in Small Towns and Rural America: The Role of Medicaid Expansion Georgetown University Center for Children and Families and the University of North Carolina NC Rural Health Research Program The Georgetown University Center for Children and Families (CCF) is an independent, nonpartisan policy and research center founded in 2005 with a mission to expand and improve high quality, affordable health coverage for America’s children and families. CCF is part of the Health Policy Institute at the McCourt School of Public Policy. Visit http://ccf.georgetown.edu/. The North Carolina Rural Health Research Program (NC RHRP) at the Cecil G. Sheps Center for Health Services Research is built upon a 44 year history of rural health research at The University of North Carolina. The program seeks to address problems in rural health care delivery through basic research, policy-relevant analyses, geographic and graphical presentation of data, and the dissemination of information to organizations and individuals who can use the information for policy or administrative purposes to address complex social issues affecting rural populations. Visit http://www.shepscenter.unc. edu/programs-projects/rural-health/. This paper was written jointly by Joan Alker of Georgetown University Center for Children and Families, Jack Hoadley of Georgetown University Health Policy Institute, and Mark Holmes of the North Carolina Rural Health Research Program. The authors would like to thank the Pritzker Children’s Initiative for its support in pursuing this research. Design and layout provided by Nancy Magill. For more information on CCF’s Rural Health Policy Project, visit http://ccf.georgetown.edu/topic/rural-health/. HEALTH INSURANCE COVERAGE IN SMALL TOWNS AND RURAL AMERICA September 2018 Health Insurance Coverage in Small Towns and Rural America: The Role of Medicaid Expansion By Jack Hoadley, Joan Alker, and Mark Holmes Key Findings Introduction zz The uninsured rate for low-income adult citizens (below 138 percent FPL) has come down since 2008/09 in nearly all states, but small towns and rural areas of states that have expanded Medicaid have seen the sharpest declines. The uninsured rate for this population dropped sharply from 35 percent to 16 percent in rural areas and small towns of Medicaid expansion states compared to a decline from 38 percent to 32 percent in non-expansion states between 2008/09 and 2015/16. zz States that experienced the biggest drop in uninsured rates for low-income adults living in small towns and rural areas are Arkansas, Colorado, Connecticut, Hawaii, Kentucky, Michigan, Nevada, New Mexico, Oregon, and West Virginia. zz Non-expansion states with the highest rate of uninsured low-income adults in small towns and rural areas are South Dakota, Georgia, Oklahoma, Florida, Texas, Alabama, Missouri, and Mississippi. Two states that more recently made decisions to expand Medicaid—Alaska and Louisiana—are also among the states with the highest uninsured rates for lowincome adults in non-metro areas. zz The non-expansion states with the biggest coverage disparities between rural areas and small towns and metro areas are Virginia (which recently decided to expand Medicaid), Utah (which will vote this fall on a Medicaid ballot initiative), Florida, and Missouri. The experience in expansion states demonstrates the great opportunity for these states to bring down the uninsured rate in small towns and rural areas and narrow the gap between metro and rural areas. September 2018 Medicaid has been a key factor in lowering the percentage of Americans who lack health insurance. Nationally, the uninsured rate for all Americans under the age of 65 (adults and children) fell dramatically between 2010 and 2016 from 18.2 percent to 10.4 percent, rising slightly to 10.7 percent in 2017.1 Expansion of Medicaid coverage and the new availability of subsidized private insurance from the health care marketplaces helped drive down the uninsured rate, in turn strengthening the health care providers who treat these individuals. In small towns and rural areas, the uninsured rate remains higher than in metropolitan areas. In a previous report, we highlighted how Medicaid offers a vital source of health coverage nationwide, but it plays an even more pronounced role in small towns and rural areas.2 We found that Medicaid covers a larger share of nonelderly adults and children in rural and small-town areas than in metropolitan areas; this trend is strongest among children. These differences result in part from demographic and economic factors that characterize small towns and rural areas. For example, rural areas tend to have lower household incomes, lower rates of workforce participation, and higher rates of disability–all factors associated with Medicaid eligibility.3 State decisions around their Medicaid programs have resulted in uneven patterns of insurance coverage from state to state. Over the same time period (2010-2017) cited above, the national uninsured rate for children and nonelderly adults in expansion states fell from 16.4 percent to 7.6 percent. The rate in non-expansion states fell less significantly—from 20.3 percent to 15.7 percent.4 HEALTH INSURANCE COVERAGE IN SMALL TOWNS AND RURAL AMERICA 1 States that have not expanded Medicaid coverage to adults below 138 percent of the federal poverty line, regardless of their age and family circumstances, have many more uninsured adults. About 2.2 million poor adults live in non-expansion states and are not eligible for either Medicaid or subsidized Marketplace coverage under current law. Another 1.5 million adults in these states are eligible for subsidized Marketplace coverage but would find Medicaid more affordable.5 This paper examines the status of insurance coverage for low-income citizen adults in the 46 states with significant rural populations.6 Nationally, 14 percent of the U.S. nonelderly population resides in small towns and rural areas. Of that, about 6 percent are in rural (“noncore”) counties and 8 percent are in small-town (“micropolitan”) counties. In 16 states, the share of the nonelderly population that lives in small towns and rural areas comprises one-third or more of the population.7 Prior to the enactment of the ACA, Medicaid coverage for adults was mostly limited to very low-income parents, pregnant women, or those with a qualifying disability. States not accepting the option for Medicaid expansion in general have no eligibility for childless adults who are not disabled, and mandatory coverage levels for parents are very low—generally below 50 percent of the poverty level.9 These significant inequities in adults’ Medicaid income eligibility nationwide lead to disparities in the rate of uninsured adults. Using data from the Census Bureau’s American Community Survey public use micro sample, this report examines uninsured rates at the county-level by age in 2008/09 and 2015/16.8 For most tables in this report, county-level data are aggregated to the state level. The county-level estimates used here are unique because they are two-year data, rather than the most recent fiveyear data (2012-2016) available from the Census Bureau. This distinction is important because the Affordable Care Act (ACA) was largely implemented in 2014, and thus the time periods analyzed here allow for an examination of the law’s effects in small towns and rural areas. Although county-level estimates are available for similar populations (e.g., the Small Area Health Insurance Estimates by the U.S. Census Bureau), we specifically wanted to estimate uninsured rates only for citizens. 