Mortality and Hospitalizations for Dually Enrolled and Nondually Enrolled Medicare Beneficiaries Aged 65 Years or Older, 2004 to 2017 | Geriatrics | JAMA | JAMA Network
[Skip to Navigation]
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
Purchase Options:
[Skip to Navigation Landing]
Figure 1.  County-Level Variation in the Proportion of Beneficiaries Aged 65 Years or Older Dually Enrolled in Medicare and Medicaid
County-Level Variation in the Proportion of Beneficiaries Aged 65 Years or Older Dually Enrolled in Medicare and Medicaid

The decline in dual enrollment in the southeastern United States may reflect the implementation of stricter requirements to obtain Medicaid benefits in some states.

Figure 2.  Annual All-Cause Mortality Rates Among Dually and Nondually Enrolled Medicare Beneficiaries Aged 65 Years or Older
Annual All-Cause Mortality Rates Among Dually and Nondually Enrolled Medicare Beneficiaries Aged 65 Years or Older

Mortality rates were adjusted for age, sex, and race. Lines were smoothed using the Loess method. The shaded areas around the curves indicate 95% CIs. Beneficiaries were dually enrolled in Medicare and Medicaid or nondually enrolled in Medicare only.

Figure 3.  County-Level Variation in All-Cause Mortality Among Beneficiaries Aged 65 Years or Older Dually Enrolled in Medicare and Medicaid
County-Level Variation in All-Cause Mortality Among Beneficiaries Aged 65 Years or Older Dually Enrolled in Medicare and Medicaid

The panels show county-level variation in annual age-, sex-, and race-standardized all-cause mortality.

Figure 4.  All-Cause Hospitalization Rates Among Dually Enrolled and Nondually Enrolled Medicare Beneficiaries Aged 65 Years or Older
All-Cause Hospitalization Rates Among Dually Enrolled and Nondually Enrolled Medicare Beneficiaries Aged 65 Years or Older

The all-cause hospitalization rates were adjusted for age, sex, and race. All hospitalizations per beneficiary within each calendar year were included. Lines were smoothed using the Loess method. The shaded areas around the curves indicate 95% CIs. Beneficiaries were dually enrolled in Medicare and Medicaid or nondually enrolled in Medicare only.

Figure 5.  Risk-Adjusted Hospitalization-Related Mortality Among Dually and Nondually Enrolled Medicare Beneficiaries Aged 65 Years or Older
Risk-Adjusted Hospitalization-Related Mortality Among Dually and Nondually Enrolled Medicare Beneficiaries Aged 65 Years or Older

Hospitalization-related mortality was adjusted for patient demographic characteristics (age, sex, and race) and clinical comorbidities. For beneficiaries with multiple hospitalizations within a calendar year, 1 hospitalization was randomly selected. Lines were smoothed using the Loess method. The shaded areas around the curves indicate 95% CIs. Beneficiaries were dually enrolled in Medicare and Medicaid or nondually enrolled in Medicare only.

