²= 0.95)	Uninsured (R²= 0.66)	Insured (R²= 0.98)
Intercept	41.14‡	33.20‡	82.49‡	24.72‡	16.43‡	26.66†	14.08‡	11.74*	19.23‡
Homeless	345.04‡	225.46‡	47.76‡	197.57‡	147.47‡	318.46‡	26.53‡	32.28‡	312.80‡
Recession	−4.53	−4.32	4.11	−2.66	−1.86	−4.45	0.07	−2.57	0.54
Homeless ACSC rate during the recession	34.72**	14.12	15.61**	27.94**	6.76	30.13*	4.19*	27.65**	7.00
Time	−1.35	−1.15‡	−0.39**	−0.82**	−0.53†	−1.17†	−0.17†	−0.71**	NS

Source: Authors’ analysis of data from the NY HCUP-SID 2006–2012.

NS: not significant.

p < 0.05; **p < 0.01; †p < 0.001; ‡p < .0001; Adjusted R-square reported. Recession was defined as the period from 1 January 2008 to 30 June 2009.

Discussion

This study demonstrated that preventable hospitalizations for homeless adults in New York State were higher than those for nonhomeless adults during 2006–2012. Overall, the proportion of preventable hospitalizations relative to total admissions among the homeless declined during the observation period, whereas the proportion among the nonhomeless remained relatively constant during the same period.

It is possible that observed postfinancial crisis declines in preventable hospitalizations among the homeless were related to the implementation of a Temporary High-Risk Pool Program in 2010 for the uninsured with preexisting health conditions that provided complete primary care and outpatient substance abuse treatment in New York State.²³ Hospitals may have enrolled uninsured homeless persons into this program after the financial crisis because of the impending Patient Protection and Affordable Care Act, which would provide partial compensation for low-income uninsured patients.²³ However, while this policy may have decreased ACSC rates for the homeless uninsured after the program’s implementation in 2010 by changing the status of the homeless uninsured to insured, it may have increased ACSC hospitalization rates for the homeless insured. This implies that while the policy intervention may have increased potential access to primary care services for these newly insured homeless persons, it may not have increased realized access because of existing barriers that did not connect them with a primary care provider. Additional study of this issue is needed to confirm these hypotheses.

Our analysis further shows that mean ACSC rates for the homeless were higher than those for the nonhomeless for all demographic groups (adults, seniors, males, females, Whites, racial minorities, insured and uninsured). These findings highlight the negative consequences of being homeless, which supersede other demographic characteristics that are commonly associated with challenges to accessing care. However, similar to other studies, our work showed that ACSC rates were highest among homeless adults from racial-ethnic minority groups.^24,25 Also, we found that the homeless uninsured and other demographic groups had significantly higher ACSC rates during the global financial crisis, pointing to the added impact of the crisis on a population that is already at risk of difficulties in accessing high-quality primary care.

Homeless adults are more likely to receive inadequate preventative care, experience illness, and have higher hospitalization rates as well as die at a younger age.²⁶ Additionally, many homeless adults experience complex medical and psychiatric problems that are oftentimes worsened by drug and alcohol abuse.²⁶ These complex challenges call for more integrated and multidisciplinary services that cater to the medical and social needs of the homeless.²⁶ Health policies must thus consider strategies to better address the multiple needs of this vulnerable population group with particular awareness of how these needs are magnified by the economic stability of the US economy.

The primary cause of unmet needs is lack of health insurance, which provides access to preventative primary care,²⁷ although it is important to note that many homeless adults frequently experience barriers to obtaining care even in systems that provide universal health care access.²⁵ The future of the Affordable Care Act is uncertain and strategies aimed at improving access to quality health care for the homeless population in the US will be increasingly challenging and costly in the absence of clear funding mechanisms to support such programs.

Study limitations

There are several limitations to this study. First, we used point-in-time estimates of the homeless population. This approach does not represent all persons that may have been homeless throughout the study period. Although the Continuum of Care Homeless Assistance Program provides a baseline estimate for the homeless, point-in-time estimates may not accurately represent the number of actual homeless persons. At the same time, it is the data source with the greatest consistency in data collection over time even though its cross-sectional data collection approach may not accurately represent the number of homeless persons over the time period of a year. The size of the error associated with the homeless estimates remains unclear; it could have overestimated or underestimated the ACSC hospitalization rates for the homeless although the direction error would have been the same for all homeless groups considered here. Second, we used the ACSC hospitalization rate as a proxy measure for primary care access; this does not provide an actual measure of access to primary care services. Third, the study relied on administrative data (NY HCUP-SID), and we cannot exclude the possibility that there may have been inaccuracies in the documentation of key demographic and clinical variables, such as coding errors in identifying homeless patients. It is possible that homelessness reporting may be increased or decreased based on patient motivation to receive services or patient refusal to acknowledge homelessness because of the potential stigmatization associated with reporting. There is a need for a consistent protocol in how hospital staff document homeless persons to ensure that this population is appropriately identified. Fourth, there were potential confounding factors that were impossible to measure in the NY HCUP-SID data sets. This included the inability to measure the length and severity of homelessness, the availability of transportation and the availability of social services for populations experiencing homelessness.

Finally, the financial crisis could have changed key characteristics of homeless persons during this period, resulting in a unique set of health care needs that were not relevant before the crisis. Based on a US Congressional report, the proportion of homeless families with children under the age of 18 years increased from 29.8% in 2007 to 34.1% in 2009. Although this study excluded children under the age of 18 years, the change in the composition of the homeless population, that is, an increase in the number of families who were homeless, could have caused changes in the types of conditions for which homeless persons were hospitalized. However, the focus of this study was on changes in hospital use as a measure of health care access. Identifying and examining differences in hospitalizations for different health conditions was beyond the scope of this study.

