Lack Of Access To Specialists Associated With Mortality And Preventable Hospitalizations Of Rural Medicare Beneficiaries

Americans living in rural areas have poorer health outcomes than their urban counterparts do. Higher mortality rates in rural versus urban areas have been documented across patient groups and care settings, such as for heart and respiratory disease patients,¹ dialysis patients,² cancer patients,³ and hospital inpatients.⁴ Furthermore, the problem is not lessening: Although mortality rates are declining nationally, the gap in mortality rates between rural and urban areas has grown substantially over time.^5,6 Rural Americans also have higher rates of other adverse health outcomes, such as preventable hospitalizations⁷ and emergency department visits.⁸

To address these health disparities, the Centers for Medicare and Medicaid Services (CMS) recently unveiled several initiatives to improve access to primary care providers and specialists. These initiatives include expanding access to telehealth and virtual services by authorizing Medicare to pay for virtual patient check-ins by telephone; remote evaluation of recorded images sent by patients to clinicians; and remote wellness visits for patients with both routine and complex care needs. Other initiatives include expanding the scope of practice of lower-level nonphysician providers to offer primary care services and increasing the wage index of rural hospitals to enable them to recruit more clinicians and staff.⁹

Understanding what factors account for poorer rural health outcomes could help CMS more efficiently target policy interventions toward reducing rural-urban health disparities. Various differences between rural and urban populations are potential contributors to these disparities. First, there are differences in population demographics and health status. Rural residents are older, and they are more likely to have chronic conditions^10,11 and experience complications of chronic disease.^11,12 They also report higher levels of disability¹² and poorer overall health.¹³ In addition, rates of tobacco use and obesity are higher in rural areas.^12,14,15 Second, social risk factors such as low education, lack of employment, and poverty are more common in rural than in urban areas, and prior research has shown that these factors help explain part of the difference in rural-urban mortality rates.⁶ Third, people in rural areas face problems related to the quality of and access to health care. Multiple studies have documented the lower quality of ambulatory care for people with chronic conditions and lower preventive screening rates in rural areas, compared to urban areas.^11,16 The overall rates of labor supply of specialist physicians in particular^17,18 and health care providers in general¹⁹ are lower in rural areas. As a result, rural residents with chronic conditions are less likely to receive specialized care from a clinician with disease-relevant expertise.¹⁹

Medicare beneficiaries who have complex chronic conditions are an especially vulnerable group and are more likely to experience poor health outcomes, such as hospitalization or death. As a result, it is important to understand the nature of rural-urban disparities and to systematically assess what factors contribute to these disparities, so that CMS can prioritize effective interventions. We therefore focused on three research questions: First, how do Medicare beneficiaries from rural, micropolitan, and metropolitan areas (defined below) differ in terms of demographics, medical comorbidities, functional status, social risk, and access to care? Second, are these differences related to mortality and preventable hospitalizations? And third, what is the relative effect of each group of factors in explaining differences in clinical outcomes across rural, micropolitan, and metropolitan beneficiaries?

Study Data And Methods

Data Sources And Study Sample

To assess differences in baseline-year patient risk factors and following-year health outcomes, we conducted a longitudinal study of rural and urban Medicare beneficiaries with complex chronic conditions. We used information from the Medicare Current Beneficiary Survey (MCBS) linked to respondents’ fee-for-service Medicare claims and administrative data for the years 2006–13. The MCBS is a nationally representative annual survey of the Medicare population, with a rotating four-year longitudinal cohort design.²⁰ We also linked data on health care supply at the level of Hospital Service Area provided by the Dartmouth Institute for Health Policy and Clinical Practice²¹ and county-level rural-urban classifications from the Area Health Resources Files provided by the Health Resources and Services Administration²² to MCBS respondents’ records, using their address of residence. Specifically, we used core-based statistical area indicator codes to classify beneficiaries’ county of residence as metropolitan (city), micropolitan (town), or rural.

