Characteristics of Hospitals by Graduate Medical Education Expense Category: Implications for Rural Residency Program Expansion.

PURPOSE: To describe how the characteristics of the hospitals and communities they serve vary across the 4 hospital graduate medical education (GME) expense categories (according to Section 131 of the Consolidated Appropriations Act of 2021) and identify the rurally located never claimer hospitals that are most similar to teaching hospitals, signaling that they might be good candidates for new rural GME programs. METHOD: Hospital categories and characteristics were gathered from the March 2022 Medicare Cost Reports; 2022 County Health Rankings & Roadmaps data were used for community characteristics. Each acute hospital was classified into 1 of the following 4 mutually exclusive hospital categories: category A, category B, established teaching hospital (ETH), and never claimer. Multinomial logistic regressions were conducted to estimate the adjusted associations of hospital characteristics with hospital categories and to identify the never claimer hospitals in rural locations that have characteristics similar to teaching hospitals (category A, category B, and ETHs). RESULTS: Out of 3,590 hospitals, 2,075 (57.8%) were never claimer hospitals. After adjusting for multiple characteristics, rural hospitals had a similar probability of being in each hospital category as that of urban hospitals. Never claimer hospitals served an older population and were located in communities with more uninsured adults and children and less availability of primary care physicians, dentists, and mental health professionals. CONCLUSIONS: This study demonstrated that most hospitals in every category, but especially teaching hospitals (i.e., category A hospitals, category B hospitals, and ETHs), were concentrated in urban areas. Larger hospitals (measured by net patient revenue) were more likely to report GME expenses (i.e., be a category A hospital, a category B hospital, or an ETH). The study suggests that there are roughly 145 rural never claimer hospitals that might be strong candidates for initiating new residency programs.

Impact of including drug spending in oncology alternative payment models.

OBJECTIVES: To develop a method for determining the effect of including drug costs in alternative payment models (APMs). STUDY DESIGN: Retrospective claims analysis. METHODS: Using the Oncology Care Model as an example, we developed an oncology episode payment model for a commercial payer using historical claims data. We defined 6-month episodes of chemotherapy. Using claims data, we characterized episodes and developed a risk adjustment model. We used bootstrapping to estimate the variation in episode cost with drugs included and without. RESULTS: Episode costs were approximately $100,000. Although absolute cost variation was higher when we included drugs, the percent of total cost represented by variation was lower. Under reasonable assumptions about potential savings from drug and nondrug spending, our results suggest that including drugs in APMs can improve the risk-benefit trade-off faced by provider groups. We introduce a risk-mitigated sharing rate that may enable inclusion of drugs in APMs without substantially increasing downside risk. CONCLUSIONS: We have developed a method to assess whether the inclusion of drug spending in APMs is a good decision for provider groups. Including drug costs in episode payments for oncology patients may be preferable for many provider groups.

Bolstering the rural physician workforce in underserved communities: Are Rural Residency Planning and Development Programs finding the sweet spot?

PURPOSE: The purpose of this study is to describe the characteristics of Rural Residency Planning and Development (RRPD) Programs, compare the characteristics of counties with and without RRPD programs, and identify rural places where future RRPD programs could be developed. METHODS: The study sample comprised 67 rural sites training residents in 40 counties in 24 US states. Descriptive statistics were used to describe RRPD programs and logistic regression to predict the probability of a county being an RRPD site as a function of population, primary care physicians (PCP) per 10,000 population, and the social vulnerability index (SVI) compared to a control sample of nonmetro counties without RRPD sites. FINDINGS: Most RRPD grantees (78%) were family medicine programs affiliated with medical schools (97%). RRPD counties were more populous (P<.01), had a higher population density (P<.05), and a higher percent of the non-White or Hispanic population (P = .05) compared to non-RRPD counties. Both higher population (P<.001) and PCP ratio (P = .046) were strong predictors, while SVI (P = .07) was a weak predictor of being an RRPD county. CONCLUSIONS: RRPD sites appear to represent a "sweet spot" of rural counties that have the population and physician supply to support a training program but also are relatively more socially vulnerable with high-need populations. Additional counties fitting this "sweet spot" could be targeted for funding to address health disparities and health workforce maldistribution.

The Landscape of State Policies Supporting Family Caregivers as Aligned With the National Academy of Medicine Recommendations.

