Rural-Urban Disparities In All-Cause Mortality Among Low-Income Medicare Beneficiaries, 2004–17

Abstract

There is growing concern about the health of older US adults who live in rural areas, but little is known about how mortality has changed over time for low-income Medicare beneficiaries residing in rural areas compared with their urban counterparts. We evaluated whether all-cause mortality rates changed for rural and urban low-income Medicare beneficiaries dually enrolled in Medicaid, and we studied disparities between these groups. The study cohort included 11,737,006 unique dually enrolled Medicare beneficiaries. Between 2004 and 2017 all-cause mortality declined from 96.6 to 92.7 per 1,000 rural beneficiaries (relative percentage change: −4.0 percent). Among urban beneficiaries, declines in mortality were more pronounced (from 86.9 to 72.8 per 1,000 beneficiaries, a relative percentage change of −16.2 percent). The gap in mortality between rural and urban beneficiaries increased over time. Rural mortality rates were highest in East North central states and increased modestly in West North central states during the study period. Public health and policy efforts are urgently needed to improve the health of low-income older adults living in rural areas.

Significant health disparities exist between Medicare beneficiaries residing in rural areas of the US and their urban counterparts.¹ Rural beneficiaries often face numerous challenges that influence their health, including poverty; unreliable access to health care services; and, more recently, hospital closures.^2–4 Given these concerning patterns, policy and public health efforts to improve the health of rural beneficiaries have intensified in recent years.⁵

One group of people at particularly high risk for poor health outcomes is low-income Medicare beneficiaries who are dually enrolled in Medicaid.⁶ They represent the poorest and sickest subset of the Medicare population, with a high burden of chronic illnesses, comorbidities, and social needs.^7,8 However, little is known about the health outcomes of dually enrolled beneficiaries living in rural areas. Although recent policy and public health initiatives have focused on improving the health of low-income people, these efforts have primarily concentrated on urban areas.^9,10 As a result, it is possible that rural beneficiaries have not benefited equally from these initiatives and that disparities between rural and urban dually enrolled beneficiaries have widened over time.^9,11–14 Understanding how health outcomes have changed for this low-income population in rural areas compared with those in urban areas is critical for public health and policy efforts to reduce health inequities.

Therefore, in this study we aimed to answer the following key questions. First, did all-cause mortality rates differ between Medicare (fee-for-service and Medicare Advantage) beneficiaries older than age sixty-five who were dually enrolled in Medicaid in rural versus urban areas? Second, how did mortality change among rural and urban dually enrolled beneficiaries from 2004 to 2017, and did the difference in mortality between these groups decrease over time? Third, is there geographic variation in mortality among rural dually enrolled beneficiaries, and, if so, have these patterns changed over time?

Study Data And Methods

Institutional Review Board approval, including waiver of the requirement of participant informed consent, was provided by Beth Israel Deaconess Medical Center.

STUDY POPULATION

Medicare Master Beneficiary Summary Files were used to identify beneficiaries older than age sixty-five who were enrolled in Medicare for at least twelve months between 2004 and 2017. We included people enrolled in fee-for-service or Medicare Advantage. Medicare beneficiaries were considered dually enrolled if they were also concurrently enrolled in Medicaid for at least one month within a given year. To identify whether Medicare beneficiaries lived in a rural or urban area, we crosswalked each beneficiary’s ZIP code of residence with the 2010 rural-urban commuting area codes from US census tract data.¹⁵ Census tracts are subdivisions of counties; because the rural-urban commuting area is linked to census tracts, it provides a more granular assessment of the rurality of a beneficiary’s place of residence than county-level data. This also allows for more reliable longitudinal comparisons of populations within these census tracts.¹⁶ We used rural-urban commuting area codes to define areas as urban (rural-urban commuting area 1–3) or rural (rural-urban commuting area 4–10), consistent with prior studies^17–20 (see online appendix exhibit 1).²¹ The final study cohort is shown in appendix exhibit 2.²¹

OUTCOMES

The primary outcome was annual all-cause mortality rates among rural and urban dually enrolled Medicare beneficiaries.