We limited the study sample to beneficiaries with selected complex chronic conditions—heart failure, ischemic heart disease, diabetes, and chronic obstructive pulmonary disease (COPD) or asthma—using either patient self-report (or the report of the patient’s proxy respondent) of having been diagnosed with at least one of those conditions or International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM), codes defined by CMS to provide a confirmatory diagnosis in their medical claims.²³ We further limited the study sample to Medicare beneficiaries who had at least two years of continuous enrollment in fee-for-service Medicare Parts A and B that overlapped with two years of participation in the MCBS, or who had at least twelve months of enrollment in the baseline year and had died during the following year. We excluded beneficiaries with no interactions with the health care system in the baseline year. We also excluded beneficiaries with no listed ZIP code to allow us to determine whether they resided in a rural or an urban county. We used MCBS survey weights to compute nationally representative estimates of the fee-for-service Medicare population. These weights account for the overall selection probability of each person sampled and include adjustments for the stratified sampling design, survey nonresponse, and coverage error.

Study Variables

Our dependent variables were the clinical health outcomes of mortality and preventable hospitalization. We measured mortality using the date of death reported in the CMS administrative records that were included in the MCBS for beneficiaries. To identify preventable hospitalizations, we used the ICD-9-CM diagnosis codes listed on study participants’ inpatient hospital claims with an algorithm provided by the Agency for Healthcare Research and Quality.²⁴

We categorized our independent variables into demographic characteristics, medical comorbidities, functional status, social risk, and access to care. For demographics, we used data from the MCBS to determine beneficiaries’ age, sex, and race. We measured medical comorbidities using the CMS Hierarchical Condition Categories risk score (version V1210.70.F2);²⁵ the original reason for Medicare eligibility (end-stage renal disease, disabled, or age sixty-five or older); and separate indicators for diabetes, heart failure, ischemic heart disease, and COPD or asthma. To measure functional status, we grouped counts of self-reported difficulty with activities of daily living (ADLs) and instrumental ADLs (IADLs) into no limitations (0), mild or moderate limitations (1 or 2), or severe limitations (3–6). We assessed frailty using a claims-based index that has been validated among Medicare beneficiaries.²⁶ Finally, we assessed social risk using self-reported measures of material capital (annual income), human capital (highest level of education attained), and social support (marital status).

We defined access to care as the local supply rates of primary care physicians, medical specialists, and hospital beds, using data from the Dartmouth Institute.²¹ We also measured realized access to (that is, contact with) primary care providers and specialists for face-to-face evaluation and management visits in the ambulatory care setting. To identify evaluation and management visits to primary care providers or specialists, we used the Current Procedural Terminology codes that CMS uses to attribute beneficiaries to providers under the Merit-based Incentive Payment System.²⁷ To identify such visits to federally qualified health centers or rural health clinics, we used revenue codes from Medicare outpatient facility claims. We defined contact with primary care providers as having one or more evaluation and management visits in a federally qualified health center or rural health clinic or with any of the following specialty types: general practice, family practice, internal medicine, geriatric medicine, preventive medicine, obstetrics-gynecology, nurse practitioner, and physician assistant. We defined contact with specialty care as having one or more evaluation and management visits with any other specialty type. We used this information to divide the sample members into four mutually exclusive groups: those with at least one primary care provider visit and at least one specialist visit, those with at least one primary care provider visit but no specialist visit, those with at least one specialist visit but no primary care provider visit, and those with no primary care provider visit or specialist visit.

Statistical Analysis

First, we computed weighted descriptive statistics for independent variables in the baseline year and dependent variables in the following year. We used the Wald test to compare differences in proportions (or means) on key characteristics between beneficiaries classified as rural, micropolitan, or metropolitan. In addition, we computed weighted descriptive statistics for the specific specialty types that beneficiaries visited in the baseline year and the characteristics of beneficiaries with and without specialist contact in the baseline year.

Second, we estimated two patient-level multivariable regression models to assess the association between the five independent variable categories (demographics, medical comorbidities, functional status, social risk, and access to care) in the baseline year and the clinical outcomes in the following year. We used weighted negative binomial regression to estimate the effects of our independent variables on the count outcome of preventable hospitalization, and we used weighted logistic regression to estimate the same effects on the binary outcome of mortality. We reported these results as marginal effects—that is, we modeled the response in the dependent variables to a one-unit change in the independent variables at the population mean. In addition, we translated the marginal effects into percentage effects by dividing them by the dependent variable means. Both models included year fixed effects to control for secular trends and were adjusted for the complex survey design of the MCBS and intraperson correlation over time. We estimated these models for our full study sample of Medicare beneficiaries with complex chronic conditions as well as for a subsample with the cardiovascular conditions of heart failure or ischemic heart disease to determine whether these conditions were particularly responsive to access to specialty care.

