Access to outpatient services in rural communities changes after hospital closure

Abstract

Objective

Between January 2005 and July 2020, 171 rural hospitals closed across the United States. Little is known about the extent that other providers step in to fill the potential reduction in access from a rural hospital closure. The objective of this analysis is to evaluate the trends of Federally Qualified Health Centers (FQHCs) and Rural Health Clinics (RHCs) in rural areas prior to and following hospital closure.

Data Sources/Study Setting

We used publicly available data from Centers for Medicare and Medicaid Provider of Services files, Cecil G. Sheps Center rural hospital closures list, and Small Area Income and Poverty Estimates.

Study Design

We described the trends over time in the number of hospitals, hospital closures, FQHC sites, and RHCs in rural and urban ZIP codes, 2006–2018. We used two‐way fixed effects and pooled generalized linear models with a logit link to estimate the probabilities of having any RHC and any FQHC within 10 straight‐line miles.

Data Collection/Extraction Methods

Not applicable.

Principal Findings

Compared to hospitals that never closed, the predicted probability of having any FQHC within 10 miles increased post closure by 5.95 and 11.57 percentage points at 1 year and 5 years, respectively (p < 0.05). The predicted probability of having any RHC within 10 miles was not significantly different following rural hospital closure. A percentage point increase in poverty rate was associated with a 1.98 and a 1.29 percentage point increase in probabilities of having an FQHC or RHC, respectively (p < 0.001).

Conclusions

In areas previously served by a rural hospital, there is a higher probability of new FQHC service‐delivery sites post closure. This suggests that some of the potential reductions in access to essential preventive and diagnostic services may be filled by FQHCs. However, many rural communities may have a persistent unmet need for preventive and therapeutic care.

What is known on this topic?

• The rate of rural hospital closure in the United States has increased dramatically in recent years.

• Rural hospitals are primarily outpatient facilities, and, currently, the landscape of primary care providers after a rural hospital closes is unknown.

What this study adds?

• After a rural hospital closes, we find an increased probability of having access to a Federally Qualified Health Center within 10 miles.

• Understanding the availability of health care services in rural areas is critical in understanding the long‐term impacts of rural hospital closure on the health of rural communities and for identifying unmet need.

1. BACKGROUND

Despite the fact that rural hospitals provide critically needed services to local communities, the rate of rural hospital closure in the United States has increased dramatically in recent years. ¹ While the United States experienced increases in rural hospital closures in the 1980s due to Medicare prospective payment reform, Congress' creation of Critical Access Hospitals in the late 1990s slowed the rate of closures until the 2008/2009 financial crisis. ² Yet, between January 2005 and July 2020, 171 rural hospitals closed across the country. ³ The potential impact of a local hospital closure on access to health care in a rural community is multifaceted with both short‐ and long‐term effects. Notwithstanding the serious implications for access to care, there has been little research into the extent that other providers step in to fill the potential reduction in access from a rural hospital closure.

Research shows that hospital closures reduce local access to care, especially emergency care, lead to an “outmigration of health care professionals, and worsen pre‐existing challenges around access to specialty care.” ² For example, one study reported that the supply of primary care physicians decreased 8.2% at least 6 years post closure. ⁴ Rural hospital closures have also been found to increase mean emergency medical service transport times, up to a 76% increase compared to before the closure. ⁵ , ⁶ In addition, when a rural hospital closes, patients must travel further to access acute care, an average of 12.5 miles to the next closest hospital—more than half of patients would have to drive 15‐30 miles. ⁷ Given rural communities are typically older with lower income, worse health, and greater dependence on public insurance, barriers to accessing care post closure are concerning. ⁸ Many residents of rural communities where the local hospital has closed do not receive needed tests or routine care due to lack of transportation to get the services, and some even delay or lack life‐saving therapies such as dialysis, cancer therapies, and treatment for catastrophic injuries. ² , ⁹