STATISTICAL ANALYSIS

Continuous variables were summarized with means and standard deviations; categorical variables were summarized with frequencies and percentages. Patient demographics (age, sex, and race/ethnicity), insurance type (fee-for-service or Medicare Advantage), and geographical region were characterized for beneficiaries residing in rural and urban areas.

Observed annual all-cause mortality rates for rural dually enrolled beneficiaries per 1,000 rural dually enrolled beneficiaries and for urban dually enrolled beneficiaries per 1,000 urban dually enrolled beneficiaries were calculated for each year. We then calculated annual all-cause mortality rates after standardization for age (five-year intervals) and sex to account for differences in demographics between rural and urban beneficiaries and for changes over time (using 2004 as a reference year). To examine whether demographics-standardized all-cause mortality rates changed over time for rural and urban dually enrolled beneficiaries, we used a Z-test to compare the first and final years of the study period. To determine whether mortality changed differentially over time between rural versus urban dually enrolled beneficiaries, we evaluated the interaction between rural status and time. We then performed the same analyses to examine these patterns by the nine US census regions. As a subgroup analysis, we also examined changes in all-cause mortality among Black and White dually enrolled beneficiaries residing in rural versus urban areas. All analyses were performed using SAS, version 15.1.

LIMITATIONS

This study had several limitations. First, many states elected to expand Medicaid under the Affordable Care Act, beginning in 2014, which changed eligibility requirements for the program. However, our study population is unlikely to have been affected, as Medicaid eligibility requirements for people age sixty-five or older are set federally and are more comparable across states. Second, Medicare beneficiaries are able to move in and out of the Medicaid program according to whether they meet eligibility requirements. However, rates of continuous enrollment in Medicaid tend to be high (greater than 85 percent) among Medicare beneficiaries older than age sixty-five.^22,23 Third, because it is linked to census tracts and therefore provides a more granular assessment of where beneficiaries reside, we used the rural-urban commuting area to evaluate whether beneficiaries resided in a rural or urban area. However, this may limit comparisons of our findings to prior studies that use county-level measures of rurality. Fourth, the 2010 rural-urban commuting area data were not able to account for changes in census tract data that may have occurred beyond this time frame. As a result, we were unable to account for minor urbanization or deurbanization changes that may have occurred during the study period. Fifth, approximately 10 percent of dually enrolled beneficiaries in our initial cohort were excluded because of missing information on rural versus urban residence. However, demographics (age and sex) were similar between the final study cohort and the excluded beneficiaries, making it unlikely that excluding these beneficiaries biased our findings.

Study Results

STUDY COHORT

Overall, there were 11,737,006 unique dually enrolled Medicare (fee-for-service and Medicare Advantage) beneficiaries in the period 2004–17. The mean age of the cohort was 74.8 years (standard deviation: 8.8), and 66.2 percent were female. During the study period there were 65,999,705 total dually enrolled beneficiary-years represented, of which 46,668,763 (70.7 percent) were fee-for-service and 19,330,942 (29.3 percent) were Medicare Advantage beneficiary-years. Of the total beneficiary-years, there were 11,929,876 (18.1 percent) rural beneficiary-years and 54,069,829 (81.9 percent) urban beneficiary-years.

Between 2004 and 2017, mean age decreased among rural (from 78.0 [SD: 8.2] to 76.4 [SD: 8.2] years; p < 0.001) and urban (from 77.5 [SD: 8.0] to 76.5 [SD: 8.1] years; p < 0.001) dually enrolled beneficiaries. In both groups the proportion of dually enrolled beneficiaries who were female declined (rural: from 71.1 percent to 66.7 percent [p < 0.001]; urban: from 70.9 percent to 65.9 percent [p < 0.001]). The proportion of dually enrolled White beneficiaries remained consistent in rural areas (at 77.2 percent; p = 0.16) but decreased slightly in urban areas (from 57.5 percent to 53.6 percent; p < 0.001). Annual rates of enrollment in Medicare Advantage for at least one month within a given year increased for both rural (from 3.1 percent to 26.0 percent; p < 0.001) and urban (from 14.0 percent to 46.6 percent; p < 0.001) dually enrolled beneficiaries (exhibit 1; appendix exhibit 3).²¹

EXHIBIT 1.