Third, we conducted a decomposition analysis to estimate the relative explanatory effect of the five independent variable categories on differences in clinical outcomes between rural and metropolitan beneficiaries. Specifically, we used the Fairlie decomposition method²⁸ to determine the extent to which the difference in the probability of having one or more preventable hospitalizations and the difference in mortality between rural and metropolitan beneficiaries were explained by differences in beneficiaries’ observed characteristics in the baseline year. The Fairlie method is a nonlinear extension of the Oaxaca-Blinder decomposition method^29,30 and is used to explain group differences in an outcome variable. It decomposes group differences into an explained component based on observable characteristics and an unexplained component that reflects the effect of omitted variables.^6,31 We reported the decomposition results as the percentages of rural versus urban difference in the probabilities of preventable hospitalization or mortality that were explained by each predictor variable and by each of the five independent variable categories.

Limitations

Our study had several limitations. First, our sample was limited to fee-for-service Medicare beneficiaries for whom both medical claims and MCBS data were available. This excluded the 25 percent of the Medicare population enrolled in managed care during this period.³²

Second, as this was an observational study, there may have been confounding by unobserved factors on which rural and urban beneficiaries (or those with and without specialist involvement) systematically differed. We attempted to mitigate confounding by limiting our study to beneficiaries with the same complex chronic conditions and by controlling for many potential confounders (such as social risk and functional status), using the rich MCBS data set linked to fee-for-service Medicare claims. Nonetheless, because we could not rule out potential confounding due to unobserved factors, our estimates might not imply causal relationships.

Third, we found higher Hierarchical Condition Categories scores for urban beneficiaries than for rural ones, which suggests the possibility of greater coding intensity in urban areas—which in turn could be correlated with greater contact with specialists. This may have artifactually caused risk-adjusted outcomes to appear better for beneficiaries who had specialist visits, compared to those who did not. However, we controlled for this by using self-reported measures of both disease and functional status in the MCBS to help mitigate the effects of coding bias in our study.

Lastly, we included nurse practitioners, physician assistants, federally qualified health centers, and rural health clinics in our definition of primary care providers, along with physicians. The beneficial effect that we found for contact with specialists could also reflect a beneficial effect of physician care versus care from nonphysician providers.

Study Results

Patient Characteristics

Of the 115,833,835 weighted patient-years that met our two-year fee-for-service Medicare enrollment criteria, 11,077,877 (9.6 percent) were associated with beneficiaries residing in rural areas, 20,381,285 (17.6 percent) with beneficiaries in micropolitan areas, and 84,374,674 (72.8 percent) with beneficiaries in metropolitan areas (see online appendix exhibit 1).³³ After we limited the study sample to beneficiaries with complex chronic conditions and excluded people who had no baseline interactions with the health care system or who had missing variables, the final sample consisted of 66,585,996 weighted patient-years, or 17,695 unweighted patient-years (exhibit 1). The patient-years represented 11,581 unique beneficiaries, some of whom appeared in the data in multiple years. In total, 6,743,430 (61 percent) of rural patient-years, 12,030,928 (59 percent) of micropolitan patient-years, and 47,811,637 (57 percent) of metropolitan patient-years that met enrollment criteria also had one or more complex chronic conditions and were included in our final study sample.

Compared to micropolitan or metropolitan beneficiaries, rural beneficiaries were more often white (appendix exhibit 2).³³ In terms of medical comorbidities (exhibit 1), rural beneficiaries had a higher burden of heart failure and ischemic heart disease, though they had lower Hierarchical Condition Categories risk scores (compared with metropolitan beneficiaries). Neither functional status, as measured by difficulties with ADLs, nor frailty differed significantly by rurality. Social risk factors were markedly more common in rural beneficiaries: They had lower incomes and less education than micropolitan or metropolitan beneficiaries did.