Table.  Characteristics of Hospitalized Medicare Beneficiaries Aged 65 Years or Older Dually Enrolled in Medicare and Medicaid or Nondually Enrolled in Medicare Only
Characteristics of Hospitalized Medicare Beneficiaries Aged 65 Years or Older Dually Enrolled in Medicare and Medicaid or Nondually Enrolled in Medicare Only
1.
US Census Bureau. Gini index of money income and equivalence-adjusted income: 1967 to 2014. https://www.census.gov/content/dam/Census/library/visualizations/2015/demo/gini-index-equil-adj-inc.pdf. Accessed November 21, 2019.
2.
Chetty  R, Stepner  M, Abraham  S,  et al.  The association between income and life expectancy in the United States, 2001-2014.  JAMA. 2016;315(16):1750-1766.PubMedGoogle ScholarCrossref
3.
Bor  J, Cohen  GH, Galea  S.  Population health in an era of rising income inequality: USA, 1980-2015.  Lancet. 2017;389(10077):1475-1490.PubMedGoogle ScholarCrossref
5.
Office of the Assistant Secretary for Planning and Evaluation; Department of Health and Human Services. Report to Congress: social risk factors and performance under Medicare’s value-based purchasing programs. https://aspe.hhs.gov/pdf-report/report-congress-social-risk-factors-and-performance-under-medicares-value-based-purchasing-programs. Accessed April 11, 2019.
6.
Congressional Budget Office. Dual-eligible beneficiaries of Medicare and Medicaid. https://www.cbo.gov/publication/44308. Accessed April 11, 2019.
7.
Elixhauser  A, Steiner  C, Harris  DR,  et al.  Comorbidity measures for use with administrative data.  Med Care. 1998;36(1):8-27.PubMedGoogle ScholarCrossref
8.
Tsugawa  Y, Figueroa  JF, Papanicolas  I,  et al.  Assessment of strategies for managing expansion of diagnosis coding using risk-adjustment methods for Medicare data.  JAMA Intern Med. 2019;179(9):1287-1290.PubMedGoogle ScholarCrossref
9.
Downing  NS, Wang  C, Gupta  A,  et al.  Association of racial and socioeconomic disparities with outcomes among patients hospitalized with acute myocardial infarction, heart failure, and pneumonia.  JAMA Netw Open. 2018;1(5):e182044.PubMedGoogle Scholar
10.
Zhang  J, Yu  KF.  What’s the relative risk?  JAMA. 1998;280(19):1690-1691.PubMedGoogle ScholarCrossref
11.
Centers for Medicare & Medicaid Services. Technical documentation for the Medicare Current Beneficiary Survey. https://www.cms.gov/Research-Statistics-Data-and-Systems/Research/MCBS/Downloads/OLD/HHC2003appendixA.pdf. Accessed November 29, 2019.
12.
Huh  J, Reif  J.  Did Medicare Part D reduce mortality?  J Health Econ. 2017;53:17-37.PubMedGoogle ScholarCrossref
13.
Afendulis  CC, He  Y, Zaslavsky  AM,  et al.  The impact of Medicare Part D on hospitalization rates.  Health Serv Res. 2011;46(4):1022-1038.PubMedGoogle ScholarCrossref
14.
Figueroa  JF, Tsugawa  Y, Zheng  J,  et al.  Association between the value-based purchasing pay for performance program and patient mortality in US hospitals.  BMJ. 2016;353:i2214.PubMedGoogle ScholarCrossref
15.
Wadhera  RK, Yeh  RW, Joynt Maddox  KE.  The hospital readmissions reduction program.  N Engl J Med. 2019;380(24):2289-2291.PubMedGoogle ScholarCrossref
16.
Ryan  AM, Krinsky  S, Maurer  KA,  et al.  Changes in hospital quality associated with hospital value-based purchasing.  N Engl J Med. 2017;376(24):2358-2366.PubMedGoogle ScholarCrossref
17.
Sankaran  R, Sukul  D, Nuliyalu  U,  et al.  Changes in hospital safety following penalties in the US Hospital Acquired Condition Reduction Program.  BMJ. 2019;366:l4109.PubMedGoogle ScholarCrossref
18.
Roberts  ET, Zaslavsky  AM, McWilliams  JM.  The value-based payment modifier.  Ann Intern Med. 2018;168(4):255-265.PubMedGoogle ScholarCrossref
19.
Wadhera  RK, Joynt Maddox  KE, Wasfy  JH,  et al.  Association of the hospital readmissions reduction program with mortality among Medicare beneficiaries hospitalized for heart failure, acute myocardial infarction, and pneumonia.  JAMA. 2018;320(24):2542-2552.PubMedGoogle ScholarCrossref
20.
Joynt  KE, Jha  AK.  Characteristics of hospitals receiving penalties under the Hospital Readmissions Reduction Program.  JAMA. 2013;309(4):342-343.PubMedGoogle ScholarCrossref
21.
Chen  LM, Epstein  AM, Orav  EJ,  et al.  Association of practice-level social and medical risk with performance in the Medicare physician value-based payment modifier program.  JAMA. 2017;318(5):453-461.PubMedGoogle ScholarCrossref
22.
Joynt Maddox  KE.  Financial incentives and vulnerable populations.  N Engl J Med. 2018;378(11):977-979.PubMedGoogle ScholarCrossref
23.
Joynt  KE, De Lew  N, Sheingold  SH,  et al.  Should Medicare value-based purchasing take social risk into account?  N Engl J Med. 2017;376(6):510-513.PubMedGoogle ScholarCrossref
24.
Verma  S. Better care for people dually eligible for Medicare and Medicaid. https://www.healthaffairs.org/do/10.1377/hblog20190423.701475/full/. Accessed June 24, 2019.
25.
Figueroa  JF, Zhou  X, Jha  AK.  Characteristics and spending patterns of persistently high-cost Medicare patients.  Health Aff (Millwood). 2019;38(1):107-114.PubMedGoogle ScholarCrossref
26.
Figueroa  JF, Lyon  Z, Zhou  X,  et al.  Persistence and drivers of high-cost status among dual-eligible Medicare and Medicaid beneficiaries.  Ann Intern Med. 2018;169(8):528-534.PubMedGoogle ScholarCrossref
27.
Medpac. Managed care plans for dual-eligible beneficiaries: report to the Congress: Medicare and the health care delivery system. http://medpac.gov/docs/default-source/reports/jun18_ch9_medpacreport_sec.pdf. Accessed June 23, 2019.
28.
Grabowski  DC.  Special needs plans and the coordination of benefits and services for dual eligibles.  Health Aff (Millwood). 2009;28(1):136-146.PubMedGoogle ScholarCrossref
30.
Ku  L, Steinmetz  E. Bridging the gap: continuity and quality of coverage in Medicaid association for community affilitated plans. https://ccf.georgetown.edu/wp-content/uploads/2013/09/GW-Continuity-Report-9-10-13.pdf. Accessed June 24, 2019.
31.
Riley  GF, Zhao  L, Tilahun  N.  Understanding factors associated with loss of medicaid coverage among dual eligibles can help identify vulnerable enrollees.  Health Aff (Millwood). 2014;33(1):147-152.PubMedGoogle ScholarCrossref
32.
Jacobson  GA, Neuman  P, Damico  A.  At least half of new Medicare advantage enrollees had switched from traditional Medicare during 2006-11.  Health Aff (Millwood). 2015;34(1):48-55.PubMedGoogle ScholarCrossref
Original Investigation
March 10, 2020