Baseline characteristics of rural and urban dually enrolled Medicare beneficiaries, 2004–05 and 2016–17

	Rural		Urban
Characteristics	2004–05	2016–17	2004–05	2016–17
Total number	1,616,427	1,811,403	6,550,268	9,144,286
Age, mean years	78.0	76.4	77.5	76.5
Female (%)	71.1	66.7	70.9	65.9
Race/ethnicity (%)
White	77.2	77.2	57.5	53.6
Black	16.7	15.4	19.5	19.7
Other	0.65	0.9	3.0	3.14
Asian	0.61	0.9	10.0	11.3
Hispanic	3.1	3.7	9.7	12.0
Native American	1.8	1.9	0.34	0.4
Insurance type (%)
Fee-for-service	96.9	74.0	86.0	54.4
Medicare Advantage	3.1	26.0	14.0	46.6
Region (%)
East North Central	13.0	13.9	10.6	10.8
East South Central	22.0	19.1	5.8	4.6
West North Central	8.3	7.3	2.8	2.6
West South Central	15.8	13.8	10.7	10.2
Mid-Atlantic	4.4	5.1	16.7	16.0
Mountain	5.4	6.9	4.1	5.0
New England	3.1	4.9	5.2	6.1
Pacific	9.1	10.7	24.4	24.0
South Atlantic	18.9	18.2	19.5	20.7
Others	0.04	0.03	0.13	0.09

SOURCE Authors’ analysis of dually enrolled Medicare (fee-for-service and Medicare Advantage) beneficiaries, 2004–17, using rural-urban commuting area codes and data from the Centers for Medicare and Medicaid Services Medicare Master Beneficiary Summary Files.

The distribution of rural and urban dually enrolled beneficiaries varied geographically. Throughout the study period, the largest share of rural beneficiaries resided in the East South Central region, and the largest share of urban beneficiaries resided in the Pacific region (exhibit 1).

ALL-CAUSE MORTALITY

In the Medicare (fee-for-service and Medicare Advantage) dually enrolled population, observed all-cause mortality rates were higher among rural beneficiaries (92.8 per 1,000 rural dually enrolled beneficiaries) when compared with their urban counterparts (75.1 per 1,000 urban dually enrolled beneficiaries) over the entire study period. Among rural dually enrolled beneficiaries, observed all-cause mortality declined from 99.5 to 87.0 per1.0 beneficiaries (difference: 12.5; 95% confidence interval: 11.6, 13.4; p < 0.001) from 2004 to 2017. In comparison, among urban dually enrolled beneficiaries, declines were more pronounced, from 86.2 to 68.6 per 1,000 beneficiaries (difference: 17.6; 95% CI: 17.2, 17.9; p < 0.001) from 2004 to 2017 (data not shown). After all-cause mortality rates were standardized for age and sex, mortality declined among the rural dually enrolled, from 96.6 to 92.7 per 1,000 beneficiaries (difference:, 3.9; 95% CI: 3.0, 4.7; p < 0.001), as well as among the urban dually enrolled, from 86.9 to 72.8 per 1,000 beneficiaries (difference: 14.1; 95% CI: 13.7, 14.5; p < 0.001), from 2004 to 2017 (exhibit 2). The relative percentage decline in all-cause mortality for rural and urban dually enrolled beneficiaries is shown in exhibit 3. The difference in all-cause mortality between rural and urban dually enrolled beneficiaries widened during the study period (p < 0.001 for the difference in rural versus urban change over time).

EXHIBIT 2. Age-sex standardized all-cause mortality rates per 1,000 dually enrolled Medicare beneficiaries in rural and urban areas of the US, 2004–17.

source Authors’ analysis of dually enrolled Medicare (fee-for-service and Medicare Advantage) beneficiaries, 2004–17, using rural-urban commuting area codes and data from the Centers for Medicare and Medicaid Services Master Beneficiary Summary files. notes All-cause mortality rates after standardization for age and sex. The interaction between rural/urban status and all-cause mortality over time was statistically significant (p < 0.001).