Access and realized access also differed significantly across levels of rurality. While local-area primary care physician supply was similar in rural and metropolitan areas, the supply of specialists was 31 percent lower in rural areas, and fewer beneficiaries in rural areas had any contact with a specialist (exhibit 1) (for a graphical representation, see appendix exhibit 3).³³ Rural beneficiaries had a lower likelihood of contact with all disease-relevant specialists, including cardiologists, pulmonologists, and endocrinologists (appendix exhibit 4).³³

Clinical Outcomes

Raw preventable hospitalization and mortality rates were higher in rural areas than in micropolitan areas, and the rates in micropolitan areas were higher than those in metropolitan areas (exhibit 2). Preventable hospitalization rates per 100 beneficiaries ranged from 14.9 in rural areas to 10.6 in metropolitan areas. Annual mortality rates ranged from 8.6 percent in rural areas to 7.0 percent in metropolitan areas. This means that rural residents experienced 40 percent higher preventable hospitalization rates and 23 percent higher mortality rates than their metropolitan counterparts did.

SOURCE Authors’ analysis of data for 2006–13 from the Medicare Current Beneficiary Survey. NOTES Rural, micropolitan, and metropolitan areas are explained in the text. All differences between area types were significant for preventable hospitalizations ($p<0.05$) and mortality rates ($p<0.01$).

Association Of Factors With Preventable Hospitalization And Mortality

An increase of one standard deviation in the supply of medical specialists was associated with an 8.3 percent lower mortality rate (exhibit 3). Receipt of more evaluation and management visits was associated with a slightly higher rate of preventable hospitalization, but this could be due to less-than-perfect risk adjustment because more visits may indicate more unmeasured severity. Having one or more visits with a specialist in addition to one or more visits with a primary care provider was associated with a 15.9 percent lower preventable hospitalization rate and a 16.6 percent lower mortality rate, relative to having only one or more visits with a primary care provider. In addition to those associations with access to care, exhibit 3 and appendix exhibit 5 show that preventable hospitalization and mortality rates were also associated with factors in the other four categories.³³ Notably, a higher Hierarchical Condition Categories risk score and the presence of heart failure were associated with higher preventable hospitalization and mortality rates. Impaired functional status, as evidenced by frailty and difficulty with ADLs and IADLs, were associated with higher preventable hospitalization or mortality rates. Lastly, having a lower income and being unmarried were associated with higher preventable hospitalization rates, and having less education was associated with higher mortality rates.

When we limited our sample to beneficiaries with cardiovascular conditions, we found larger beneficial effects of access to specialty care than we did in our main analysis (appendix exhibit 6).³³ Specifically, we found that each standard deviation of additional specialist supply was associated with a lower mortality rate of 11.3 percent and that having one or more visits with a specialist in addition to having one or more visits with a primary care provider was associated with a 17.5 percent lower preventable hospitalization rate and a 33.8 percent lower mortality rate, relative to having only one or more visits with a primary care provider.

In a sensitivity analysis, we excluded the 2.5 percent of our sample who did not have any visits with either a specialist or a primary care provider, to test whether those beneficiaries were skewing our findings, and we reran the regression models described above. We found very similar results (appendix exhibit 7).³³

To determine whether the relationships we found between specialist involvement and outcomes reflected underlying differences in the patients who were able to obtain specialty care, we examined characteristics of these groups. We found that beneficiaries with specialist involvement in their care appeared to be sicker than beneficiaries without it, as evidenced by their substantially higher Hierarchical Condition Categories risk scores, higher prevalences of heart failure and ischemic heart disease, and greater numbers of evaluation and management visits (appendix exhibit 8).³³ However, beneficiaries with specialist involvement were also less likely to have severe functional limitations and more likely to have higher education and more income and be white and married.

Decomposition Results

Rural residence was associated with a 33 percent higher probability of having one or more preventable hospitalizations (11.2 percent versus 8.4 percent) and a 23 percent higher probability of mortality (8.6 percent versus 7.0 percent) than metropolitan residence was (appendix exhibit 9).³³ Our decomposition analyses showed that access-to-care factors explained 89 percent of the difference in the probability of preventable hospitalization and 32 percent of the difference in mortality between rural and urban residents (exhibit 4). Social risk factors explained 7 percent of the difference in preventable hospitalization and 12 percent of the difference in mortality. The predictor variable with the single greatest explanatory effect was local-area supply of specialists, which explained 55 percent of the difference in preventable hospitalization rates and 40 percent of the difference in mortality (appendix exhibit 9).³³ Some factors, notably demographic characteristics and medical comorbidities, had negative explanatory effects—meaning that they contributed to a lower, not a higher, probability of the adverse clinical outcomes in rural residents. However, these factors were more than counterbalanced by the effects of the other factors that led to a higher probability of adverse outcomes in rural residents.