2 HEALTH INSURANCE COVERAGE IN SMALL TOWNS AND RURAL AMERICA September 2018 Low-Income Adults Are More Likely to Have Insurance Coverage in Medicaid Expansion States with Larger Differences in Small Towns and Rural Areas As described above, Medicaid eligibility for non-pregnant adults was limited prior to enactment of the ACA. All states are required to cover some parents but eligibility levels are very low.10 Furthermore, many states provided little or no eligibility for adults without a dependent child, regardless of income, unless they qualified based on a disability. States that expanded Medicaid as a result of the ACA offer coverage to more adult citizens. As of January 1, 2014, when generous federal funding first became available for expanded Medicaid coverage, 24 states and the District of Columbia had implemented the new Medicaid eligibility levels.11 Another seven states have implemented expanded Medicaid since that date—meaning that the full effect is not reflected in the data used for this report. Another two (Maine and Virginia) have made decisions to expand but have not yet implemented those decisions. The impact of Medicaid expansion is dramatic. On average, the uninsured rate for adult citizens up to 138 percent of FPL was 13 percent in the states that expanded Medicaid by the end of 2014. By contrast, the rate was more than twice as high (27 percent) in non-expansion states (Figure 1). The difference is similarly great in small towns and rural areas: 16 percent uninsured in these areas of expansion states versus 32 percent in the non-expansion states. In the states that have expanded Medicaid, the uninsured rate in small towns and rural areas has fallen to a level that comes closer to that in metro areas (16 percent versus 12 percent). Figure 1: Percent of low-income citizen adults who are uninsured by expansion status, 2015/16 Expansion Status of States Number of States Uninsured Citizen Adults with Incomes to 138 Percent FPL All Areas Metro Non-metro 22 13% 12% 16% Late Expanders 5 23% 22% 30% No Expansion 19 27% 26% 32% All states 46 20% 18% 26% Yes, by end of 2014 Note: States with few or no non-metro counties are excluded (DC, DE, MA, NJ, RI). States that expanded Medicaid effective in 2014 are categorized as “yes, by end of 2014.” States that expanded Medicaid between January 1, 2015, and December 31, 2016, are categorized as “Late Expanders.” Two states that have made decisions to expand Medicaid but where enrollment has not begun (ME, VA) are categorized as “no expansion.” For a list of states in each category, see the methodology. September 2018 HEALTH INSURANCE COVERAGE IN SMALL TOWNS AND RURAL AMERICA 3 Eight non-expansion states have more than one-third of their low-income adults uninsured in their small towns and rural areas. The states with the highest rate of uninsured adults in rural areas are South Dakota, Georgia, Oklahoma, Florida, Texas, Alabama, Missouri, and Mississippi (Figure 2). This means that these states have significant room to improve coverage for low-income adults and strengthen the providers and hospitals that serve rural areas and small towns. Figure 2: Share of low-income uninsured citizen adults in rural and metro areas in non-expansion states, 2015/16 Non-metro adults to 138% FPL uninsured, 2015/16 (percent) Metro adults to 138% FPL uninsured, 2015/16 (percent) South Dakota 47 41 Georgia 38 30 Oklahoma 38 32 Florida 37 24 Texas 36 29 Alabama 36 29 Missouri 35 26 Mississippi 35 33 South Carolina 32 27 Utah 31 20 Virginia* 31 21 North Carolina 29 25 Tennessee 29 25 Wyoming 28 29 Idaho 28 31 Nebraska 24 19 Kansas 24 25 Maine* 23 19 Wisconsin 18 13 States without Medicaid expansion *Maine and Virginia have made decisions to expand, but enrollment has not yet begun. In most non-expansion states there is a substantial gap in the uninsured rate, with a greater share of the low-income adult population lacking insurance in small towns and rural areas compared to those in metropolitan areas. Florida, Missouri, Utah, and Virginia have gaps of 10 percentage points or more between these rates in non-metro versus metro areas (Figure 3). Low-income adults in the rural areas and small towns of these states would likely see sharp improvements in their ability to obtain insurance coverage if state officials expanded Medicaid. 4 HEALTH INSURANCE COVERAGE IN SMALL TOWNS AND RURAL AMERICA September 2018 Figure 3: States with the largest rural-metro gaps in uninsured rates for low-income adults 37 35 Non-metro 31 24 31 20 Florida Metro 26 Missouri Utah 21 Virginia Most states that have expanded their Medicaid programs now have substantially lower uninsured rates (Figure 4). In all the original expansion states, the uninsured rate for low-income adults in small towns and rural areas is below onefourth of the population, and in four states that rate is 10 percent or lower. Figure 4: Share of low-income uninsured citizen adults in rural areas and metro areas of expansion states, 2015/16 Expansion States Non-metro adults to 138% FPL uninsured, 2015/16 (percent) Metro adults to 138% FPL uninsured, 2015/16 (percent) Hawaii 9 9 Connecticut 9 11 Vermont Maryland Illinois Minnesota Kentucky Washington Colorado Nevada West Virginia New York Iowa California Michigan Oregon Ohio New Hampshire New Mexico Arkansas Arizona North Dakota 10 10 12 13 13 13 13 14 14 14 15 15 16 17 18 20 21 22 23 24 3 12 12 9 13 11 11 17 13 10 13 11 13 13 15 19 15 21 18 17 *The five states that expanded Medicaid through the ACA after December 31, 2014, are excluded from this table (Alaska, Indiana, Louisiana, Montana, and Pennsylvania). September 2018 HEALTH INSURANCE COVERAGE IN SMALL TOWNS AND RURAL AMERICA 5 Furthermore, most expansion states have no gap in the level of uninsured adults between non-metro and metro areas. In half (11 of 22) of the states that expanded Medicaid by the end of 2014, the uninsured rate in the nonmetro areas is lower than in metro areas or no more than two percentage points higher (Arkansas, Colorado, Connecticut, Hawaii, Illinois, Kentucky, Maryland, Nevada, New Hampshire, Washington, and West Virginia). These states have experienced the greatest success in