Mortality and Hospitalizations for Dually Enrolled and Nondually Enrolled Medicare Beneficiaries Aged 65 Years or Older, 2004 to 2017

Author Affiliations
  • 1Richard A. and Susan F. Smith Center for Outcomes Research in Cardiology, Division of Cardiology, Beth Israel Deaconess Medical Center, Harvard Medical School, Harvard University, Boston, Massachusetts
  • 2Department of Biostatistics, T. H. Chan School of Public Health, Harvard University, Boston, Massachusetts
  • 3Department of Health Policy and Management, T. H. Chan School of Public Health, Harvard University, Boston, Massachusetts
  • 4Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School, Harvard University, Boston, Massachusetts
  • 5Center for Health Economics and Policy, Institute for Public Health and Cardiovascular Division, School of Medicine, Washington University in St Louis, St Louis, Missouri
JAMA. 2020;323(10):961-969. doi:10.1001/jama.2020.1021
Key Points

Question  Did mortality or hospitalization rates differ between beneficiaries dually enrolled in Medicare and Medicaid compared with nondually enrolled (Medicare only) beneficiaries aged 65 years or older from 2004 to 2017, and did these differences decrease over time?

Findings  In this serial cross-sectional study that included 71 017 608 Medicare fee-for-service beneficiaries, dually enrolled beneficiaries compared with nondually enrolled beneficiaries had higher all-cause mortality (odds ratio, 2.22), all-cause hospitalizations (41 121 vs 22 601 per 100 000 beneficiary-years), and hospitalization-related 30-day mortality (odds ratio, 1.15) in 2017. Between 2004 and 2017, these differences did not decrease.

Meaning  Between 2004 and 2017, there were persistent differences in mortality and hospitalizations between dually enrolled beneficiaries and nondually enrolled beneficiaries.

Abstract

Importance  Medicare beneficiaries who are also enrolled in Medicaid (dually enrolled beneficiaries) have drawn the attention of policy makers because they comprise the poorest subset of the Medicare population; however, it is unclear how their outcomes have changed over time compared with those only enrolled in Medicare (nondually enrolled beneficiaries).