SOURCE Authors’ analysis of data for 2006–13 from the Medicare Current Beneficiary Survey (MCBS). NOTES Baseline years were 2006–12. We used the Fairlie decomposition method (see note 28 in text) to determine the extent to which differences in the probability of having one or more preventable hospitalizations or of dying in the following year by rural versus metropolitan residence were explained by differences in beneficiaries’ observed characteristics in the baseline year. We conducted a weighted decomposition with random ordering of variables over 1,000 replications and used the MCBS sampling weights. We report the results as patient effects, summing the individual predictor variable effects over the five categories of demographic characteristics, medical comorbidities, functional status, social risk, and access to care. The categories are explained in the text.

In a sensitivity analysis, we grouped micropolitan and rural beneficiaries together and decomposed differences between that group and metropolitan beneficiaries (appendix exhibits 10 and 11).³³ We found the same general pattern of results as in our main analysis, except that access-to-care factors explained more of the differences, medical comorbidities had a greater negative explanatory effect, and the total explained difference was less.

Discussion

In this nationally representative study of older adult Medicare beneficiaries, we found that rural beneficiaries with complex chronic conditions had higher preventable hospitalization and mortality rates than their urban peers did. Rural beneficiaries had less access to and contact with specialists for ambulatory care, and contact with specialists was associated with substantially lower preventable hospitalization and mortality rates among all beneficiaries. Furthermore, we found that access to care, particularly to specialists, explained a sizable portion of the difference in preventable hospitalization and mortality rates between rural and urban beneficiaries.

We also found that rural beneficiaries faced markedly higher levels of social risk in the form of lower income and less education and that social risk factors helped explain some of the rural-urban difference in preventable hospitalization and mortality rates. Prior research has found that socioeconomic factors such as low income, lack of education, and living in deprived areas are associated with more hospitalizations,³⁴ greater use of health care,³⁵ and increased likelihood of mortality.^36,37

Access to primary care does not appear to drive rural-urban health outcome disparities.

However, we found that lack of access to specialists was the primary driver of higher mortality and preventable hospitalization rates among rural Medicare beneficiaries with chronic conditions. This was reinforced by our finding of even larger outcome disparities among rural and urban beneficiaries with cardiovascular diseases, which were explained by disparities in access to specialists such as cardiologists. It is important to note that although current policy recommendations for reducing rural health disparities are targeted at increasing access to primary care,^38,39 we did not find disparities in access to primary care between rural and urban beneficiaries, nor did we find primary care access to be a driver of existing outcome disparities. Thus, to the extent that current policy interventions focus on expanding primary care but not specialist care in rural areas, they appear to be misguided and unlikely to reduce disparities in rural health outcomes. Notably, multiple studies have found that regular treatment by specialist physicians in the ambulatory care setting is associated with better quality of care and reduced risk of death or hospitalization for people with chronic conditions.^40–43 This does not detract from the value of primary care. However, access to primary care does not appear to drive rural-urban health outcome disparities.

Of course, specialist involvement is not random. Our analyses showed that patients with specialist involvement had greater disease severity than those without it (which could have biased our estimates of the beneficial effects of specialist involvement toward the null), but they also had better functional status and fewer social risk factors (which could have biased our estimates away from the null). Further study, including analyses of the specific differences in care delivered by specialists versus primary care providers in rural areas, is needed to determine the mechanisms that underlie our findings.

Rural beneficiaries without access to specialist care are a medically and socially vulnerable population.

Rural beneficiaries without access to specialist care are a medically and socially vulnerable population. Policy makers should consider targeting interventions to these beneficiaries to improve care and reduce rural-urban disparities in health outcomes. Our results have implications for CMS’s Rural Health Strategy⁹ and suggest that the expansion of telemedicine in rural areas to expand access to key specialty areas (such as cardiology) for Medicare beneficiaries with complex chronic conditions could have a particularly large impact. Additionally, workforce reforms to increase the supply of specialists in rural areas (for example, loan forgiveness programs and differential payment rates) could be useful. As noted above, CMS recently increased the wage index of rural hospitals,⁹ which will increase their reimbursement. A similar approach might be considered for medical specialists as well. Health systems could implement other creative models, such as incentives for rural and urban hospitals to form partnerships and the creation of opportunities for specialists to provide intermittent service for rural areas.