eliminating the metro-nonmetro disparity in insurance coverage. Furthermore, most expansion states have no gap in the level of uninsured adults between nonmetro and metro areas. Three states—Arkansas, Kentucky, and West Virginia—where expansion has dramatically lowered uninsured rates for all low-income citizen adults have no rural-metro gap. The experience in expansion states demonstrates the potential for states that have not yet expanded Medicaid. Not only do they have the chance to reduce the number of uninsured adults overall, but they have a significant opportunity to bring down the uninsured rate in small towns and rural areas that currently have more uninsured adults and narrow the gap between metro and rural areas. Medicaid Expansion States Experienced Large Declines in Uninsured Rates, Especially in Small Towns and Rural Areas The impact of Medicaid expansion can be examined by comparing the uninsured rates for low-income citizen adults before and after the implementation of the ACA—from 2008/09 to 2015/16. In nearly all states, uninsured rates have come down. Across all states with a substantial number of small towns and rural areas, the average decline in the uninsured rate for this adult population below 138 percent of the poverty line was 14 percentage points, dropping from 34 percent to 20 percent uninsured. The drop was a little less in non-metro counties (11 percentage points) than in metro counties (15 percentage points). (Figure 5). Not surprisingly, the decline in the uninsured rate was much greater in expansion states, where the rate fell from 30 percent to 13 percent (18 percentage points) than in non-expansion states where it fell from 38 percent to 27 percent (11 percentage points). Figure 5: Decline in uninsured rate for low-income citizen adults in all areas and in non-metro counties, by expansion status, 2008/09 to 2015/16 (percentage points) Expansion Status of States Yes, by end of 2014 Late Expanders No Expansion All states All Areas Non-metro Areas Metro Areas Percent Uninsured, 2008/09 Percent Uninsured, 2015/16 Decline in Uninsured (pct. pts.) Percent Uninsured, 2008/09 Percent Uninsured, 2015/16 Decline in Uninsured (pct. pts.) Percent Uninsured, 2008/09 Percent Uninsured, 2015/16 Decline in Uninsured (pct. pts.) 30 13 18 35 16 19 29 12 17 37 23 13 41 30 11 35 22 14 38 27 11 38 32 5 38 26 13 34 20 14 37 26 11 33 18 15 Note: States with few or no non-metro counties are excluded. States that expanded Medicaid effective in 2014 are categorized as “yes, by end of 2014.” States that expanded Medicaid between January 1, 2015, and December 31, 2016, are categorized as “Late Expanders.” Two states that have made decisions to expand Medicaid but where enrollment has not begun (ME, VA) are categorized as “no expansion.” For a list of states in each category, see the methodology. Differences may vary due to rounding. 6 HEALTH INSURANCE COVERAGE IN SMALL TOWNS AND RURAL AMERICA September 2018 The contrast was especially striking in small towns and rural areas: a decline of 19 percentage points in expansion states versus just 5 points in non-expansion states (Figure 6). In expansion states, the non-metro uninsured rate for low-income citizen adults fell from 35 percent to 16 percent between 2008/09 to 2015/16. In the non-expansion states, the decline was from 38 percent to 32 percent. This result underscores that the Medicaid expansion has been the key driver as the ACA was implemented in reducing the number of uninsured adults in rural areas and small towns nationwide. Figure 6. Decline in uninsured rate for low-income citizen adults, by expansion status, 2008/09 to 2015/16 38 38 35 32 29 26 16 12 Non-metro areas, expansion states Non-metro areas, non-expansion states 2008/09 pecent uninsured Metro areas, expansion states Metro areas, non-expansion states 2015/16 pecent uninsured Note: States with few or no non-metro counties and “late expander” states are excluded. For a list of states in each category, see the methodology. A look at individual states shows how dramatic the change has been in the small towns and rural areas of these states. In 10 expansion states (Colorado, Nevada, Kentucky, Oregon, New Mexico, Arkansas, Connecticut, Hawaii, Michigan, and West Virginia), the drop in the uninsured rate for low-income citizen adults in these areas has been 20 percentage points or greater (Figure 7). None of the non-expansion states has experienced a drop in the uninsured rate of this magnitude. September 2018 HEALTH INSURANCE COVERAGE IN SMALL TOWNS AND RURAL AMERICA 7 Figure 7: Decline in uninsured rate for low-income citizen adults in non-metro counties, by state and expansion status, 2008/09 to 2015/16 (percentage points) Expansion State 2008/09 2015/16 Drop Colorado 42% 13% 29 Nevada 42% 14% Kentucky 40% Oregon Non-Expansion State 2008/09 2015/16 Drop Wyoming 47% 28% 19 28 Florida 53% 37% 16 13% 27 Nebraska 39% 24% 15 43% 17% 27 Idaho 38% 28% 10 New Mexico 46% 21% 25 Oklahoma 47% 38% 9 Arkansas 45% 22% 23 Wisconsin 27% 18% 9 Connecticut 32% 9% 23 North Carolina 35% 29% 7 Hawaii 31% 9% 22 Kansas 30% 24% 6 Michigan 38% 16% 22 Tennessee 35% 29% 6 West Virginia 35% 14% 21 South Carolina 38% 32% 5 Maryland 29% 10% 18 Mississippi 39% 35% 5 Washington 31% 13% 18 Georgia 43% 38% 4 Ohio 35% 18% 18 Texas 40% 36% 4 Illinois 29% 12% 17 Utah 34% 31% 3 New Hampshire 36% 20% 17 Virginia 33% 31% 2 California 30% 15% 15 Missouri 35% 35% 1 Vermont 22% 10% 12 Maine 22% 23% -1 Iowa 27% 15% 12 Alabama 35% 36% -1 Minnesota 24% 13% 11 South Dakota 37% 47% -10 New York 24% 14% 10 North Dakota 32% 24% 9 Arizona 31% 23% 8 Note: States with few or no non-metro counties and states that expanded Medicaid between January 1, 2015, and December 31, 2016, are excluded. States that expanded Medicaid effective in 2014 are categorized as “expansion states.” Two states that have made decisions to expand Medicaid but where enrollment has not begun (ME, VA) are categorized as “non-expansion” states. Differences may vary due to rounding. 