Policy Points In the absence of federal policy, states adopted policies to support family caregivers, but availability and level of support varies. We describe, compare, and rank state policies to support family caregivers as aligned with National Academy of Medicine recommendations. Although the landscape of state policies supporting caregivers has improved over time, few states provide financial supports as recommended, and benefit restrictions hinder accessibility for all types of family caregivers. Implementing policies supporting family caregivers will become more critical over time, as the reliance on family caregivers as essential providers of long-term care is only expected to grow as the population ages. CONTEXT: In the United States in 2020, approximately 26 million individuals provided unpaid care to a family member or friend. On average, 60% of caregivers were employed, and they provided 20.4 hours of care per week on top of employment. Although a handful of patchwork laws exist to aid family caregivers, systematic supports, including comprehensive training, respite, and financial support, remain limited. In the absence of federal supports, states have adopted policies to provide assistance, but they vary in availability and level of support provided. Our objectives were to describe, compare, and rank state policies to support family caregivers over time. METHODS: We used publicly available data from the AARP Long-Term Services and Supports State Scorecard, the National Academy for State Health Policy, and Tax Credits for Workers and Families for all 50 states and the District of Columbia (2015-2019). FINDINGS: We found that states had increased supports to family caregivers over this five-year period, although significant variability in adoption and implementation of policies persists. Approximately 20% of states had enacted policies that exceed the federal Family and Medical Leave Act requirements, and 18% offered paid family leave. However, most states had not improved spousal impoverishment protections for Medicaid beneficiaries. For example, from 2016 to 2019, 24% of states provided fewer or no protections, while 71% of states did not improve spousal impoverishment protections over time. Access to training for caregivers varied based on eligibility criteria (e.g., select populations and/or only co-residing caregivers). CONCLUSIONS: Overall, state approaches to support family caregivers vary by eligibility and scope of services. Substantial gaps in support of caregivers, particularly economic supports, persist. Although the landscape of state policies supporting caregivers has improved over time, few states provide financial supports as recommended by the National Academy of Medicine, and benefit restrictions hinder accessibility for all family caregivers.

The economic effects of rural hospital closures.

OBJECTIVE: To provide an updated analysis of the economic effects of rural hospital closures. STUDY SETTING: Our study sample was national in scope and consisted of nonmetro counties from 2001 to 2018. STUDY DESIGN: We used a difference-in-differences study design to estimate the effect of a hospital closure on county income, population, unemployment, and size of the labor force. Specifically, we compared economic changes over time in nonmetro counties experiencing a hospital closure to changes in a control group of nonmetro counties over the same time period. We also leveraged insight from recent research to control for estimation bias due to heterogeneity in the closure effect over time or across groups defined by when closure was experienced. DATA EXTRACTION: Data on (adjusted gross) annual income (in real dollars), annual population size, and monthly unemployment rate and labor force size were sourced from the Internal Revenue Service, Census Bureau, and Bureau of Labor Statistics, respectively. We used data from the North Carolina Rural Health Research Program to identify counties that experienced a hospital closure. PRINCIPAL FINDINGS: Of the 1759 nonmetro counties in our study sample, 109 experienced a hospital closure during the study period. Relative to the nonclosure counterfactual, closures significantly decreased labor force size, on average, by 1.4% (95% CI: [-2.1%, -0.8%]). Results also suggest that Prospective Payment System (PPS) hospital closures significantly decreased population size, on average, by 1.1% (95% CI: [-1.7%, -0.5%]), relative to the nonclosure counterfactual. CONCLUSIONS: Our analysis suggests that rural hospital closures often have adverse effects on local economic outcomes. Importantly, the negative economic effects of closure appear to be strongest following Prospective Payment System hospital closures and attenuated when the closed hospital is converted to another type of health care facility, allowing for the continued provision of services other than inpatient care.

The contributions of survey-based versus administrative measures of socioeconomic status in predicting type of post-acute care for hospitalized Medicare beneficiaries.