Conclusion

Rural Medicare beneficiaries with complex chronic conditions have higher preventable hospitalization and mortality rates than their urban peers do. Part of this difference appears to be explained by access to specialists, which could either represent a direct effect of specialist involvement or be a marker for access to differential quality and sophistication of care overall.

ACKNOWLEDGMENTS

Saint Louis University purchased and provided access to the data used in this study for Kenton Johnston. Karen Joynt Maddox receives research support from the National Heart, Lung, and Blood Institute (Grant No. R01HL143421) and National Institute on Aging (Grant No. R01AG060935), and previously did contract work for the Department of Health and Human Services. The funders had no role in the design and conduct of the study; collection, management, analysis, or interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. The authors thank Julia Clarke of Saint Louis University for providing assistance with the literature review. The authors thank the peer reviewers and the editors for helpful feedback that improved the article.

NOTES

• 1 Moy E, Garcia MC, Bastian B, Rossen LM, Ingram DD, Faul Met al. Leading causes of death in nonmetropolitan and metropolitan areas—United States, 1999–2014. MMWR Surveill Summ. 2017;66(1):1–8. Crossref, Medline, Google Scholar
• 2 Maripuri S, Arbogast P, Ikizler TA, Cavanaugh KL. Rural and micropolitan residence and mortality in patients on dialysis. Clin J Am Soc Nephrol. 2012;7(7):1121–9. Crossref, Medline, Google Scholar
• 3 Singh GK, Williams SD, Siahpush M, Mulhollen A. Socioeconomic, rural-urban, and racial inequalities in US cancer mortality: part I—all cancers and lung cancer and part II—colorectal, prostate, breast, and cervical cancers. J Cancer Epidemiol. 2011;2011:107497. Crossref, Medline, Google Scholar
• 4 Villapiano N, Iwashyna TJ, Davis MM. Worsening rural-urban gap in hospital mortality. J Am Board Fam Med. 2017;30(6):816–23. Crossref, Medline, Google Scholar
• 5 Singh GK, Siahpush M. Widening rural-urban disparities in all-cause mortality and mortality from major causes of death in the USA, 1969–2009. J Urban Health. 2014;91(2):272–92. Crossref, Medline, Google Scholar
• 6 Spencer JC, Wheeler SB, Rotter JS, Holmes GM. Decomposing mortality disparities in urban and rural U.S. counties. Health Serv Res. 2018;53(6):4310–31. Crossref, Medline, Google Scholar
• 7 Laditka JN, Laditka SB, Probst JC. Health care access in rural areas: evidence that hospitalization for ambulatory care–sensitive conditions in the United States may increase with the level of rurality. Health Place. 2009;15(3):731–40. Crossref, Medline, Google Scholar
• 8 Greenwood-Ericksen MB, Kocher K. Trends in emergency department use by rural and urban populations in the United States. JAMA Netw Open. 2019;2(4):e191919. Crossref, Medline, Google Scholar
• 9 Centers for Medicare and Medicaid Services. CMS Rural Health Strategy [Internet]. Baltimore (MD): CMS; [cited 2019 Oct 18]. Available from: https://www.cms.gov/About-CMS/Agency-Information/OMH/Downloads/Rural-Strategy-2018.pdf Google Scholar
• 10 Weeks WB, Kazis LE, Shen Y, Cong Z, Ren XS, Miller Det al. Differences in health-related quality of life in rural and urban veterans. Am J Public Health. 2004;94(10):1762–7. Crossref, Medline, Google Scholar
• 11 Hale NL, Bennett KJ, Probst JC. Diabetes care and outcomes: disparities across rural America. J Community Health. 2010;35(4):365–74. Crossref, Medline, Google Scholar
• 12 Jackson BE, Coultas DB, Suzuki S, Singh KP, Bae S. Rural-urban disparities in quality of life among patients with COPD. J Rural Health. 2013;29(Suppl 1):s62–9. Crossref, Medline, Google Scholar