Objective  To evaluate annual changes in all-cause mortality, hospitalization rates, and hospitalization-related mortality among dually enrolled beneficiaries and nondually enrolled beneficiaries.

Design, Setting, and Participants  Serial cross-sectional study of Medicare fee-for-service beneficiaries aged 65 years or older between January 2004 and December 2017. The final date of follow-up was September 30, 2018.

Exposures  Dual vs nondual enrollment status.

Main Outcomes and Measures  Annual all-cause mortality rates; all-cause hospitalization rates; and in-hospital, 30-day, 1-year hospitalization-related mortality rates.

Results  There were 71 017 608 unique Medicare beneficiaries aged 65 years or older (mean age, 75.6 [SD, 9.2] years; 54.9% female) enrolled in Medicare for at least 1 month from 2004 through 2017. Of these beneficiaries, 11 697 900 (16.5%) were dually enrolled in Medicare and Medicaid for at least 1 month. After adjusting for age, sex, and race, annual all-cause mortality rates declined from 8.5% (95% CI, 8.45%-8.56%) in 2004 to 8.1% (95% CI, 8.05%-8.13%) in 2017 among dually enrolled beneficiaries and from 4.1% (95% CI, 4.08%-4.13%) in 2004 to 3.8% (95% CI, 3.76%-3.79%) in 2017 among nondually enrolled beneficiaries. The difference in annual all-cause mortality between dually and nondually enrolled beneficiaries increased between 2004 (adjusted odds ratio, 2.09 [95% CI, 2.08-2.10]) and 2017 (adjusted odds ratio, 2.22 [95% CI, 2.21-2.23]) (P < .001 for interaction between dual enrollment status and time). All-cause hospitalizations per 100 000 beneficiary-years declined from 49 888 in 2004 to 41 121 in 2017 among dually enrolled beneficiaries (P < .001) and from 29 000 in 2004 to 22 601 in 2017 among nondually enrolled beneficiaries (P < .001); however, the difference between these groups widened between 2004 (adjusted risk ratio, 1.72 [95% CI, 1.71-1.73]) and 2017 (adjusted risk ratio, 1.83 [95% CI, 1.82-1.83]) (P < .001 for interaction). Among hospitalized beneficiaries, the risk-adjusted 30-day mortality rates declined from 10.3% (95% CI, 10.22%-10.33%) in 2004 to 10.1% (95% CI, 10.02%-10.20%) in 2017 among dually enrolled beneficiaries and from 8.5% (95% CI, 8.50%-8.56%) in 2004 to 8.1% (95% CI, 8.06%-8.13%) in 2017 among nondually enrolled beneficiaries. In contrast, 1-year mortality increased among hospitalized beneficiaries from 23.1% (95% CI, 23.05%-23.20%) in 2004 to 26.7% (95% CI, 26.58%-26.84%) in 2017 among dually enrolled beneficiaries and from 18.1% (95% CI, 18.11%-18.17%) in 2004 to 20.3% (95% CI, 20.21%-20.31%) in 2017 among nondually enrolled beneficiaries. The difference in hospitalization-related outcomes between dually and nondually enrolled beneficiaries persisted during the study period.

Conclusions and Relevance  Among Medicare fee-for-service beneficiaries aged 65 years or older, dually enrolled beneficiaries had higher annual all-cause mortality, all-cause hospitalizations, and hospitalization-related mortality compared with nondually enrolled beneficiaries. Between 2004 and 2017, these differences did not decrease.

Introduction

In the United States, income inequality has steadily increased since the 1970s.1 As inequities have widened, there has been increasing concern about the health outcomes of individuals experiencing poverty, given the strong link between socioeconomic status and health.2,3 In the Medicare program, more than 7 million individuals aged 65 years or older are also dually enrolled in Medicaid due to poverty.4 Dually enrolled beneficiaries have a higher burden of chronic disease, worse clinical outcomes, and greater overall Medicare spending levels compared with nondually enrolled (Medicare only) beneficiaries.5,6 As a result, the population of dually enrolled beneficiaries has drawn the attention of policy makers, particularly as efforts to reduce health inequity have intensified in the United States.