OBJECTIVES: To assess and compare the associations between socioeconomic status (SES) measures from two sources (claims vs. survey data) and the type of post-acute care (PAC) locations following hospital discharge. METHODS: This observational study included Medicare Fee-for-Service (FFS) beneficiaries age 65.5 years or older who participated in the Medicare Current Beneficiary Survey (MCBS) and were hospitalized in 2006-2011. Multiple data sets were used including: Area Deprivation Index; Medicare Cost Reports, Provider of Services files, and Area Health Resource File. Multinomial regression models estimated associations between beneficiary's SES and PAC type. SES measures came from surveys (income and education) and administrative records (dual enrollment and area deprivation). PAC types included home with self-care, home health agency, skilled nursing facility (SNF), or inpatient rehabilitation facility. RESULTS: Low income and dual enrollment were associated with higher SNF use while living in a deprived area was associated with lower SNF use and higher use of home with self-care. Dual enrollment and area deprivation were associated with the largest differences. CONCLUSIONS: If policies to modify payment based on SES are considered, administrative measures (dual enrollment and area deprivation) rather than survey measures (education and income) may be sufficient.

Which Definition of Rurality Should I Use?: The Relative Performance of 8 Federal Rural Definitions in Identifying Rural-Urban Disparities.

BACKGROUND: The federal government uses multiple definitions for identifying rural communities based on various geographies and different elements of rurality. OBJECTIVES: The objectives of this study were to: (1) assess the degree to which rural definitions identify the same areas as rural; and (2) assess rural-urban disparities identified by each definition across socioeconomic, demographic, and health access and outcome measures. RESEARCH DESIGN: We determined the rural status of each census tract and calculated the rural-urban disparity resulting from each definition, as well as across the number of definitions in which tracts were designated as rural (rurality agreement). SUBJECTS: The population in 72,506 census tracts. MEASURES: We used 8 federal rural definitions. Population characteristics included percent with a bachelor's degree, income below 200% poverty, population density, percent with health insurance and whether various health care services were within 30 minutes driving time of the tract centroid. RESULTS: The rural population varied from slightly < 6.9 million people to >75.5 million across definitions. The largest rural-urban disparities were found using Urban Influence Codes. Urbanized Area and Urbanized Cluster tended to generate smaller disparities. Population characteristics such as population density and percent White had notable discontinuities across levels of rurality, while others such as percent with a bachelor's degree and income below 200% poverty varied continuously. CONCLUSIONS: Rural-urban populations and disparities were sensitive to the specific definition and the relative strength of definitions varied across population characteristics. Researchers and policymakers should carefully consider the choice of outcome and region when deciding the most appropriate rural definition.

US Primary Care vs Specialty Care Trainee Positions and Physician Incomes: Trends From 2001 to 2019.

BACKGROUND: Much of the Affordable Care Act (ACA) and subsequent US health care policies were designed to address deficiencies in health care access and enhance primary care services. How residency positions and physician incomes have changed in the post-ACA era is not well characterized. OBJECTIVE: We evaluated the growth of US trainee positions and physician income, in the pre- vs post-ACA environment by specialty and among primary care vs specialty care. METHODS: Total resident complement by specialty and year was extracted from the National Graduate Medical Education (GME) Census and stratified into primary care vs specialty care. Median incomes were extracted from Medical Group Management Association surveys. Piecewise linear regression with interaction terms (pre-ACA, 2001-2010, vs post-ACA, 2011-2019) assessed growth rate by specialty and growth rate differences between primary care and specialty care. Sensitivity analyses were performed by focusing on family medicine and excluding additional GME positions contributed by the introduction of the 2015 single GME accreditation system. RESULTS: Resident complements increased for primary care (+0.16%/year pre-ACA to +2.06%/year post-ACA, P < .001) and specialty care (+1.49%/year to +2.07%/year, P = .005). Specialty care growth outpaced primary care pre-ACA (P < .001) but not post-ACA (P = .10). Family medicine had the largest increase in the pre- vs post-ACA era (-0.77%/year vs +2.09%/year, P < .001). Excluding positions contributed by the single GME accreditation system transition did not result in any statistically significant changes to the findings. Income growth increased for primary care (+0.84%/year to +1.37%/year, P = .044), but decreased for specialty care (+1.44%/year to +0.49%/year, P = .011). Specialty care income growth outpaced primary care pre-ACA (P < .001), but not post-ACA (P = .22). CONCLUSIONS: We found significant growth differences in resident complement and income among primary care versus specialty care in the pre-/post-ACA eras.

Decline in Inpatient Volume at Rural Hospitals.