• 13 Henning-Smith C, Kozhimannil K, Casey M, Prasad S, Moscovice I. Rural-urban differences in Medicare quality outcomes and the impact of risk adjustment. Med Care. 2017;55(9):823–9. Crossref, Medline, Google Scholar
• 14 Atkins GT, Kim T, Munson J. Residence in rural areas of the United States and lung cancer mortality. Disease incidence, treatment disparities, and stage-specific survival. Ann Am Thorac Soc. 2017;14(3):403–11. Crossref, Medline, Google Scholar
• 15 Hales CM, Fryar CD, Carroll MD, Freedman DS, Aoki Y, Ogden CL. Differences in obesity prevalence by demographic characteristics and urbanization level among adults in the United States, 2013–2016. JAMA. 2018;319(23):2419–29. Crossref, Medline, Google Scholar
• 16 Henning-Smith C, Prasad S, Casey M, Kozhimannil K, Moscovice I. Rural-urban differences in Medicare quality scores persist after adjusting for sociodemographic and environmental characteristics. J Rural Health. 2019;35(1):58–67. Crossref, Medline, Google Scholar
• 17 Chen X, Orom H, Hay JL, Waters EA, Schofield E, Li Yet al. Differences in rural and urban health information access and use. J Rural Health. 2019;35(3):405–17. Crossref, Medline, Google Scholar
• 18 Shugarman LR, Sorbero ME, Tian H, Jain AK, Ashwood JS. An exploration of urban and rural differences in lung cancer survival among Medicare beneficiaries. Am J Public Health. 2008;98(7):1280–7. Crossref, Medline, Google Scholar
• 19 Carr BG, Bowman AJ, Wolff CS, Mullen MT, Holena DN, Branas CCet al. Disparities in access to trauma care in the United States: a population-based analysis. Injury. 2017;48(2):332–8. Crossref, Medline, Google Scholar
• 20 CMS.gov. Medicare Current Beneficiary Survey: data documentation and codebooks [Internet]. Baltimore (MD): Centers for Medicare and Medicaid Services; [cited 2019 Oct 21]. Available from: https://www.cms.gov/Research-Statistics-Data-and-Systems/Research/MCBS/Codebooks.html Google Scholar
• 21 Goodman DC, Fisher ES, Bronner KK. Hospital and physician capacity update [Internet]. Lebanon (NH): Dartmouth Institute for Health Policy and Clinical Practice; 2009 Mar 30 [cited 2019 Oct 18]. (Brief Report). Available from: http://www.dartmouthatlas.org/downloads/reports/Capacity_Report_2009.pdf Google Scholar
• 22 Health Resources and Services Administration. Area Health Resources Files [Internet]. Rockville (MD): HRSA; [cited 2019 Oct 18]. Available from: https://data.hrsa.gov/topics/health-workforce/ahrf Google Scholar
• 23 Chronic Conditions Data Warehouse. Condition categories: CCW chronic condition algorithms [Internet]. Baltimore (MD): Centers for Medicare and Medicaid Services; [last revised 2019 Feb; cited 2019 Oct 17]. Available for download from: https://www.ccwdata.org/web/guest/condition-categories Google Scholar
• 24 Agency for Healthcare Research and Quality. AHRQ quality indicators: guide to Prevention Quality Indicators: hospital admission for ambulatory care sensitive conditions, version 3.1 [Internet]. Rockville (MD): AHRQ; 2007 Mar 12 [cited 2019 Oct 18]. Available from: https://www.qualityindicators.ahrq.gov/Downloads/Modules/PQI/V31/pqi_guide_v31.pdf Google Scholar
• 25 Pope GC, Kautter J, Ellis RP, Ash AS, Ayanian JZ, Lezzoni LIet al. Risk adjustment of Medicare capitation payments using the CMS-HCC model. Health Care Financ Rev. 2004;25(4):119–41. Medline, Google Scholar
• 26 Kim DH, Schneeweiss S, Glynn RJ, Lipsitz LA, Rockwood K, Avorn J. Measuring frailty in Medicare data: development and validation of a claims-based frailty index. J Gerontol A Biol Sci Med Sci. 2018;73(7):980–7. Crossref, Medline, Google Scholar
• 27 Centers for Medicare and Medicaid Services, Quality Payment Program. Merit-based Incentive Payment System (MIPS): total per capita costs for all attributed beneficiaries [Internet]. Baltimore (MD): CMS; 2018 [cited 2019 Oct 28]. Available for download from: https://qpp-cm-prod-content.s3.amazonaws.com/uploads/147/Cost%20Measures.zip Google Scholar