8 HEALTH INSURANCE COVERAGE IN SMALL TOWNS AND RURAL AMERICA September 2018 Parents Represent About One-Fourth of Uninsured Low-Income Adults About one-fourth of all remaining uninsured citizen adults are parents. This uninsured share is similar in both small towns and rural areas (24 percent) and in metro counties (27 percent). Parents include adults with a child in the household and so may include grandparents or other adult caretakers. When parents have insurance coverage, children benefit as well. Children with insured parents are more likely to be insured themselves.12 Children are also more likely to get preventive care, and the entire family is assured of the financial protection that Medicaid offers from medical debt and bankruptcy.13 As discussed above, some parents have been eligible for Medicaid coverage prior to the ACA, although only at very low income levels. Furthermore, eligible adults without children are less like to enroll in Medicaid than eligible parents, and both are less likely to enroll than eligible children.14 These factors help explain why more of the uninsured adults have no children. In addition, parents may be more aware of insurance options when their children are covered by Medicaid or CHIP and their state has made outreach efforts to get families enrolled. There is a soft dividing line between the categories of parents and “childless” adults. Those classified as childless adults may include parents whose children are no longer in the household because they are older or are currently living with a different family member. They may be fathers no longer living in the household that includes their children. Still others are younger women of childbearing age who may become parents in the near future. A recent study of women in Ohio after Medicaid was expanded found higher rates of prenatal vitamin use and recommended prenatal screenings.15 This finding underscores the importance of expansion in covering women before they qualify for pregnancy related Medicaid. Case Studies: Virginia, Florida, Utah, and Missouri Four states—Virginia, Florida, Utah, and Missouri—have especially wide gaps in uninsured rates between the non-metro and metro counties. Virginia Virginia acted in 2018 to expand its Medicaid program, and enrollment is expected to start on January 1, 2019. As Figure 8 shows, rural areas and small towns in Virginia have more room to gain from the decision to expand than Virginia’s metropolitan counties. Although statewide, the uninsured rate for low-income adults came down from 34 percent to 24 percent since the ACA was implemented, the uninsured rate for adults below 138 percent of the poverty line in Virginia’s small towns and rural areas remained considerably higher (31 percent) than in its metropolitan cities and counties (21 percent). Nine of the 10 nonmetro counties or cities with the largest number of low-income uninsured adults have uninsured rates of at least 30 percent. By contrast, nine of the 10 metro counties or cities with the largest number of low-income uninsured adults have rates no higher than 24 percent. Uninsured rates for adults should come down considerably statewide once expansion is implemented. September 2018 HEALTH INSURANCE COVERAGE IN SMALL TOWNS AND RURAL AMERICA 9 Figure 8: Virginia Counties and Cities with the Most Uninsured Citizen Adults Under 138% FPL Non-metro counties or cities with the most uninsured adults Adults to 138% FPL uninsured, 2015/16 (percent) Metro counties or cities with the most uninsured adults Adults to 138% FPL uninsured, 2015/16 (percent) All non-metro counties and cities 31 All metro counties and cities 21 Pittsylvania County 30 Fairfax County 15 Henry County 30 Virginia Beach City 22 Wise County 34 Norfolk City 22 Danville City 31 Richmond City 24 Tazewell County 34 Henrico County 20 Buchanan County 37 Prince William County 21 Lee County 36 Chesapeake City 20 Halifax County 31 Newport News City 20 Carroll County 28 Hampton City 26 Russell County 34 Chesterfield County 18 If Virginia can achieve the same results as in the neighboring state of Kentucky (see text box on page 12), which expanded Medicaid in the first year permitted under the ACA, Virginia could achieve similarly dramatic results for those living in small towns and rural areas. The experience in Kentucky and many other expansion states suggest that the uninsured gap between those living in metro areas compared to those in rural areas and small towns should be reduced or eliminated. Florida Florida is another state where our data show that Medicaid expansion would have a large effect on uninsured rates for adults statewide and a substantially disproportionate benefit for rural areas and small towns. In Florida the uninsured rate for low-income adults in rural counties is 37 percent as opposed to 24 percent in metro counties of the state (Figure 9). In several rural counties, uninsured rates for low-income adults are 40 percent or greater. Figure 9: Florida Counties with the Most Uninsured Citizen Adults Under 138% FPL Non-metro counties with the most uninsured adults Adults to 138% FPL uninsured, 2015/16 (percent) Metro counties with the most uninsured adults Adults to 138% FPL uninsured, 2015/16 (percent) All non-metro counties 37 All metro counties 24 Putnam 37 Miami-Dade 22 Columbia 40 Broward 22 Jackson 42 Orange 25 Suwannee 34 Hillsborough 22 Okeechobee 36 Palm Beach 20 10 HEALTH INSURANCE COVERAGE IN SMALL TOWNS AND RURAL AMERICA September 2018 Utah Utah is actively considering expansion with a ballot initiative scheduled this year. The five non-metro counties with the largest number of low-income uninsured adults, have uninsured rates of at least 30 percent, whereas none of the metro counties with the most uninsured adults reaches this level (Figure 10). Figure 10: Utah Counties with the Most Uninsured Citizen Adults Under 138% FPL Non-metro counties with the most uninsured adults Adults to 138% FPL uninsured, 2015/16 (percent) Metro counties with the most uninsured adults Adults to 138% FPL uninsured, 2015/16 (percent) All non-metro counties 31 All metro counties 20 Iron 30 Salt Lake 22 Uintah 35 Utah 17 Sanpete 30 Davis 13 Sevier 31 Weber 24 San Juan 42 Washington 26 Missouri Missouri also has a much higher uninsured rate in non-metro counties (35 percent) than in metro counties (26 percent) (Figure 11). Like Virginia, there was some success in achieving lower uninsured rates for low-income citizen adults statewide as a result of the implementation of the ACA’s Marketplace subsidies—a drop from 39 percent to 29 percent. But that gain was not seen in small towns and rural areas, where the uninsured rate was virtually unchanged at 35 percent. In other words, rural areas and small towns in Missouri will only see significant coverage gains from the ACA if the state chooses to expand Medicaid. Figure 11: Missouri Counties and Cities with the Most Uninsured Citizen Adults Under 138% FPL Non-metro counties with the most uninsured adults Adults to 138% FPL uninsured, 2015/16 (percent) Metro counties or cities with the most uninsured adults Adults to 138% FPL uninsured, 2015/16 (percent) All non-metro counties and cities 35 All metro counties and cities 26 Taney County 39 St. Louis County 20 Butler County 41 Jackson County 27 St. Francois County 37 St. Louis City 28 Howell County 37 Greene County 27 Dunkirk County 35 Jefferson County 32 September 2018 HEALTH INSURANCE COVERAGE IN SMALL TOWNS AND RURAL AMERICA 11 Success of Medicaid Expansion in Kentucky for Rural Areas Kentucky was an early adopter of Medicaid expansion with considerable success in dramatically reducing the uninsured rate for low-income adults from 43 percent in 2008/09 statewide to 13 percent in 2015/16. The low level was