Federally Qualified Health Centers (FQHCs) and Rural Health Clinics (RHCs) may augment or substitute care provided in a community when the local hospital closes. ² FQHCs, also referred to as community health centers, can be located in urban or rural areas as long as the area is determined to be a Medically Underserved Area (MUA), that is an area with a shortage of primary care services as defined by the Health Resources and Services Administration (HRSA). ¹⁰ In addition to a minimum set of required primary care services, FQHCs provide behavioral health, chronic care management, and other specialty and ancillary services. ¹¹ Similarly, FQHC look‐a‐like sites operate, provide services, and bill for Medicaid and Medicare reimbursement similar to FQHCs but are not funded through the Health Center Program. ¹¹ RHCs offer essential primary care services to rural communities in MUAs and Health Professional Shortage Areas (HPSAs), which are areas with primary care, dental, and/or mental health care provider shortages as defined by HRSA. ¹² The services available at RHC can be narrower in scope than FQHCs, and they do not have the same requirements, including the types of services available, sliding scale charges, and after‐hours services. ¹³ Rural hospitals are primarily outpatient facilities, for example, the average critical access hospital derives 79.4% of patient revenue from outpatient services. ¹⁴ Rural hospitals, FQHCs, and RHCs serve similar patients ¹⁵ and case examples describe how FQHCs have increased services after a hospital closes. ² Thus, primarily outpatient care, similar patients, and anecdotal evidence suggest that FQHCs may replace at least some of the lost care after a rural hospital closes.

As rural hospitals continue to close, it is important to know whether new FQHC and RHC service sites are filling potential gaps in primary health care. The objective of this analysis is to describe the trends of FQHCs and RHCs availability in rural areas prior to and following hospital closure. Additionally, we explore the associations between poverty rates and access to FQHCs and RHCs following hospital closure.

2. METHODS

2.1. Data and sample

The sample contains short‐term, acute care, and critical access hospitals for years 2006–2018 but excludes long‐term care, psychiatric, rehabilitation, transplant, and children's hospitals. We define rural hospitals as those hospitals in a rural ZIP code and urban hospitals as those in a nonrural ZIP code. We defined rural using the Metropolitan Statistical Area code which is based upon the Office of Management and Budget designation and designates zip codes that reside mostly in nonmetro counties as rural. ¹⁶ Rural hospitals were identified in all states in the continental United States except for Rhode Island and Washington D.C., which have no eligible rural hospitals.

RHCs, FQHCs (including look‐a‐like sites), and rural hospitals were identified using the Centers for Medicare and Medicaid (CMS) provider of service files. We identified the location and year of each rural hospital closure using the publicly available Cecil G. Sheps Center list of rural hospital closures from 2006 to 2018. In one instance, two rural hospitals within 10 miles of each other closed the same year, and this was categorized as one hospital closure. We identified county‐specific poverty estimates using the Small Area Income and Poverty Estimates files for 2006–2018.

First, we calculated the straight‐line (Haversine) distance of every rural hospital (open and closed) from 2006 to 2018 to the nearest RHC and FQHC. We then calculated the percent of rural hospitals with any RHC within a 10‐mile straight‐line distance radius and any FQHC within a 10‐mile straight‐line distance radius. Second, using Google Maps, we calculated driving distances between each closed rural hospital address and each RHC and FQHC. We then calculated the percent of rural hospital closures with each of the facility types within a 10‐mile driving distance radius. Driving distance was calculated for all years available prior to the closure, the year of closure, and all the years available post closure (minimum −10 years prior to closure and maximum is +10 years post closure). Eight facilities that were inaccessible via roads, for example, an RHC on an island, were included in straight‐line distance calculations and excluded from driving distance calculations. For RHCs and FQHCs with invalid addresses due to road name changes over time (<1%), we calculated distances using the centroid of the ZIP code of the address.

2.2. Descriptive statistics: Time trends

First, we describe the trends over time in the number of rural hospitals, rural hospital closures, FQHC sites, and RHCs in rural ZIP codes from 2006 to 2018. Second, we conduct the Chow test to test for differences in trends of rural hospital closures over time by Census regions. Finally, we describe the trend in rural FQHCs and RHCs pre closure and post closure. Although ZIP codes may change from rural to urban or urban to rural over time, the proportion of ZIP codes that were rural was consistent (~45%) over time.