PURPOSE: To investigate (1) all-payer inpatient volume changes at rural hospitals and (2) whether trends in inpatient volume differ by organizational and geographic characteristics of the hospital and characteristics of the patient population. METHODS: We used a retrospective, longitudinal study design. Our study sample consisted of rural hospitals between 2011 and 2017. Inpatient volume was measured as inpatient average daily census (ADC). Additional measured hospital characteristics included census region, Medicare payment type, ownership type, number of beds, local competition, total margin, and whether the hospital was located in a Medicaid expansion state. Measured characteristics of the local patient population included total population size, percent of population aged 65 years or older, and percent of population in poverty. To identify predictors of inpatient volume trends, we fit a linear multiple regression model using generalized estimating equations. FINDINGS: Rural hospitals experienced an average change in ADC of -13% between 2011 and 2017. We found that hospital characteristics (eg, census region, Medicare payment type, ownership type, total margin, whether the hospital was located in a Medicaid expansion state) and patient population characteristics (eg, percent of population in poverty) were significant predictors of inpatient volume trends. CONCLUSIONS: Trends in inpatient volume differ by organizational and geographic characteristics of the hospital and characteristics of the patient population. Researchers and policy makers should continue to explore the causal mechanisms of inpatient volume decline and its role in the financial viability of rural hospitals.

What Characteristics Influence Whether Rural Beneficiaries Receiving Care From Urban Hospitals Return Home for Skilled Nursing Care?

PURPOSE: Skilled nursing care (SNC) provides Medicare beneficiaries short-term rehabilitation from an acute event. The purpose of this study is to assess beneficiary, market, and hospital factors associated with beneficiaries receiving care near home. METHODS: The population includes Medicare beneficiaries who live in a rural area and received acute care from an urban facility in 2013. "Near home" was defined 3 different ways based on distances from the beneficiary's home to the nearest source of SNC. Results include unadjusted means and odds ratios from logistic regression. FINDINGS: About 69% of rural beneficiaries receiving acute care in an urban location returned near home for SNC. Beneficiaries returning home were white (odds ratio [OR] black: 0.69; other race: 0.79); male (OR: 1.07); older (OR age 85+ [vs 65-69]: 1.14); farther from SNC (OR: 1.01 per mile); closer to acute care (OR: 0.28, logged miles); and received acute care from hospitals that did not own a skilled nursing facility (owned OR: 0.77) and hospitals with: no swing bed (swing bed OR: 0.47), high case mix (OR: 3.04), and nonprofit status (for-profit OR: 0.85). Results varied somewhat across definitions of "near home." CONCLUSIONS: Rural Medicare beneficiaries who received acute care far from home were more likely to receive SNC far from home. Because Medicare beneficiaries have the choice of where to receive SNC, policy makers may consider ensuring that new payment models do not incentivize provision of SNC away from home.

What Is Rural? Challenges And Implications Of Definitions That Inadequately Encompass Rural People And Places.

Monitoring and improving rural health is challenging because of varied and conflicting concepts of just what rural means. Federal, state, and local agencies and data resources use different definitions, which may lead to confusion and inequity in the distribution of resources depending on the definition used. This article highlights how inconsistent definitions of rural may lead to measurement bias in research, the interpretation of research outcomes, and differential eligibility for rural-focused grants and other funding. We conclude by making specific recommendations on how policy makers and researchers could use these definitions more appropriately, along with definitions we propose, to better serve rural residents. We also describe concepts that may improve the definition of and frame the concept of rurality.

The Association of Discharge Before Noon and Length of Stay in Hospitalized Pediatric Patients.

BACKGROUND AND OBJECTIVES: To optimize patient throughput, many hospitals set targets for discharging patients before noon (DCBN). However, it is not clear whether DCBN is an appropriate measure for an efficient discharge. This study aims to determine whether DCBN is associated with shorter length of stay (LOS) in pediatric patients and whether that relationship is different between surgical and medical discharges. METHODS: From May 2014 to April 2017, we performed a retrospective data analysis of pediatric medical and surgical discharges belonging to a single academic medical center. Patients were included if they were 21 years or younger with at least one night in the hospital. Propensity score weighted multivariate ordinary least squares models were used to evaluate the association between DCBN and LOS. RESULTS: Of the 8,226 pediatric hospitalizations, 1,531 (18.61%) patients were DCBN. In our multivariate model of all the discharges, DCBN was associated with an average of 0.27 day (P = .014) shorter LOS when compared to discharge in the afternoon. In our multivariate medical discharge model, DCBN was associated with an average of 0.30 (P = .017) day decrease in LOS while the association between DCBN and LOS was not significant among surgical discharges. CONCLUSIONS: On average, at a single academic medical center, DCBN was associated with a decreased LOS for medical but not surgical pediatric discharges. DCBN may not be an appropriate measure of discharge efficiency for all services.