particularly striking in small towns and rural areas of the state. Figure 12. Kentucky Counties with the Most Uninsured Citizen Adults Under 138% FPL Non-metro counties with the most uninsured adults All nonmetro counties Adults to 138% FPL uninsured, 2015-2016 (percent) Metro counties with the most uninsured adults Adults to 138% FPL uninsured, 2015-2016 (percent) 13 All metro counties 13 Pike 9 Jefferson 10 Pulaski 14 Fayette 14 Madison 6 Kenton 16 Floyd 9 Warren 11 Laurel 13 Daviess 10 Harlan 11 Christian 18 Knox 12 Hardin 17 McCreary 16 Campbell 12 Clay 14 Boone 20 Bell 12 Boyd 14 Conclusion States that have expanded Medicaid under the ACA have seen broad gains in insurance coverage for low-income adults. Rural areas and small towns have seen disproportionate benefits with little or no disparity in coverage rates between metro and rural areas. Medicaid expansion makes a difference for many reasons, including the likelihood that Marketplace coverage is more challenging to sell in rural areas and small towns,16 and rural areas have higher rates of poverty on average. Increased insurance coverage in turn benefits the clinics, hospitals, and other providers that operate in these states, especially in rural communities. An earlier study we conducted found significant differences between providers operating in states that opted to expand Medicaid and those that did not.17 The benefits of Medicaid expansion have been experienced beyond the walls of health care facilities such as clinics and hospitals with positive ripple effects throughout the communities they serve. Community health centers that serve patients in Medicaid expansion states have experienced a 11 percentage point decline in their share of uninsured patients and a 13 percentage point increase in Medicaid patients.18 Clinics in rural areas of expansion states also have experienced increases on quality measures such as asthma treatment and hypertension control and provided more patient visits in areas such as mammograms and substance abuse disorders. These gains at rural clinics were not duplicated in urban clinics in those same states, perhaps because patients in urban areas have more access to providers other than these clinics. Another recent study found that Medicaid expansion contributed to a rosier financial picture for hospitals and less likelihood of hospital closures, especially those in rural areas.19 Notably, the study found higher rates of rural hospital closures in states that failed to expand Medicaid. The revenue hospitals receive from Medicaid when patients have this source of coverage improves 12 HEALTH INSURANCE COVERAGE IN SMALL TOWNS AND RURAL AMERICA September 2018 the bottom line and keeps them open. Keeping a rural community hospital open means that care is available to all residents of the community on a timelier basis and maintains a major employer for the community.20 Our study found that those states that have not yet expanded Medicaid have some of the largest gaps in uninsured rates with rural areas and small towns having substantially higher rates of uninsured low-income adults. Many states that have not yet expanded have sizable rural populations (including Idaho, Mississippi, Nebraska, Oklahoma, South Dakota and Wyoming). Improved coverage rates typically translate to a more stable health care system, and in particular help rural areas and small towns maintain the availability of health care providers in areas where shortages are all too common. Methodology This report relies primarily on data from the Census Bureau’s American Community Survey (ACS) public use micro sample to calculate uninsured rates at the county level for nonelderly adults (age 19 to 64) for 2016. We restrict the sample of adults in two ways. First, we study only those individuals with incomes below 138 percent of the federal poverty guidelines. Second, we study only adults who are citizens, including naturalized citizens. For one section of the analysis, we differentiate low-income citizen adults by whether they have children under age 18 who live in the household. Similar data for both children and nonelderly adults who were covered by Medicaid or who were uninsured in 2009 and 2015 were calculated for an earlier report.21 Although the estimates are for 2016, we use data from 2015 and 2016 in order to increase the precision of the estimates, with totals normed to 2016 levels. The two-year time frames used in this report provide a different perspective compared to the single-year ACS summary estimates. Those are available at the national and state levels, as well as for a selection of counties. The five-year ACS summary estimates are available for all counties in the United States. However, these data are from 2012 through 2016, whereas the analytical approach in this report provides us with more recent complete county-level data for 2015 through 2016. State tables shown in the report are aggregated from the county estimates. Method for Estimating the Number of Uninsured Individuals per County Annual, county-level numbers of uninsured citizens do not exist in a consistent manner across all years and states. Thus, we developed synthetic estimates using the Public Use Microdata Sample (PUMS) of the ACS to estimate annual, county-level estimates for each of three age groups September 2018 using a three-step approach to calculate.22 Effectively, the approach takes the statewide estimated number of uninsured adults, using the insurance coverage variable (HICOV) to define whether the respondent had insurance coverage. The approach then allocates them across counties according to the degree to which the county’s demographics make them likely to be uninsured. Step 1: Modeling individual probabilities First, we used the PUMS to model factors associated with an individual’s probability of being uninsured. We pooled data for 2015/16, adding an indicator for whether the observation was from 2016. An individual was identified as being “enrolled” if they indicated they were uninsured. We estimated a separate linear probability model for each state and the District of Columbia, age category (0 to 18, 19 to 64, 65 or older), for a total of 51 states x 3 age categories for 153 models. We estimated the probability an individual was uninsured as a function of 18 age indicators (five year increments: 0-4, 5-9, continuing through 80-84, and 85 or more), sex, age interacted with sex, 14 race/ethnicity categories (Hispanic status crossed with race, including “other” and “two or more races”), 5 income categories (under 50, 50-99 percent FPL, 100-149 percent FPL, 150-199 percent FPL, 200 percent FPL), family status (marriage status interacted with whether there are children in the household), disability interacted with income category, indicators for whether the individual was born in the United States or was a naturalized citizen, and indicators for the