2.3. Two‐way fixed effects

To estimate the association of rural hospital closures on the predicted probability of having (1) any RHC and (2) any FQHC within 10 straight‐line miles, we conducted a two‐way fixed effects analysis for each outcome. We controlled for whether a hospital ever closed, the year, number of years pre and post closure, the percent living in poverty in the county of the hospital, and the interaction of years pre/post closure, and the percent living in poverty. We controlled for poverty to confirm that counties with a higher percentage of the population living in poverty are associated with a higher likelihood of an FQHC or RHC site. We used pooled generalized linear models with a binomial distribution and logit link to estimate the association of time since a rural hospital closed and the probability of having any of a facility type within 10 miles. Standard errors were clustered at the level of the hospital to allow for the correlation of repeated outcomes observed for a single hospital over time. The standard errors for marginal effects were calculated using the delta method. The Sheps Center defines a closure as ceasing to provide “general, short‐term, acute inpatient care […] We did not consider a hospital closed if it: merged with, or was sold to, another hospital but the physical plant continued to provide inpatient acute care, converted to critical access status, or Both closed and reopened during the same calendar year and at the same physical location.” To address the resultant possibility that the CMS file could list a hospital as open and the Sheps Center could list a hospital as closed, we removed duplicates (rural hospitals within 1 mile of each other in a given year). ³ If a hospital was closed according to the closed list, but was still active according to CMS, then we marked the hospital as closed. We found 68 hospitals were conflicting in between the two data sources. Of those 68 hospitals, 56 were conflicting based on 1 year (eg, Sheps marked a hospital as closed in 2009 while the hospital was last observed in the CMS data in 2008). Finally, we required all hospitals to contribute to at least one pre‐period and one post‐period to be included in the analysis.

2.4. Sensitivity analysis

We conducted a sensitivity analysis to test whether our results were robust using an event study design. ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ Unlike the primary analysis described above, the sensitivity analysis addresses underlying differences between hospitals that close and hospitals that never close through a flexible model including hospital‐level fixed effects. We examine the effect of a rural hospital closure on the probability of having any FQHC or RHC within 10 miles, as shown in Equation (1), where Y is whether hospital h has any FQHC (or RHC) within 10 miles in year t, α is hospital fixed effects, $\delta$ is year fixed effects, and $I\left[t-E_i=k\right]$ represents an indicator for the number of years since closure for hospital h at time t.

$$Y_{\mathit{ht}}={\alpha }_h+{\delta }_t+\sum _k{\beta }_{k}I\left[t-E_h=k\right]+{\epsilon }_{\mathit{ht}}$$ (1)

In the sensitivity analysis, trends in the predicted probability of having any FQHC or RHC over time are similar to the primary analysis, see Appendix  C Figures C1 and C2, respectively.

All statistical testing was two‐sided with a level of significance set at 0.05. All analyses were conducted using Stata, version 16 (StataCorp LLC, College Station, TX). The Institutional Review Board of our institution reviewed and determined the research to be exempt from IRB review.

3. RESULTS

3.1. Hospitals, Federally Qualified Health Centers and Rural Health Clinics, 2006–2018

Between 2006 and 2018, we find a 5.7% decrease in the overall number of rural and urban hospitals in the continental United States (Figure 1(A)). In general, the number of rural FQHC and RHC health facilities in the United States increased between 2006 and 2018 (Figure 1(B)). We found a 131.0% increase in the number of FQHC sites from 1289 in 2006 to 2978 in 2018. The percent decrease in the number of hospitals in rural areas was larger than hospitals in urban ZIP codes, at 5.9% and 5.5%, respectively.

FIGURE 1.

Overall trends in facilities, 2006‐2018. (A) Trends in number of acute care hospitals, 2006–2018. B. Trends in Federally Qualified Health Centers and Rural Health Clinics, 2006–2018. Source: Created by the authors. Data from Centers for Medicare and Medicaid Services Provider of Services Files: 2006‐2018

The number of rural hospital closures varied significantly by geographic region, with higher rates of closures in the South Census region beginning in 2012 compared to other regions (p < 0.01), see Figure 2(A). Numbers of rural FQHC sites were highest in the South Census region at 1604 in 2018 and increased from 2006 to 2018 by 153% (Figure 2(B)). The Midwest, West, and Northeast had 142%, 91%, and 95% increases, respectively. We also found a 18.3% increase in the number of RHCs from 3099 in 2006 to 3665 in 2018. The number of RHCs increased over time with more clinics in the Midwest and West regions at 1532 and 1512 in 2018, respectively (Figure 2(B)). The South, Midwest, West, and Northeast had 23%, 15%, 17%, and 6% changes, respectively.