EXHIBIT 3.

Change in age- and sex-standardized all-cause mortality among rural and urban dually enrolled Medicare beneficiaries, by census region, 2004 and 2017

	Rural			Urban
Region (states)	2004	2017	Relative percent change	2004	2017	Relative percent change
Overall	96.6	92.7	−4.0	86.9	72.8	−16.2
New England (CT, ME, MA, NH, RI, VT)	90.0	74.2	−17.6	81.4	66.3	−18.6
Mid-Atlantic (NJ, NY, PA)	90.5	79.7	−11.9	73.5	59.9	−18.5
East North Central (IL, IN, MI, OH, WI)	114.9	108.9	−5.2	109.6	92.0	−16.1
West North Central (IA, KS, MN, MO, NE, ND, SD)	100.4	103.9	+ 3.5	101.1	94.4	−6.6
South Atlantic (DE, FL, GA, MD, NC, SC, VA, DC, WV)	97.1	94.1	−3.1	99.0	81.8	−17.4
East South Central (AL, KY, MS, TN)	90.0	91.8	+ 2.0	97.5	92.1	−5.5
West South Central (AR, LA, OK, TX)	88.8	87.9	−1.0	83.6	71.7	−14.2
Mountain (AZ, CO, ID, MT, NV, NM, UT, WY)	98.1	88.4	−9.9	99.6	84.3	−15.4
Pacific (AK, CA, HI, OR, WA)	99.3	86.9	−12.5	73.4	59.7	−18.7

source Authors’analysis of dually enrolled Medicare (fee-for-service and Medicare Advantage) beneficiaries, 2004–17, using rural-urban commuting area codes and data from the Centers for Medicare and Medicaid Services Medicare Master Beneficiary Summary Files. note All p values were <0.001 for difference in rural versus urban change over time except for the Mid-Atlantic (p = 0.17), New England (p = 0.80), Pacific (p = 0.35), and Mountain (p = 0.007) regions.

All-cause mortality varied geographically across the nine US census regions (exhibit 3; appendix exhibit 4).²¹ In 2004 mortality rates were highest for rural dually enrolled beneficiaries in the East North Central region (114.9 per 1,000 beneficiaries) and lowest in the West South Central region (88.8 per 1,000 beneficiaries). By 2017 mortality rates for the rural dually enrolled remained highest in the East North Central region (108.9 per 1,000 beneficiaries) but were lowest in New England (74.2 per 1,000 beneficiaries). Between 2004 and 2017 mortality among rural dually enrolled beneficiaries increased modestly in West North Central (from 100.4 to 103.9 per 1,000 beneficiaries; p = 0.03) and East South Central (from 90.0 to 91.8 per 1,000 beneficiaries; p = 0.06) states. Exhibit 3 shows the relative percentage change in all-cause mortality for rural and urban dually enrolled beneficiaries by US census region. The difference in all-cause mortality between rural and urban dually enrolled beneficiaries increased significantly (p < 0.001 for difference in rural versus urban change over time) from 2004 to 2017 in all US census regions except the Mid-Atlantic, New England, Pacific, and Mountain regions.

In our subgroup analysis of dually enrolled beneficiaries stratified by race, age- and sex-standardized all-cause mortality rates for White rural beneficiaries declined from 102.2 to 98.2 per 1,000 beneficiaries (difference: 4.0; 95% CI: 3.0, 5.0; p < 0.001) between 2004 and 2017, whereas all-cause mortality for Black rural beneficiaries declined from 80.7 to 78.5 per 1,000 beneficiaries (difference: 2.2; 95% CI: 0.13, 4.2; p < 0.001). In comparison, declines in all-cause mortality were more pronounced for both White (from 101.3 to 87.6 per 1,000 beneficiaries; difference: 13.7; 95% CI: 13.1,14.3; p < 0.001) and Black (from 88.2 to 75.4 per 1,000 beneficiaries; difference: 12.8; 95% CI: 11.9, 13.7; p < 0.001) urban beneficiaries from 2004 to 2017 (exhibit 4).