Public Use Microdata Area (PUMA) of the respondent. For adults, labor force status (industry of employment, unemployed, or not in labor force) was also included. Sampling weights were used to ensure the sample was representative of the state population. A separate analysis was done to calculate uninsured rates for 2008/09. HEALTH INSURANCE COVERAGE IN SMALL TOWNS AND RURAL AMERICA 13 Step 2: Developing Small Area Estimates We collected county-level data on corresponding characteristics from the ACS summary data. For example, for each county we calculated the proportion working in each industry, the age/ income profile, and the age/sex/nativity profile. Usually, these data were pulled from the five-year (2012-2016) estimates published through American FactFinder. Using the Missouri Master Area Block Level Equivalency (MABLE) data engine provided by the Missouri Census Data Center,23 we developed crosswalks from county to PUMA so the PUMA of the ACS PUMS could be used to generate county-specific estimates that could be allocated to PUMAs. For example, if 60 percent of the population of a county was in PUMA 101, and 40 percent was in PUMA 102, the PUMA indicators from the PUMS models would have .6 for PUMA 101 and .4 for PUMA 102, with 0 for the rest of the PUMA indicators (counties spanning multiple PUMAs were allocated proportionally by 2010 Census population). Thus, we generate a county-level dataset of the population in each county in the state. These data were then used with the parameter estimates from Step 1 to develop the average probability in the county of being uninsured. This probability, multiplied by the county population in the age group, served as the initial estimate of the number of uninsured individuals in the county. Step 3: Raking Estimates The sum of the county estimates aggregated to the state may differ from the direct state estimates in ACS. Therefore, the county estimates were adjusted (raked) to ensure the sum of the county estimates in a state equals the estimated state total.24 For example, if the number of uninsured summed across counties was 100 but the state estimate was 110, each county estimate was increased by 10 percent as long as the county’s uninsured count did not exceed its total population. The number of uninsured in the second year of the two-year time period (i.e. 2016) was used as the “target” for each state/age group/period; this approach trades off the increased precision and sample size from the two-year time period against the accuracy from using the second year only. For example, the number of enrollees in 2016 may be considerably higher than in 2015 due to reduced uninsured rates resulting from Medicaid expansion and other changes initiated by the Affordable Care Act. This approach ensures the county-level estimates aggregate to the state estimates. Estimating Income Levels for Adults From the ACS, we know (for each county) the number of families by family type by ratio of income to poverty.25 For example, table B17022 indicates the number of families with no children with income below 130 percent of poverty and between 130 and 149 percent of poverty. Table B17025 indicates the number of citizens below 100 percent of poverty. We triangulate among these data, using the ACS microdata, to estimate the number of citizens with incomes below 138 percent of poverty using an approach similar to that for estimating the number of uninsured. Classifying Counties as Small Towns and Rural Areas In this report, we classify counties as metropolitan and nonmetropolitan. The latter category combines the Census Bureau categories of micropolitan or small town counties (those with central urban areas of no more than 50,000 people) and noncore or rural counties. We characterize non-metro counties as representing America’s small towns and rural areas. In four states (DC, DE, NJ, RI), no counties are classified as non-metro and are thus excluded from this report. In addition, we exclude Massachusetts, where the total non-metro population is less than 2 percent of the state’s population (only 100,000 people). The limitation of a county-based definition of small towns and rural areas is that county size and county boundaries vary considerably by state. For example, San Bernardino County, California, has 2 million people and runs from urbanized areas near Los Angeles through deserts and mountains to the Nevada border. Its classification as a metropolitan county thus effectively misclassifies people living in the small town and rural areas of that county. By contrast, states such as Georgia and Kansas have much smaller counties allowing more residents to be accurately classified as metro or non-metro. The Census Bureau also uses another definition of urban and rural; but it is built up from census tract data and thus is not readily amenable to classifying counties.26 One recent report by the Kaiser Family Foundation defines rural counties based on an index of relative rurality, which is based on population size, population density, extent of urbanized area, and distance to the nearest metro area.27 This produces a different classification of the population, which could lead to different findings. Classifying States Based on Medicaid Expansion In this report, states are classified for their Medicaid expansion status based on analysis by the Kaiser Family Foundation.29 States that expanded Medicaid effective by the end of 2014 are categorized as “yes, by end of 2014.” States that expanded Medicaid between January 1, 2015, and December 31, 2016, (Alaska, Indiana, Louisiana, Montana, and Pennsylvania) are categorized as “late expanders.” The Census data for this analysis are based on surveys conducted throughout 2015 and 2016, so the Medicaid expansion in these states was not effective throughout the survey period. Two states that have made decisions to expand Medicaid but where enrollment has not begun (Maine and Virginia) are categorized as “no expansion.” 