FIGURE 2.

Trends in rural facilities by geographic region. (A) Rural hospital closures, 2005–2019. (B) Federally Qualified Health Centers and Rural Health Clinics by region, 2006‐2018. Source: Created by the authors using publicly available data from Centers for Medicare and Medicaid Provider of Services Files: 2006‐2018 and from the NC Rural Health Research Program. Citation: Rural Hospital Closures: January 2005—Present. The Cecil G. Sheps Center for Health Services Research. The University of North Carolina at Chapel Hill. Copyright 2014. South Census Region: Delaware, District of Columbia, Florida, Georgia, Maryland, North Carolina, South Carolina, Virginia, West Virginia, Alabama, Kentucky, Mississippi, Tennessee, Arkansas, Louisiana, Oklahoma, and Texas. Midwest: Indiana, Illinois, Michigan, Ohio, Wisconsin, Iowa, Nebraska, Kansas, North Dakota, Minnesota, South Dakota, and Missouri. West: Arizona, Colorado, Idaho, New Mexico, Montana, Utah, Nevada, Wyoming, Alaska, California, Hawaii, Oregon, and Washington. Northeast: Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island, and Vermont

3.2. Federally Qualified Health Centers and Rural Health Clinics pre‐ and postrural hospital closure

Using driving distance, the number of RHCs and FQHCs within 10 miles of a closed hospital increased over time (Figure 3). We found that 2 years prior to closure, 33.7% of rural hospitals had an FQHC within 10 miles driving distance, and 2 years after closure, 42.0% of those areas had an FQHC. Similarly, 2 years before closing, 51.5% of closing rural hospitals had an RHC within 10 miles driving distance, and 52.0% of those same areas had an RHC within 10 miles 2 years after closure. The slope of lines prior to and post closure was significantly different for RHCs but not for FQHCs (see Appendix A for additional details). The trend in FQHCs continued through the hospital closure, and we do not observe the classic dramatic increase we may expect if the relationship was causal.

FIGURE 3.

Proportion of rural communities with at least one FQHC or RHC facility within 10 miles before and after hospital closure. FQHC, Federally Qualified Health Center; RHC, Rural Health Clinic. Source: Created by the authors. Data from Centers for Medicare and Medicaid Provider of Services Files: 2006–2018

Using straight‐line distance, we examined the association of years since a rural hospital closure and the predicted probability of having at least one FQHC within 10 miles (see Table 1). Compared to the year of closure and hospitals that never closed, a rural hospital closure was associated with an increase of 5.95 percentage points at 1 year post closure; 7.41 percentage points at 2 years post closure; 11.94 percentage points increase at 3 years post closure; 11.56 percentage points at 5 years post closure; and 15.44 percentage points at 6 years post closure (all estimates with a p < 0.05). The predicted probability of having any RHC within 10 miles was not significantly different following rural hospital closure. Please see Appendix B for complete model output.

TABLE 1.

Association of rural hospital closures and poverty on availability of Federally Qualified Health Center (FQHC) or Rural Health Clinics (RHC) services