However, little is known about whether health outcomes have improved for dually enrolled beneficiaries compared with nondually enrolled beneficiaries. One possibility is that over time, changes to the Medicare program and a greater policy focus on improving the delivery and coordination of inpatient and ambulatory care have disproportionately benefited dually enrolled beneficiaries. Alternatively, in the face of increasing socioeconomic inequality, major differences in health outcomes between dually enrolled beneficiaries and nondually enrolled beneficiaries could remain or may have even widened.

Therefore, this study aimed to answer 3 questions. Did the annual all-cause mortality rates among dually enrolled beneficiaries aged 65 years or older change between 2004 and 2017? How did hospitalization rates, hospitalization-related mortality, and hospitalization-related expenditures change during this period? Have the differences in mortality between dually enrolled beneficiaries and nondually enrolled beneficiaries narrowed or widened over time?

Methods

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

Study Population

Centers for Medicare & Medicaid Services denominator files were used to identify individuals aged 65 years or older enrolled for at least 1 month in the Medicare fee-for-service program between January 2004 and December 2017. These files provide information on demographic characteristics, enrollment status, and mortality. Medicare beneficiaries were considered dually enrolled if they were also enrolled in Medicaid for at least 1 month during a given year. We excluded beneficiaries who were aged 64 years or younger and those with end-stage kidney disease.

We linked the Medicare denominator files to the Medicare inpatient files to identify fee-for-service beneficiaries with at least 1 hospitalization and who were admitted to a short-term acute care hospital in the United States during the study period. Patient comorbidities were characterized using the Elixhauser comorbidity index,7 and were obtained from the secondary diagnosis codes recorded during the index hospitalization. Because the maximum number of diagnosis codes allowed for Medicare increased from 10 to 25 in 2011, we restricted the 2011-2017 data to the first 10 diagnosis codes, which has been done in prior studies.8

Race/ethnicity was classified based on self-report using categories specified by Medicare at the time of Medicare enrollment. Race/ethnicity was included as a covariate in the analysis because it is associated with mortality.9

Study Outcomes

The primary outcomes were (1) annual all-cause mortality rates; (2) all-cause hospitalization rates; and (3) in-hospital, 30-day, and 1-year hospitalization-related mortality rates among Medicare beneficiaries by dual enrollment status. The final date of follow-up was September 30, 2018. Medicare denominator files were used to calculate annual all-cause mortality rates among Medicare beneficiaries. All-cause hospitalization rates for Medicare beneficiaries were calculated using beneficiary-years of enrollment in fee-for-service as a denominator.

In addition, among Medicare beneficiaries hospitalized during each year, in-hospital, 30-day, and 1-year mortality rates were calculated. Trends in hospital length of stay and discharge disposition (home care vs long-term care or skilled nursing facility) also were evaluated. To measure Medicare inpatient expenditures, we determined the total annual Medicare inpatient reimbursements, adjusting for inflation using the consumer price index with 2017 as the index year.

Statistical Analysis

We fit a mixed-effects model with a Poisson link function and state-specific random intercepts to evaluate annual changes in all-cause mortality by dual enrollment status among Medicare beneficiaries (adjusted for age, sex, and race). The total number of Medicare beneficiaries was used as an offset. The model incorporated dummy variables representing each year to examine yearly changes in mortality rates. The baseline year of 2004 served as the reference for each subsequent year. The coefficients of dummy variables in the model for mortality represent changes in the risk ratio (RR) relative to 2004. The adjusted mortality rates were then calculated using the relative RRs for each year and represent what the rate would have been if the case mix of patients was identical to 2004.10

We fit the same model to evaluate changes in all-cause hospitalization rates with the beneficiary year as the offset. To evaluate hospitalization-related mortality, we fit a mixed model with logit-link function and adjusted this outcome for age, sex, race, and clinical comorbidities. For beneficiaries with multiple hospitalizations within a calendar year, we randomly selected 1 hospitalization per beneficiary per year. To compare the difference in clinical outcomes between dually enrolled beneficiaries and nondually enrolled beneficiaries at the beginning and end of the study period, we fit a mixed model to calculate the adjusted RR (for hospitalizations) or odds ratio (OR; for mortality) during 2004 and 2017 separately. To assess if the difference in outcomes changed over time, we also fit the mixed model with an interaction term for dual enrollment status and time (2004 and 2017).