EXHIBIT 4. Age-sex standardized all-cause mortality rates per 1,000 dually enrolled Medicare beneficiaries in rural and urban areas in the US, by race, 2004–17.

source Authors’ analysis of dually enrolled Medicare (fee-for-service and Medicare Advantage) beneficiaries, 2004–17, using rural-urban commuting area codes and data from the Centers for Medicare and Medicaid Services Medicare Master Beneficiary Summary Files. note The exhibit shows all-cause mortality rates by race (Black or White) after standardization for age and sex.

Discussion

In this study of dually enrolled Medicare beneficiaries older than age sixty-five, rural beneficiaries had higher all-cause mortality rates than their urban counterparts. Although all-cause mortality rates decreased overall for both groups from 2004 to 2017, declines among rural beneficiaries were more modest and stalled in more recent years, resulting in an increased gap in mortality. There was also substantial geographic variation in all-cause mortality rates among rural dually enrolled beneficiaries. Rural mortality rates were highest in East North Central states and increased modestly in West North Central states during the study period.

In recent years there has been increasing concern about the health of rural communities in the US. Approximately 60 million Americans currently reside in rural areas,²⁴ and these people experience worse outcomes than their urban counterparts, including higher rates of death from heart disease, cancer, and unintentional injuries.^10,25,26 These disparities have been attributed to the complex social, behavioral, and structural factors that rural residents face, which result in disproportionally higher rates of poverty, chronic disease, and fragmented medical care.^{3,9,10,25,27,28} Our study extends this evidence by focusing specifically on low-income (dually enrolled) Medicare beneficiaries and demonstrates an alarming, growing gap in all-cause mortality between those living in rural and in urban areas.

We found that there was marked geographic variation in patterns and trends in all-cause mortality rates in the dually enrolled population. For example, rural beneficiaries living in the East North Central region persistently experienced the highest mortality rates throughout the study period. Further, although mortality in urban areas decreased in West North Central states, this region experienced a modest increase in rural mortality. These patterns may be explained, at least in part, by several factors, including the opioid epidemic and the growing shortage of health care providers that has primarily affected rural and low-income settings in midwestern states.^29–32 In addition, mortality generally did not improve in the South, an area that has experienced a rapid rise in chronic diseases and has disproportionately been affected by rural hospital closures.^11,33 In contrast, New England, a region that includes states with generally high-performing health care systems, experienced the sharpest declines in rural mortality during the study period.³⁴ Overall, these findings indicate that local and regional public health efforts are needed to improve health outcomes for low-income Medicare beneficiaries residing in rural areas.

Federal policy efforts that aim to improve the health of dually enrolled Medicare beneficiaries have intensified in recent years. The Centers for Medicare and Medicaid Services has implemented integrated care models, such as the Program of All-Inclusive Care for the Elderly, that specifically target this population.⁸ However, these initiatives, which were developed based on urban infrastructure, have achieved limited success in rural areas because of challenges associated with implementation.^35–37 In addition, home-based services and nursing home facilities also remain limited within rural areas, often leaving dually enrolled beneficiaries living in these communities without reliable access to skilled medical care.^38–40 As a result, rural dually enrolled beneficiaries continue to disproportionately experience worse access to medical services, as well as uncoordinated care, that may negatively affect their outcomes. Therefore, policy efforts that specifically target these beneficiaries may be needed to improve care delivery and long-term health.

Conclusion

In the dually enrolled Medicare (fee-for-service and Medicare Advantage) population older than age sixty-five, beneficiaries living in rural areas had higher all-cause mortality rates than their urban counterparts. Although all-cause mortality rates decreased for both groups from 2004 to 2017, declines among rural beneficiaries lagged behind those living in urban areas and have stalled during the last several years, resulting in a widening rural-urban gap in mortality. Given increasing inequities in health outcomes for low-income Medicare beneficiaries living in rural areas, there is an urgent need for public health and policy efforts to focus on improving access to care, care delivery, and the health of this vulnerable population. ■