	Change in probability of any RHC	Change in probability of any FQHC
	Marginal effect (standard error)	Marginal effect (standard error)
10 years Pre‐closure	0.0909 (0.0949)	−0.2048** (0.0746)
9 years Pre‐closure	0.0398 (0.0631)	0.0446 (0.0800)
8 years Pre‐closure	0.0796 (0.0545)	−0.0145 (0.0655)
7 years Pre‐closure	0.0346 (0.0571)	0.0129 (0.0586)
6 years Pre‐closure	0.0124 (0.0524)	−0.0084 (0.0522)
5 years Pre‐closure	0.0083 (0.0501)	−0.0190 (0.0493)
4 years Pre‐closure	0.0303 (0.0527)	−0.0013 (0.0495)
3 years Pre‐closure	0.0730 (0.0444)	0.0286 (0.0420)
2 years Pre‐closure	0.0633 (0.0422)	0.0255 (0.0354)
1 year Pre‐closure	0.0534 (0.0373)	0.0320 (0.0260)
Year of hospital closure	‐ref‐	‐ref‐
1 year Post‐closure	−0.0165 (0.0351)	0.0595** (0.0272)
2 years Post‐closure	0.0470 (0.0461)	0.0741** (0.0360)
3 years Post‐closure	0.0457 (0.0509)	0.1195** (0.0452)
4 years Post‐closure	0.0826 (0.0559)	0.0814 (0.0498)
5 years Post‐closure	0.0555 (0.0650)	0.1157** (0.0588)
6 years Post‐closure	0.0895 (0.0736)	0.1545** (0.0682)
7 years Post‐closure	0.1341* (0.0781)	0.0596 (0.0763)
8 years Post‐closure	0.1381 (0.0884)	0.0724 (0.0843)
9 years Post‐closure	0.1268 (0.0954324)	0.0691 (0.0884)
10 years Post‐closure	0.1698 (0.1136)	0.0447 (0.1244)
Percent in poverty	0.0129** (0.0015)	0.0198** (0.0013)

Note: The above table shows the average marginal effects of rural hospital closures and poverty rates on the probabilities of having any FQHC within 10 straight‐line miles and any RHC within 10 straight‐line miles after a closure. For example, on average, compared to the year of closure or never having closed, 1‐year postrural hospital closure is associated with a 5.95 percentage point increase in probability of having any FQHC within 10 miles, controlling for all else in the model. Poverty rates are defined as percent of hospital county population living in poverty.

Abbreviations: FQHC, Federally Qualified Health Center; RHC, Rural Health Clinic.

^* p < 0.10.

^** p < 0.05.

Source: Created by the authors.

3.3. Federally Qualified Health Centers and Rural Health Clinics and percent of people living in poverty

Finally, we examined the association of percent living in poverty in the county of the hospital on the predicted probability of having at least one FQHC and at least one RHC within 10 miles (Figure 4). On average, every percentage point increase in percent living in poverty in the county is associated with a 1.98 percentage point increase in probability of having an FQHC (p < 0.001) and a 1.27 percentage point increase in probability of having an RHC (p < 0.001), controlling for all else in the model (see Table 1). We also observe an increased probability of having an FQHC in areas with a higher percent of the county living in poverty, but the difference is not statistically significant (Figure 4).

FIGURE 4.

Effect of percent of a county living in poverty on probability of having FQHC and RHC within 10 miles. FQHC, Federally Qualified Health Center; RHC, Rural Health Clinic. Source: Created by the authors. Note: The figure illustrates probabilities of a rural hospital having any FQHC within 10 miles, and any RHC within 10 miles by high and low percent of a county living in poverty, conditional on whether or not a rural hospital ever closed, year and years since closure. We define “high” percent of county living in poverty as at or above 20.4% as it is the 75th percentile of the distribution of percent of county living in poverty. We define “low” percent of county living in poverty as at or below 12.3% as it is the 25th percentile of the poverty. Predicted probabilities were calculated using the method of recycled predictions at each percentile

4. DISCUSSION

This study examined outpatient services availability from FQHCs and RHCs following hospital closure. Nationally, the number of FQHCs and RHCs in rural and urban areas has increased substantially since 2006. Over the same time period, there was a substantial increase in the number of rural hospitals that closed in the United States This was particularly evident in the South census region where rural hospitals closed at a faster rate compared to other regions. After a rural hospital closes, the number of FQHCs within 10 miles of the closed facility increased over time. Additionally, as expected, communities with a higher percent of the population living in poverty have increased probability of an FQHC or RHC within 10 miles. This has implications for the health of rural communities and what FQHCs and RHCs should do for communities vulnerable to decreased access to outpatient care.