The Effect of the Magnet Recognition Signal on Hospital Financial Performance.

EXECUTIVE SUMMARY: The objective of this study was to investigate the effect of the Magnet Recognition (MR) signal on hospital financial performance. MR is a quality designation granted by the American Nurses Credentialing Center (ANCC). Growing evidence shows that MR hospitals are associated with various interrelated positive outcomes that have been theorized to affect hospital financial performance.In this study, which covered the period from 2000 to 2010, we applied a pre-post research design using a longitudinal, unbalanced panel of MR hospitals and hospitals that had never received MR designation located in urban areas in the United States. We obtained data for this analysis from Medicare's Hospital Cost Report Information System, the American Hospital Association Annual Survey Database, the Health Resources & Services Administration's Area Resource File, and the ANCC website. Propensity score matching was used to construct the final study sample. We then applied a difference-in-difference model with hospital fixed effects to the matched hospital sample to test the effect of the MR signal, while controlling for both hospital and market characteristics.According to signaling theory, signals aim to reduce the imbalance of information between two parties, such as patients and providers. The MR signal was found to have a significant positive effect on hospital financial performance. These findings support claims in the literature that the nonfinancial benefits resulting from MR lead to improved financial performance. In the current healthcare environment in which reimbursement is increasingly tied to delivery of quality care, healthcare executives may be encouraged to pursue MR to help hospitals maintain their financial viability while improving quality of care.

Decomposing Mortality Disparities in Urban and Rural U.S. Counties.

OBJECTIVE: To understand the role of county characteristics in the growing divide between rural and urban mortality from 1980 to 2010. DATA SOURCE: Age-adjusted mortality rates for all U.S. counties from 1980 to 2010 were obtained from the CDC Compressed Mortality File and combined with county characteristics from the U.S. Census Bureau, the Area Health Resources File, and the Inter-University Consortium for Political and Social research. STUDY DESIGN: We used Oaxaca-Blinder decomposition to assess the extent to which rural-urban mortality disparities are explained by observed county characteristics at each decade. PRINCIPAL FINDINGS: Decomposition shows that, at each decade, differences in rural/urban characteristics are sufficient to explain differences in mortality. Furthermore, starting in 1990, rural counties have significantly lower predicted mortality than urban counties when given identical county characteristics. We find changes in the effect of characteristics on mortality, not the characteristics themselves, drive the growing mortality divide. CONCLUSIONS: Differences in economic and demographic characteristics between rural and urban counties largely explain the differences in age-adjusted mortality in any given year. Over time, the role these characteristics play in improving mortality has increased differentially for urban counties. As characteristics continue changing in importance as determinants of health, this divide may continue to widen.

Factors Driving Live Discharge From Hospice: Provider Perspectives.

CONTEXT: The proportion of patients disenrolling from hospice before death has increased over the decade with significant variations across hospice types and regions. Such trends have raised concerns about live disenrollment's effect on care quality. Live disenrollment may be driven by factors other than patient preference and may create discontinuities in care, disrupting ongoing patient-provider relationships. Researchers have not explored when and how providers make this decision with patients. OBJECTIVE: The objective of this study was to ascertain provider perspectives on key drivers of live discharge from the Medicare hospice program. METHODS: We conducted semistructured telephone interviews with 18 individuals representing 14 hospice providers across the country. Transcriptions were coded and analyzed using a template analysis approach. RESULTS: Analysis generated four themes: 1) difficulty estimating patient prognosis, 2) fear of Centers for Medicare & Medicaid Services audits, 3) rising market competition, and 4) challenges with inpatient contracting. Participants emphasized challenges underlying each decision to discharge patients alive, stressing that there often exists a gray line between appropriate and inappropriate discharges. Discussions also focused on scenarios in which financial motivations drive enrollment and disenrollment practices. CONCLUSION: This study provides significant contributions to existing knowledge about hospice enrollment and disenrollment patterns. Results suggest that live discharge patterns are often susceptible to market and regulatory forces, which may have contributed to the rising national rate.

Developing Physician Migration Estimates for Workforce Models.