Despite these limitations, our findings demonstrate the importance of adequately preparing health care systems to meet the complex needs of disenfranchised populations such as the homeless, particularly if they are newly insured. This involves implementing strategies to facilitate connections with primary care providers and addressing existing barriers to health care access.²⁸

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

United States Bureau of Labor Statistics. The Recession of 2007–2009, http://www.bls.gov/spotlight/2012/recession/ (2012, accessed 27 January 2017).

Holahan

. The 2007–09 recession and health insurance coverage. Health Aff (Millwood) 2011; 30: 145–152.

Martin

Lassman

Whittle

et al.

Recession contributes to slowest annual rate of increase in health spending in five decades. Health Aff (Millwood) 2011; 30: 11–22.

Vargas Bustamante

Chen

. The Great Recession and health spending among uninsured US immigrants: implications for the Affordable Care Act implementation. Health Serv Res 2014; 49: 1900–1924.

White

Newman

. Access to primary care services among the homeless: a synthesis of the literature using the equity of access to medical care framework. J Prim Care Community Health 2015; 6: 77–87.

White

Logan

Magwood

. Access to diabetes care for populations experiencing homelessness: an integrated review. Curr Diab Rep 2016; 16: 112–112.

US Agency for Healthcare Research and Quality. 2015 National Healthcare Quality and Disparities Report and 5th Anniversary Update on the National Quality Strategy, https://www.ahrq.gov/research/findings/nhqrdr/nhqdr15/index.html (2016, accessed 25 September 2017).

US Department of Housing and Urban Development. Annual Homelessness Assessment Reports to the US Congress, https://www. hudexchange.info/programs/hdx/guides/ahar/#reports (2010, accessed 12 January 2017).

Wright

Tompkins

. How can health services effectively meet the health needs of homeless people? Br J Gen Pract 2006; 56: 286–293.

10.

National Alliance to End Homelessness. The state of homelessness in America 2014, https://www.endhomelessness.org/library/entry/the-state-of-homelessness-2014 (2014, accessed 25 January 2017).

11.

National Academies of Sciences, Engineering, and Medicine. America’s uninsured crisis: consequences for health and health care, https://www.nap.edu/catalog/12511/americas-uninsured-crisis-consequences-for-health-and-health-care (2009, accessed 26 January 2017).

12.

Shi

Chen

Nie

et al.

Racial and socioeconomic disparities in access to primary care among people with chronic conditions. J Am Board Fam Med 2014; 27: 189–198.

13.

Bodenheimer

Pham

. Primary care: current problems and proposed solutions. Health Aff (Millwood) 2010; 29: 799–805.

14.

Watts

David Bryan

Tarwater

. Changes in insurance status and emergency department visits after the 2008 economic downturn. Acad Emerg Med 2015; 22: 73–80.

15.

US Bureau of Labor Statistics. Local area unemployment statistics: New York, 2004-2014, https://www.bls.gov/lau/ (2016, accessed 17 January 2017).

16.

Moy

Chang

Barrett

et al.

Potentially preventable hospitalizations – United States, 2001–2009. MMWR 2013; 62: 139–143.

17.

Fingar KR, Barrett ML, Elixhauser A, et al. Trends in potentially preventable inpatient hospital admissions and emergency department visits: statistical brief #195, https://www.hcup-us.ahrq.gov/reports/statbriefs/sb195-Potentially-Preventable-Hospitalizations.pdf (2015, accessed 17 January 2017).

18.

Solberg

. Preventable hospital admissions: are they? Fam Pract 2015; 32: 245–246.

19.

US Agency for Healthcare Research and Quality. Healthcare Cost and Utilization Project (HCUP): State Inpatient Databases, https://www.hcup-us.ahrq.gov/sidoverview.jsp (2017, accessed 9 January 2017).

20.

US Agency for Healthcare Research and Quality. Guide to prevention quality indicators: hospital admission for ambulatory care sensitive conditions, https://www.ahrq.gov/downloads/pub/ahrqqi/pqiguide.pdf (2001, accessed 9 January 2017).

21.

Ricketts

Randolph

Howard

et al.

Hospitalization rates as indicators of access to primary care. Health Place 2001; 7: 27–38.

22.

US Census Bureau. Population and Housing Unit Estimates, http://www.census.gov/programs-surveys/popest.html (2017, accessed 12 January 2017).

23.

Boozang P, Dutton M, Lam A, et al. Implementing federal health care reform: a roadmap for New York State, http://www.healthcarereform.ny.gov/research_and_resources/ docs/roadmap_for_nys.pdf (2010, accessed 27 January 2017).

24.

Kertesz

McNeil

Cash

et al.

Unmet need for medical care and safety net accessibility among Birmingham’s homeless. J Urban Health 2014; 91: 33–45.

25.

Hwang

Ueng

Chiu

et al.

Universal health insurance and health care access for homeless persons. Am J Public Health 2010; 100: 1454–1461.

26.

Maness

Khan

. Care of the homeless: an overview. Am Fam Physician 2014; 89: 634–640.

27.

Baggett

O’Connell

Singer

et al.

The unmet health care needs of homeless adults: a national study. Am J Public Health 2010; 100: 1326–1333.

28.

Tsai

Rosenheck

Culhane

et al.

Medicaid expansion: chronically homeless adults will need targeted enrollment and access to a broad range of services. Health Aff (Millwood) 2013; 32: 1552–1559.