Our findings suggest that some of the potential reduction in access to outpatient primary care services caused by rural hospital closures may be offset by new FQHC and RHC sites within a few miles of a closed hospital. In general, the increase in availability of services from FQHCs and RHCs following hospital closures is good news for rural health and access to care. RHCs and FQHCs play a critical role in serving rural communities in the United States. As of 2019, approximately 20% of rural residents in the United States are served by community health centers. ²¹ Health centers have expanded capacity and services through the addition of behavioral health specialists and telehealth, despite reimbursement barriers. ²¹ , ²² Compared to urban hospitals, most rural hospitals provide a much higher percentage of their care on an outpatient basis than urban hospitals. Rural hospital outpatient department and FQHC/RHC patients are also similar, for example, the most common diagnoses of rural Medicare beneficiaries at FQHCs, RHCs, critical access hospitals, and rural prospective payment system hospitals are similar. ¹⁵ Thus, health centers play a meaningful role in replacing some services when a rural hospital closes, though many gaps remain.

The increase in FQHC service‐delivery sites following hospital closure suggests that new approvals reflect the need in the community in the years post closure to some extent. Furthermore, the relationship between poverty and access to FQHCs reflects the use of poverty as a HRSA criterion. This finding confirms that policy is effective in increasing access in high‐poverty areas. HRSA is implementing improvements to better measure need in their approval process for new FQHCs. The new Service Area Needs Assessment Methodology (SANAM) developed by HRSA in 2019 aims to reflect persistent unmet needs in rural communities. The SANAM model consists of 24 weighted measures that emphasize the determinants that shape lack of access to primary and preventive health care and the disparities in health status and determinants that are especially relevant to Health Center Program populations and other underserved communities. Tracking SANAM scores in areas with financially vulnerable hospitals and in areas that have lost a hospital may assist policy makers in identifying areas most likely to benefit from additional resources supporting new health center sites. For example, additional pilot grant programs supporting the expansion of health centers in rural communities losing the local rural hospital could address issues with accessing sufficient capital to rapidly pursue opening a new site.

Despite the increase in FQHCs following hospital closure, rural communities may experience unmet need for care. Limited outpatient resources and distance to acute and emergency care needs remain a concern for rural communities, particularly following hospital closure. Since 2018, rural hospital closures have continued with 17 closed in 2019 and 20 closed in 2020. ³ Prior to the coronavirus disease 2019 (COVID‐19) pandemic, approximately 39% of “rural hospitals [operated] in the red.” ²³ The potential effects of COVID‐19 pandemic and response may create challenges for rural hospitals already struggling to maintain financial viability. In fact, recent news coverage has highlighted the tenuous financial situation rural hospitals face due to the pandemic as hospitals experience a dramatic reduction in patients pursuing usual care. ²⁴ , ²⁵ , ²⁶ , ²⁷ A recent study estimated that rural residents will generate an estimated 10% more hospitalizations for COVID‐19 per capita than urban residents. ²⁸ In May 2020, the American Hospital Association estimated hospitals and health systems would lose approximately $202.6 billion between March 1 to June 30, 2020 as a result of the epidemic and response. ²⁹

FQHCs and RHCs may be better positioned to respond to the forces that drive hospital closures but face other threats. First, FQHCs largely serve uninsured and underinsured patients, and 82% of FQHC funding is from federal and state sources. ³⁰ However, federal funding for FQHCs established by the Affordable Care Act requires reauthorization every 2 years, creating uncertainty in long‐term financial support. The 2021 appropriations act recently extended core funding for health centers but failed to provide the additional 7.6 billion in emergency funding requested for health centers to maintain services and personnel during the COVID‐19 pandemic. ³¹ The 2020 CARES act allocated $1.3 billion for health centers in response to the pandemic. ³² Prior to the pandemic, over 40% of community health centers had negative operating margins. ³³ During the COVID‐19 pandemic, community health centers lost substantial revenues due to the sharp decrease in clinic visits. Despite these challenges, community health centers have stepped up to provide COVID‐19 testing and vaccination, working in coordination with public health departments. Second, payment rates are higher for RHCs based in small hospitals (<50 beds) than for independent RHCs. As a result, the loss of the hospital removes the incentive for locating an RHC in the area, and the net effect of a closing hospital may negate any RHC expansion in other communities.