14 HEALTH INSURANCE COVERAGE IN SMALL TOWNS AND RURAL AMERICA September 2018 Appendix Table 1: Share of uninsured citizen adults with incomes up to 138 percent of FPL in small towns and rural areas and in metro areas, 2015/16 State Expanded United States Alabama Alaska Arizona Arkansas California Colorado Connecticut Florida Georgia Hawaii Idaho Illinois Indiana Iowa Kansas Kentucky Louisiana Maine Maryland Michigan Minnesota Mississippi Missouri Montana Nebraska Nevada New Hampshire New Mexico New York North Carolina North Dakota Ohio Oklahoma Oregon Pennsylvania South Carolina South Dakota Tennessee Texas Utah Vermont Virginia Washington West Virginia Wisconsin Wyoming All areas uninsured rate, 2015/16 (percent) Non-metro uninsured rate, 2015/16 (percent) Metro uninsured rate, 2015/16 (percent) 20 26 18 N Y** Y Y Y Y 31 34 19 21 11 12 36 43 23 22 15 13 29 26 18 21 11 11 Y N 11 25 9 37 11 24 N 32 38 30 Y 9 9 9 N Y Y** 30 12 24 28 12 26 31 12 23 Y 14 15 13 N Y 24 13 24 13 25 13 Y** 33 39 31 N*** Y Y 21 12 14 23 10 16 19 12 13 Y N 10 34 13 35 9 33 N 29 35 26 Y** 33 34 32 N Y Y Y 21 17 19 17 24 14 20 21 19 17 19 15 Y 10 14 10 N Y 26 21 29 24 25 17 Y 16 18 15 N 35 38 32 Y 14 17 13 Y** N N N N 17 28 45 26 30 21 32 47 29 36 16 27 41 25 29 N 21 31 20 Y 8 10 3 N*** Y 24 11 31 13 21 11 Y N 13 14 14 18 13 13 N 28 28 29 See appendix table notes on page 18. September 2018 HEALTH INSURANCE COVERAGE IN SMALL TOWNS AND RURAL AMERICA 15 Appendix Table 2: Change in share of uninsured citizen adults with incomes up to 138 percent of FPL in small towns and rural areas and in metro areas, 2008/09 and 2015/16 Metro counties State United States* Alabama Alaska Arizona Arkansas California Colorado Connecticut Florida Georgia Hawaii Idaho Illinois Indiana Iowa Kansas Kentucky Louisiana Maine Maryland Michigan Minnesota Mississippi Missouri Montana Nebraska Nevada New Hampshire New Mexico New York North Carolina North Dakota Ohio Oklahoma Oregon Pennsylvania South Carolina South Dakota Tennessee Texas Utah Vermont Virginia Washington West Virginia Wisconsin Wyoming Expanded Non-metro counties Uninsured rate (percent) decline (pct points) 2008/09 2015/16 2008/09 to 2015/16 2008/09 2015/16 uninsured rate (percent) decline (pct points) 2008/09 to 2015/16 33 18 15 37 26 11 N Y** Y Y Y Y 44 56 28 47 28 29 29 26 18 21 11 11 15 30 9 26 17 18 35 45 31 45 30 42 36 43 23 22 15 13 -1 2 8 23 15 29 Y N 22 38 11 24 12 14 32 53 9 37 23 16 N 42 30 12 43 38 4 Y 19 9 9 31 9 22 N Y Y** 43 33 40 31 12 23 12 21 17 38 29 39 28 12 26 10 17 13 Y 30 13 17 27 15 12 N Y 38 48 25 13 13 35 30 40 24 13 6 27 Y** 45 31 15 50 39 11 N*** Y Y 29 29 34 19 12 13 10 17 21 22 29 38 23 10 16 -1 18 22 Y N 24 54 9 33 15 21 24 39 13 35 11 5 N 41 26 15 35 35 1 Y** 49 32 17 45 34 10 N Y Y Y 28 37 36 31 19 17 19 15 9 19 17 16 39 42 36 46 24 14 20 21 15 28 17 25 Y 21 10 11 24 14 10 N Y 37 42 25 17 13 24 35 32 29 24 7 9 Y 37 15 22 35 18 18 N Y 46 37 32 13 14 23 47 43 38 17 9 27 Y** N N N N 28 44 50 38 39 16 27 41 25 29 12 16 9 12 10 35 38 37 35 40 21 32 47 29 36 13 5 -10 6 4 N 28 20 8 34 31 3 Y 24 3 21 22 10 12 N*** Y 34 32 21 11 13 21 33 31 31 13 2 18 Y N 42 29 13 13 29 16 35 27 14 18 21 9 N 37 29 8 47 28 19 See appendix table notes on page 18. 16 HEALTH INSURANCE COVERAGE IN SMALL TOWNS AND RURAL AMERICA September 2018 Appendix Table 3: Change in share of uninsured citizen adults with incomes up to 138 percent of FPL in all areas, 2008/09 and 2015/16 All areas State Expanded United States* Alabama Alaska Arizona Arkansas California Colorado Connecticut Florida Georgia Hawaii Idaho Illinois Indiana Iowa Kansas Kentucky Louisiana Maine Maryland Michigan Minnesota Mississippi Missouri Montana Nebraska Nevada New Hampshire New Mexico New York North Carolina North Dakota Ohio Oklahoma Oregon Pennsylvania South Carolina South Dakota Tennessee Texas Utah Vermont Virginia Washington West Virginia Wisconsin Wyoming Uninsured rate (percent) decline (pct points) 2008/09 2015/16 34 20 14 40 51 28 46 28 31 31 34 19 21 11 12 10 17 9 25 17 20 23 39 11 25 12 14 N 42 32 10 Y 21 9 12 N Y Y** 41 33 39 30 12 24 11 20 16 Y 28 14 15 N Y 34 43 24 13 10 Y** N*** 47 33 14 Y Y 26 29 35 21 12 14 4 17 21 Y N 24 44 10 34 14 10 N Y** Y Y Y Y Y N 2008/09 to 2015/16 30 N 39 29 10 Y** 46 33 12 N 33 37 36 37 21 17 19 17 12 20 17 20 21 10 11 Y 37 36 26 21 11 16 Y 37 16 21 N 47 38 35 14 12 24 29 42 42 37 39 17 28 45 26 30 12 14 -3 11 9 N 29 21 8 Y N*** 23 8 15 Y 34 32 24 11 10 21 Y N 39 28 13 14 26 14 N 44 28 15 Y Y Y Y N Y Y** N N N N See appendix table notes on page 18. September 2018 HEALTH INSURANCE COVERAGE IN SMALL TOWNS AND RURAL AMERICA 17 Appendix Table Notes Table 1: * Delaware, New Jersey, Rhode Island, and the District of Columbia were excluded from the analysis because they have no non-metro counties. Massachusetts was excluded because less than 2 percent of its population resides in non-metro counties. These states and D.C. are also excluded from the U.S. total. ** Five states that expanded Medicaid after December 31, 2014, (Alaska, Indiana, Louisiana, Montana, and Pennsylvania) are treated separately since much of the data collection occurred before expansion was effective. *** Maine and Virginia have made decisions to expand, but enrollment has not begun. Table 2: * Delaware, New Jersey, Rhode Island, and the District of Columbia were excluded from the analysis because they have no non-metro counties. Massachusetts was excluded because less than 2 percent of its population resides in non-metro counties. These states and D.C. are also excluded from the U.S. total. ** Five states that expanded Medicaid after December 31, 2014, (Alaska, Indiana, Louisiana, Montana, and Pennsylvania) are treated separately since much of the data collection occurred before expansion was effective. *** Maine and Virginia have made decisions to expand, but enrollment has not begun. Note: Differences may vary due to rounding. Table 3: * Delaware, New Jersey, Rhode Island, and the District of Columbia were excluded from the analysis because they have no non-metro counties. Massachusetts was excluded because less than 2 percent of its population resides in non-metro counties. These states and D.C. are also excluded from the U.S. total. ** Five states that expanded Medicaid after December 31, 2014, (Alaska, Indiana, Louisiana, Montana, and Pennsylvania) are treated separately since much of the data collection occurred before expansion was effective. *** Maine and Virginia have made decisions to expand, but enrollment has not begun. Note: Differences may vary due to rounding. 18 HEALTH INSURANCE COVERAGE IN SMALL TOWNS AND RURAL AMERICA September 2018 Endnotes These data measure insurance status at the time of the survey interview; uninsured rates are lower if calculated as how many are uninsured for an entire year (13.3 percent in 2010 to 6.3 percent in 2017). R. Cohen, E. Zammitti, and M. Martinez, “Health Insurance Coverage: Early Release of Estimates from the National Health Interview Survey, 2017” (Washington: Centers for Disease Control and Prevention: National Center for Health Statistics, May 2018), available at https://www.cdc. gov/nchs/data/nhis/earlyrelease/insur201805.pdf. 1 J. Hoadley et al., “Medicaid in Small Towns and Rural America: A Lifeline for Children, Families, and Communities” (Washington: Georgetown University Center for Children and Families and North Carolina Rural Health Research Program, June 2017), available at https://ccf.georgetown. edu/2017/06/06/rural-health-report/. 2 J. Foutz, S. Artiga, and R. Garfield, “The Role of Medicaid in Rural America” (Washington: Kaiser Family Foundation, April 25, 2017), available at https://www.kff.org/medicaid/issuebrief/the-role-of-medicaid-in-rural-america/. 