OBJECTIVE: To understand factors affecting specialty heterogeneity in physician migration. DATA SOURCES/STUDY SETTING: Physicians in the 2009 American Medical Association Masterfile data were matched to those in the 2013 file. Office locations were geocoded in both years to one of 293 areas of the country. Estimated utilization, calculated for each specialty, was used as the primary predictor of migration. Physician characteristics (e.g., specialty, age, sex) were obtained from the 2009 file. Area characteristics and other factors influencing physician migration (e.g., rurality, presence of teaching hospital) were obtained from various sources. STUDY DESIGN: We modeled physician location decisions as a two-part process: First, the physician decides whether to move. Second, conditional on moving, a conditional logit model estimates the probability a physician moved to a particular area. Separate models were estimated by specialty and whether the physician was a resident. PRINCIPAL FINDINGS: Results differed between specialties and according to whether the physician was a resident in 2009, indicating heterogeneity in responsiveness to policies. Physician migration was higher between geographically proximate states with higher utilization for that specialty. CONCLUSIONS: Models can be used to estimate specialty-specific migration patterns for more accurate workforce modeling, including simulations to model the effect of policy changes.

A Methodology for Using Workforce Data to Decide Which Specialties and States to Target for Graduate Medical Education Expansion.

OBJECTIVE: To outline a methodology for allocating graduate medical education (GME) training positions based on data from a workforce projection model. DATA SOURCES: Demand for visits is derived from the Medical Expenditure Panel Survey and Census data. Physician supply, retirements, and geographic mobility are estimated using concatenated AMA Masterfiles and ABMS certification data. The number and specialization behaviors of residents are derived from the AAMC's GMETrack survey. DESIGN: We show how the methodology could be used to allocate 3,000 new GME slots over 5 years-15,000 total positions-by state and specialty to address workforce shortages in 2026. EXTRACTION METHODS: We use the model to identify shortages for 19 types of health care services provided by 35 specialties in 50 states. PRINCIPAL FINDINGS: The new GME slots are allocated to nearly all specialties, but nine states and the District of Columbia do not receive any new positions. CONCLUSIONS: This analysis illustrates an objective, evidence-based methodology for allocating GME positions that could be used as the starting point for discussions about GME expansion or redistribution.

Predicting Financial Distress and Closure in Rural Hospitals.

PURPOSE: Annual rates of rural hospital closure have been increasing since 2010, and hospitals that close have poor financial performance relative to those that remain open. This study develops and validates a latent index of financial distress to forecast the probability of financial distress and closure within 2 years for rural hospitals. METHODS: Hospital and community characteristics are used to predict the risk of financial distress 2 years in the future. Financial and community data were drawn for 2,466 rural hospitals from 2000 through 2013. We tested and validated a model predicting a latent index of financial distress (FDI), measured by unprofitability, equity decline, insolvency, and closure. Using the predicted FDI score, hospitals are assigned to high, medium-high, medium-low, and low risk of financial distress for use by practitioners. FINDINGS: The FDI forecasts 8.01% of rural hospitals to be at high risk of financial distress in 2015, 16.3% as mid-high, 46.8% as mid-low, and 28.9% as low risk. The rate of closure for hospitals in the high-risk category is 4 times the rate in the mid-high category and 28 times that in the mid-low category. The ability of the FDI to discriminate hospitals experiencing financial distress is supported by a c-statistic of .74 in a validation sample. CONCLUSION: This methodology offers improved specificity and predictive power relative to existing measures of financial distress applied to rural hospitals. This risk assessment tool may inform programs at the federal, state, and local levels that provide funding or support to rural hospitals.

To What Extent do Community Characteristics Explain Differences in Closure among Financially Distressed Rural Hospitals?

From January 2005 through December 2015, 105 rural hospitals closed. This study examined associations between community characteristics and rural hospital closure. Compared with other rural hospitals that were at high risk of financial distress but remained open over the same time period, closed rural hospitals had a smaller market share (p < .0001) despite being in areas with higher population density (p < .05), were located nearer to another hospital (p < .0001), and were located in markets that had a higher rate of unemployment (p < .05) and a higher percentage of Black (p < .05) and Hispanic (p < .01) residents. These results have three implications for rural health policy: rural hospital closures may disproportionately affect racial and ethnic minorities, community characteristics in combination with other factors make it likely that rural hospital closures will continue, and rural hospital closures illuminate the need for new models of reimbursement and health care delivery to meet the needs of rural communities.