This study should be interpreted in the context of the limitations. First, we are unable to evaluate whether the increase in FQHC and RHC sites opening post rural hospital closure is fully addressing the needs of the community. Second, potential for measurement error may reduce the precision of our results; for example, address and street changes over time could impact the distance calculations and facilities which were separated by bodies of water, and other undrivable barriers are not reflected in calculations for driving distances. Third, the primary study design does not support causal inference or address selection into treatment (hospital closure). Our analysis demonstrates that FQHCs are more likely to open new sites following a rural hospital closure, but this increase may not be caused by the hospital closure. The secular trend in FQHC expansion may manifest as a post‐closure increase. Indeed, we observed no statistically significant difference in trends pre and post closure for FQHCs (Appendix A). Additionally, an area may be designated an MUA prior to a hospital closure, or after a hospital closure, which may mediate the relationship between hospital closure and FQHC availability. Finally, the analysis focuses on outpatient services pre‐ and post‐ rural hospital closures, and thus, results are not generalizable to hospital closures or prevalence of FQHCs in urban areas. A strength of this study is the use of national data for urban and rural hospitals as well as two measures of distance (Haversine and driving distance) to increase generalizability of the findings. Our findings were robust to multiple specifications and robustness checks.

Rural residents benefit from increased availability of FQHC and RHC services following rural hospital closure, improving access to essential preventive and diagnostic services; however, rural communities may have excess need for preventive and therapeutic care that remains unmet. Understanding the availability of health care services in rural areas is critical in understanding the long‐term impacts of rural hospital closure on the health of rural communities. Future work should examine the degree to which unmet need may or may not persist as health centers expand to improve access to care, for example, through the designation of MUA, and how policies and payment reforms can be designed to maintain access to care.

APPENDIX A.

A.1. Federally Qualified Health Centers

We use a linear probability model with time since closure trends (both pre and post‐closure) interacted with an indicator for post‐closure to examine pre‐ and post‐closure trends. Each 1‐year increase is associated with a 1.3 percentage point increase in having an FQHC within 10 miles of driving distance (p < 0.05). The coefficient on the postclosure indicators corresponds to a 3.77 percentage point increase in probability of having an FQHC within 10 miles of driving distance. The coefficient of the interaction of time since closure and a post period was not statistically significantly different (p = 0.470), suggesting no statistical significance between pre‐ and postclosure trends.

A.2. Rural Health Clinics

We use a linear probability model with time since closure trends (both pre and post‐closure) interacted with an indicator for post‐closure to examine pre‐ and post‐closure trends. Each 1‐year increase is associated with a 1.1 percentage point decrease in having an RHC within 10 miles of driving distance (p < 0.05). The coefficient on the postclosure indicators was not statistically significant. The coefficient of the interaction of time since closure and a post period was statistically significantly different (p < 0.001), suggesting statistical significance between pre‐ and postclosure trends in probability of having an RHC within 10 miles of driving distance.

APPENDIX B.

TABLE B1.