3 R. Cohen, E. Zammitti, and M. Martinez, “Health Insurance Coverage: Early Release of Estimates from the National Health Interview Survey, 2017” (Washington: Centers for Disease Control and Prevention: National Center for Health Statistics, May 2018), available at https://www.cdc.gov/nchs/data/nhis/ earlyrelease/insur201805.pdf. 4 R. Garfield, A. Damico, and K. Orgera, “The Coverage Gap: Uninsured Poor Adults in States that Do Not Expand Medicaid” (Washington: Kaiser Family Foundation, June 12, 2018), available at https://www.kff.org/medicaid/issue-brief/thecoverage-gap-uninsured-poor-adults-in-states-that-do-notexpand-medicaid/. 5 Delaware, New Jersey, Rhode Island, and the District of Columbia were excluded from the analysis because they have no micropolitan or noncore counties. Massachusetts was excluded because less than 2 percent of its population resides in counties that are micropolitan or noncore. 6 These states are: Arkansas, Idaho, Iowa, Kentucky, Maine, Mississippi, Montana, Nebraska, New Hampshire, New Mexico, North Dakota, Oklahoma, South Dakota, Vermont, West Virginia, and Wyoming. 7 See the Methodology section for a full description of the analytic approach in this report. Unlike the counts in Figure 1, these counts include the five states with few or no non-metro counties. All these states expanded Medicaid as of January 1, 2014. 11 M. Karpman and G. Kenney, “QuickTake: Health Insurance Coverage for Children and Parents: Changes between 2013 and 2017” (Washington: The Urban Institute, September 2017), available at http://hrms.urban.org/quicktakes/health-insurancecoverage-children-parents-march-2017.html; J. Hudson and A. Moriya, “Medicaid Expansion for Adults Had Measurable ‘Welcome Mat’ Effects on Their Children,” Health Affairs 36, no. 9 (September 2017), available at https://www.healthaffairs.org/ doi/10.1377/hlthaff.2017.0347. 12 “Health Coverage for Parents and Caregivers Helps Children” (Washington: Georgetown University Center for Children and Families, March 2017), available at https://ccf.georgetown.edu/ wp-content/uploads/2017/03/Covering-Parents-v2.pdf. 13 J. Haley et al., “Medicaid/CHIP Participation Reached 93.7 Percent Among Eligible Children In 2016,” Health Affairs 37, no. 8 (August 2018), available at https://www.healthaffairs.org/ doi/full/10.1377/hlthaff.2018.0417. 14 E. K. Adams et al., “Prepregnancy Insurance and Timely Prenatal Care for Medicaid Births: Before and After the Affordable Care Act in Ohio,” Journal of Women’s Health, (August 29, 2018), available at https://www.liebertpub.com/doi/ abs/10.1089/jwh.2017.6871. 15 M. Holmes et al., “Geographic Variation in Plan Uptake in the Federally Facilitated Marketplace” (Chapel Hill: North Carolina Rural Research Program, September 2014, available at https://www.ruralhealthresearch.org/alerts/30; C. Drake, J. Abraham, and J. McCullough, “Rural Enrollment in the Federally Facilitated Marketplace,” Journal of Rural Health 32, no. 3 (September 24, 2016): 332-339, available at https:// onlinelibrary.wiley.com/doi/abs/10.1111/jrh.12149. 16 A. Searing and J. Hoadley, “Beyond the Reduction in Uncompensated Care: Medicaid Expansion Is Having a Positive Impact on Safety Net Hospitals and Clinics” (Washington: Georgetown University Center for Children and Families, June 2016), available at https://ccf.georgetown. edu/2016/06/07/medicaid_expansion_positive_impact_safety_ net_hospitals_clinics/. 17 8 18 T. Brooks et al., “Medicaid and CHIP Eligibility, March 2018 Enrollment, Renewal, and Cost Sharing Policies as of January 2018: Findings from a 50-State Survey” (Washington: Kaiser Family Foundation, March 2018), available at https://www.kff. org/medicaid/report/medicaid-and-chip-eligibility-enrollmentrenewal-and-cost-sharing-policies-as-of-january-2018findings-from-a-50-state-survey/. 19 9 These parents are covered through a provision in the Social Security Act – §1902(a)(10)(A)(i)(1) – which requires states to cover parents based on a standard equivalent to states’ previous AFDC standards. 10 September 2018 M. Cole et al., “Medicaid Expansion and Community Health Centers: Care Quality and Service Use Increased for Rural Patients” Health Affairs 37, no. 6 (June 2018): 900–907, available at https://www.healthaffairs.org/doi/pdf/10.1377/ hlthaff.2017.1542. R. Lindrooth, et al., “Understanding the Relationship Between Medicaid Expansions and Hospital Closures” Health Affairs 37, no. 1, (January 2018): 111-120, available at https://www.healthaffairs.org/doi/10.1377/hlthaff.2017.0976; A. Searing, “Study Documents How Medicaid Expansion Helps Keep Rural Hospitals Open” (Washington: Georgetown University Center for Children and Families, January 12, HEALTH INSURANCE COVERAGE IN SMALL TOWNS AND RURAL AMERICA 19 2018), available at https://ccf.georgetown.edu/2018/01/12/ study-documents-how-medicaid-expansion-helps-keep-ruralhospitals-open/. M. Holmes, “Financially Fragile Rural Hospitals: Mergers and Closures,” North Carolina Medical Journal 76, no. 1, (January 2015): 37-40, available at http://www.ncmedicaljournal.com/ content/76/1/37.full. 20 J. Hoadley et al., “Medicaid in Small Towns and Rural America: A Lifeline for Children, Families, and Communities” (Washington: Georgetown University Center for Children and Families and North Carolina Rural Health Research Program, June 2017), available at https://ccf.georgetown. edu/2017/06/06/rural-health-report/. 21 Applications of this approach can be found in T.C. Ricketts III, M. Holmes, “The Uninsured in North Carolina, 2004,” North Carolina Medical Journal 67 no. 3 (2006): 235-236; and M. Holmes, “County-Level Estimates of the Number of Individuals in North Carolina Who Would Be Eligible for Coverage Under the Affordable Care Act’s Expanded Insurance Options,” North Carolina Medical Journal 74 no. 4 (2013): 343-347. 22 Missouri Census Data Center, University of Missouri, http:// mcdc.missouri.edu/websas/geocorr12.html. 23 J.N.K. Rao, “Small Area Estimation” (Hoboken: John Wiley & Sons, Inc., 2003). 24 The tables in this paragraph are from the American Fact Finder, U.S. Census Bureau, available at https://factfinder. census.gov/faces/tableservices/jsf/pages/productview. xhtml?src=bkmk. 25 M. Ratcliffe et al., “Defining Rural at the U.S. Census Bureau” (Washington: United States Census Bureau, December 2016), available at https://www.census.gov/library/publications/2016/ acs/acsgeo-1.html. 26 J. Foutz, S. Artiga, and R. Garfield, “The Role of Medicaid in Rural America” (Washington: Kaiser Family Foundation, April 25, 2017), available at https://www.kff.org/medicaid/issuebrief/the-role-of-medicaid-in-rural-america/. 28 Kaiser Family Foundation, “Status of State Action on the Medicaid Expansion Decision” (Washington: Kaiser Family Foundation, July 2018), available at https://www.kff.org/ health-reform/state-indicator/state-activity-around-expandingmedicaid-under-the-affordable-care-act/. 27 20 HEALTH INSURANCE COVERAGE IN SMALL TOWNS AND RURAL AMERICA September 2018