To evaluate geographic variation in mortality between dually enrolled beneficiaries and nondually enrolled beneficiaries, we used a Poisson link function and county-specific random intercepts to model the number of deaths as a function of age, sex, and race while accounting for geographic differences between counties. Using this model, we calculated the rates of demographics-standardized all-cause mortality for each county or county equivalent during 2004 and 2017. The county-specific demographics-standardized rates were then mapped (the lowest rates to the highest rates).

We conducted supplemental analyses using Medicare Current Beneficiary Survey files to examine fully risk-adjusted trends in all-cause mortality among dually enrolled beneficiaries and nondually enrolled beneficiaries after accounting for differences in the clinical characteristics of these groups. The Medicare Current Beneficiary Survey is a nationally representative survey of Medicare beneficiaries that uses multiple in-person interviews to collect detailed information on demographic variables, clinical comorbidities, health status, and functional status, the latter 2 of which are not available in the Medicare denominator files.11

We fit a mixed model with a logit-link function and state-specific random intercepts to evaluate the difference in all-cause mortality between dually enrolled beneficiaries and nondually enrolled beneficiaries after accounting for demographic characteristics (age, sex, and race), clinical comorbidities, functional status, and health status (variables appear in the eText in the Supplement). In the model, we included 3 dummy variables to represent dual enrollment status, year (2004-2013), and the dual enrollment status × time interaction.

The Medicare enrollment files and inpatient data were used to crosswalk any missing demographic information for beneficiaries during the study period. No data were missing for enrollment status (dual or nondual). Bonferroni adjustment was applied to the tests of temporal trends for the clinical outcomes and the difference in clinical outcomes between dually enrolled beneficiaries and nondually enrolled beneficiaries to ensure a family-wise type I error rate of 0.05 (2-sided). The analyses were conducted using SAS version 9.4 (SAS Institute Inc).

Results
Study Cohort

Overall, there were 71 017 608 unique Medicare beneficiaries aged 65 years or older (mean age, 75.6 [SD, 9.2] years; 54.9% female) enrolled in the Medicare program for at least 1 month between January 2004 and December 2017. Of these beneficiaries, 11 697 900 (16.5%), representing 72 193 078 beneficiary-years of enrollment, were dually enrolled for at least 1 month. Dually enrolled beneficiaries were older compared with nondually enrolled beneficiaries (mean age, 76.5 [SD, 8.5] years vs 74.9 [SD, 8.2] years, respectively) and more likely to be female (67.9% vs 54.3%).

The proportion of Medicare beneficiaries who were dually enrolled was 13.2% in 2004 and 11.9% in 2017. Between 2004 and 2017, the mean age of dually enrolled beneficiaries declined from 77.1 (SD, 8.3) years to 75.6 (8.7) years. In addition, there were declines between 2004 and 2017 in the proportion of dually enrolled beneficiaries who were female (from 70.5% in 2004 to 64.7% in 2017; P < .001), white (from 61.8% to 59.0%; P < .001), and black (from 18.9% to 16.7%; P < .001). These patterns were generally similar for nondually enrolled beneficiaries between 2004 and 2017 (from a mean age of 75.4 [SD, 8.0] years in 2004 to 74.2 [SD, 8.1] years in 2017; female, from 55.8% to 53.1%; white, from 90.0% to 85.8%); however, the proportion of black beneficiaries increased between 2004 and 2017 (from 6.5% to 6.9%; P < .001).

Marked geographic variation in the distribution of dually enrolled beneficiaries was observed at the county level and ranged from less than 0.01% to 54.7% in 2004. Medicare beneficiaries who resided in the southern United States were more likely to be dually enrolled, which is a pattern that persisted in 2017 (Figure 1).

All-Cause Mortality

From 2004-2017, dually enrolled beneficiaries had higher observed annual all-cause mortality rates compared with nondually enrolled beneficiaries (7.8% vs 3.8%, respectively; P < .001). Annual all-cause mortality rates declined for dually enrolled beneficiaries (from 8.5% in 2004 to 7.3% in 2017; difference, 1.2 percentage points [95% CI, 1.15-1.22 percentage points]) and nondually enrolled beneficiaries (from 4.1% in 2004 to 3.5% in 2017; difference, 0.62 percentage points [95% CI, 0.61-0.63 percentage points]) (eFigure 1 in the Supplement).