• 28 Fairlie RW. Addressing path dependence and incorporating sample weights in the nonlinear Blinder-Oaxaca decomposition technique for logit, probit, and other nonlinear models [Internet]. Stanford (CA): Stanford Institute for Economic Policy Research; 2017 Apr [cited 2019 Oct 18]. (SIEPR Discussion Paper No. 17-013). Available from: https://siepr.stanford.edu/sites/default/files/publications/17-013.pdf Google Scholar
• 29 Blinder AS. Wage discrimination: reduced form and structural estimates. J Hum Resour. 1973;8(4):436–55. Crossref, Google Scholar
• 30 Oaxaca R. Male-female wage differentials in urban labor markets. Int Econ Rev (Philadelphia). 1973;14(3):693–709. Crossref, Google Scholar
• 31 Mortensen K, Song PH. Minding the gap: a decomposition of emergency department use by Medicaid enrollees and the uninsured. Med Care. 2008;46(10):1099–107. Crossref, Medline, Google Scholar
• 32 Medicare Payment Advisory Commission. A data book: health care spending and the Medicare program [Internet]. Washington (DC): MedPAC; 2013 Jun [cited 2019 Oct 18]. Available from: http://67.59.137.244/documents/Jun13DataBookEntireReport.pdf Google Scholar
• 33 To access the appendix, click on the Details tab of the article online.
• 34 Moy E, Chang E, Barrett M. Potentially preventable hospitalizations—United States, 2001–2009. MMWR Suppl. 2013;62(3):139–43. Medline, Google Scholar
• 35 Fitzpatrick T, Rosella LC, Calzavara A, Petch J, Pinto AD, Manson Het al. Looking beyond income and education: socioeconomic status gradients among future high-cost users of health care. Am J Prev Med. 2015;49(2):161–71. Crossref, Medline, Google Scholar
• 36 Calvillo-King L, Arnold D, Eubank KJ, Lo M, Yunyongying P, Stieglitz Het al. Impact of social factors on risk of readmission or mortality in pneumonia and heart failure: systematic review. J Gen Intern Med. 2013;28(2):269–82. Crossref, Medline, Google Scholar
• 37 Coady SA, Johnson NJ, Hakes JK, Sorlie PD. Individual education, area income, and mortality and recurrence of myocardial infarction in a Medicare cohort: the National Longitudinal Mortality Study. BMC Public Health. 2014;14(1):705. Crossref, Medline, Google Scholar
• 38 Bolin JN, Bellamy GR, Ferdinand AO, Vuong AM, Kash BA, Schulze Aet al. Rural Healthy People 2020: new decade, same challenges. J Rural Health. 2015;31(3):326–33. Crossref, Medline, Google Scholar
• 39 Bodenheimer T, Pham HH. Primary care: current problems and proposed solutions. Health Aff (Millwood). 2010;29(5):799–805. Go to the article, Google Scholar
• 40 Johnston KJ, Hockenberry JM. Are two heads better than one or do too many cooks spoil the broth? The trade-off between physician division of labor and patient continuity of care for older adults with complex chronic conditions. Health Serv Res. 2016;51(6):2176–205. Crossref, Medline, Google Scholar
• 41 Ansari M, Alexander M, Tutar A, Bello D, Massie BM. Cardiology participation improves outcomes in patients with new-onset heart failure in the outpatient setting. J Am Coll Cardiol. 2003;41(1):62–8. Crossref, Medline, Google Scholar
• 42 Smetana GW, Landon BE, Bindman AB, Burstin H, Davis RB, Tjia Jet al. A comparison of outcomes resulting from generalist vs specialist care for a single discrete medical condition: a systematic review and methodologic critique. Arch Intern Med. 2007;167(1):10–20. Crossref, Medline, Google Scholar
• 43 Ezekowitz JA, van Walraven C, McAlister FA, Armstrong PW, Kaul P. Impact of specialist follow-up in outpatients with congestive heart failure. CMAJ. 2005;172(2):189–94. Crossref, Medline, Google Scholar