Model coefficients

	(1)	(2)
	Any Rural Health Clinic within 10 miles	Any Federally Qualified Health Center within 10 miles
Hospital ever closed	−0.326	0.0251
	(0.196)	(0.217)
Hospital never closed	‐ref‐	‐ref‐
Years since hospital closed
−10	−0.528	−1.976
	(1.651)	(2.164)
−9	−2.588	1.018
	(1.411)	(1.249)
−8	−0.462	0.609
	(0.922)	(1.077)
−7	−0.898	0.750
	(1.030)	(0.908)
−6	−0.408	0.484
	(0.826)	(0.888)
−5	−0.455	0.00591
	(0.777)	(0.881)
−4	−0.0605	0.528
	(0.820)	(0.876)
−3	0.846	1.012
	(0.692)	(0.712)
−2	0.549	0.999
	(0.649)	(0.628)
−1	0.373	1.014
	(0.671)	(0.572)
0	‐ref‐	‐ref‐
1	0.206	1.004
	(0.638)	(0.649)
2	0.470	1.050
	(0.695)	(0.680)
3	0.516	1.401 *
	(0.711)	(0.686)
4	1.670 *	0.902
	(0.703)	(0.760)
5	1.554 *	1.403
	(0.746)	(0.791)
6	0.863	0.816
	(0.974)	(0.971)
7	1.708	0.551
	(1.124)	(1.014)
8	1.105	−0.0520
	(1.341)	(1.255)
9	1.353	−0.270
	(1.386)	(1.163)
10	2.250	−0.0599
	(1.917)	(1.726)
Percent of county living in poverty, all ages	0.0545 ***	0.0976 ***
	(0.00697)	(0.00730)
Interaction of years since hospital closed and percent of county living in poverty.
−10 X percent of county living in poverty, all ages	0.0581	0.0377
	(0.0913)	(0.0960)
−9 X percent of county living in poverty, all ages	0.176 *	−0.0469
	(0.0843)	(0.0635)
−8 X percent of county living in poverty, all ages	0.0504	−0.0388
	(0.0519)	(0.0528)
−7 X percent of county living in poverty, all ages	0.0648	−0.0397
	(0.0556)	(0.0436)
−6 X percent of county living in poverty, all ages	0.0281	−0.0300
	(0.0429)	(0.0424)
−5 X percent of county living in poverty, all ages	0.0298	−0.00561
	(0.0398)	(0.0408)
−4 X percent of county living in poverty, all ages	0.0115	−0.0307
	(0.0413)	(0.0411)
−3 X percent of county living in poverty, all ages	−0.0321	−0.0507
	(0.0331)	(0.0337)
−2 X percent of county living in poverty, all ages	−0.0168	−0.0508
	(0.0315)	(0.0302)
−1 X percent of county living in poverty, all ages	−0.00878	−0.0500
	(0.0337)	(0.0284)
0 X percent of county living in poverty, all ages	‐ref‐	‐ref‐
1 X percent of county living in poverty, all ages	−0.0165	−0.0424
	(0.0330)	(0.0331)
2 X percent of county living in poverty, all ages	−0.0163	−0.0413
	(0.0355)	(0.0342)
3 X percent of county living in poverty, all ages	−0.0195	−0.0506
	(0.0365)	(0.0347)
4 X percent of county living in poverty, all ages	−0.0781 *	−0.0308
	(0.0345)	(0.0396)
5 X percent of county living in poverty, all ages	−0.0784 *	−0.0516
	(0.0366)	(0.0418)
6 X percent of county living in poverty, all ages	−0.0288	−0.00675
	(0.0553)	(0.0591)
7 X percent of county living in poverty, all ages	−0.0666	−0.0159
	(0.0609)	(0.0593)
8 X percent of county living in poverty, all ages	−0.0297	0.0229
	(0.0802)	(0.0765)
9 X percent of county living in poverty, all ages	−0.0477	0.0349
	(0.0781)	(0.0697)
10 X percent of county living in poverty, all ages	−0.0874	0.0158
	(0.105)	(0.113)
Year
2006	‐ref‐	‐ref‐
2007	0.0637 ***	0.150 ***
	(0.0172)	(0.0230)
2008	0.0430	0.259 ***
	(0.0231)	(0.0318)
2009	0.00553	0.245 ***
	(0.0280)	(0.0362)
2010	0.0248	0.256 ***
	(0.0337)	(0.0401)
2011	0.0598	0.245 ***
	(0.0373)	(0.0440)
2012	0.0976 *	0.355 ***
	(0.0391)	(0.0474)
2013	0.149 ***	0.440 ***
	(0.0413)	(0.0499)
2014	0.194 ***	0.654 ***
	(0.0424)	(0.0521)
2015	0.258 ***	0.828 ***
	(0.0444)	(0.0541)
2016	0.301 ***	0.971 ***
	(0.0453)	(0.0562)
2017	0.382 ***	1.122 ***
	(0.0472)	(0.0579)
2018	0.468 ***	1.200 ***
	(0.0489)	(0.0595)
Constant	−0.776 ***	−2.814 ***
	(0.121)	(0.136)
Observations	27 422	27 422

Note: Standard errors in parentheses.

^* p < 0.05.

^*** p < 0.001.

APPENDIX C.

FIGURE C1.

Probability of having any Federally Qualified Health Center within 10 miles relative to rural hospital closure [Color figure can be viewed at wileyonlinelibrary.com]

FIGURE C2.

Probability of having any Rural Health Clinic within 10 miles relative to rural hospital closure [Color figure can be viewed at wileyonlinelibrary.com]

Miller KEM, Miller KL, Knocke K, Pink GH, Holmes GM, Kaufman BG. Access to outpatient services in rural communities changes after hospital closure. Health Serv Res. 2021;56(5):788–801. 10.1111/1475-6773.13694