Age-, sex-, and race-adjusted all-cause mortality among dually enrolled beneficiaries declined from 8.5% (95% CI, 8.45%-8.56%) in 2004 to 8.1% (95% CI, 8.05%-8.13%) in 2017 (Figure 2). Among nondually enrolled beneficiaries, adjusted all-cause mortality declined from 4.1% (95% CI, 4.08%-4.13%) in 2004 to 3.8% (95% CI, 3.76%-3.79%) in 2017. The difference in all-cause mortality between dually enrolled beneficiaries and nondually enrolled beneficiaries increased between 2004 (adjusted OR, 2.09 [95% CI, 2.08-2.10]) and 2017 (adjusted OR, 2.22 [95% CI, 2.21-2.23]) (P < .001 for interaction between dual enrollment status and time).

There was marked geographic variation in all-cause mortality among dually enrolled beneficiaries at the county level in 2004 and ranged from 5.1% to 10.7%. This pattern persisted in 2017 and ranged from 3.5% to 9.0% (Figure 3). The weighted Pearson correlation coefficient for county-specific all-cause mortality in 2004 vs 2017 was 0.76 (95% CI, 0.74-0.77).

All-Cause Hospitalizations

From 2004-2017, there were 34 273 533 unique Medicare beneficiaries aged 65 years or older who were hospitalized of whom 18.6% were dually enrolled. Patient demographic characteristics and clinical comorbidities appear in the Table. Dually enrolled beneficiaries had higher hospitalization rates compared with nondually enrolled beneficiaries and this pattern persisted during the study period. The age-, sex-, and race-adjusted all-cause hospitalizations per 100 000 beneficiary-years declined for dually enrolled beneficiaries (from 49 888 in 2004 to 41 121 in 2017; P < .001) and nondually enrolled beneficiaries (from 29 000 in 2004 to 22 601 in 2017; P < .001) (Figure 4).

The difference in all-cause hospitalizations between dually enrolled beneficiaries and nondually enrolled beneficiaries increased between 2004 (adjusted RR, 1.72 [95% CI, 1.71-1.73]) and 2017 (adjusted RR, 1.83 [95% CI, 1.82-1.83]) (P < .001 for interaction between dual enrollment status and time). These patterns were similar in the evaluation of the number of dually enrolled beneficiaries and nondually enrolled beneficiaries with at least 1 hospitalization (eFigure 2 in the Supplement). The proportion of hospitalized dually enrolled beneficiaries and nondually enrolled beneficiaries with 1, 2, 3, or 4 or more hospitalizations for 2004 and 2017 appears in eTable 1 in the Supplement.

Hospitalization-Related Mortality

Dually enrolled beneficiaries had higher observed in-hospital, 30-day, and 1-year mortality rates compared with nondually enrolled beneficiaries (eFigure 3 in the Supplement). After adjusting for patient demographic characteristics (age, sex, and race) and clinical comorbidities, in-hospital mortality rates declined for dually enrolled beneficiaries (from 5.5% [95% CI, 5.46%-5.55%] in 2004 to 4.1% [95% CI, 4.01%-4.12%] in 2017) and nondually enrolled beneficiaries (4.7% [95% CI, 4.68%-4.72%] in 2004 to 3.3% [95% CI, 3.29%-3.34%] in 2017) (Figure 5). In-hospital mortality was higher for dually enrolled beneficiaries compared with nondually enrolled beneficiaries in 2004 (adjusted OR, 1.08 [95% CI, 1.07-1.10]) and in 2017 (adjusted OR, 1.12 [95% CI, 1.10-1.13]).

Risk-adjusted 30-day mortality rates declined among dually enrolled beneficiaries (from 10.3% [95% CI, 10.22%-10.33%] in 2004 to 10.1% [95% CI, 10.02%-10.20%] in 2017) and among nondually enrolled beneficiaries (from 8.5% [95% CI, 8.50%-8.56%] in 2004 to 8.1% [95% CI, 8.06%-8.13%] in 2017). Dually enrolled beneficiaries had higher 30-day mortality rates compared with nondually enrolled beneficiaries in 2004 (adjusted OR, 1.14 [95% CI, 1.13-1.15]) and in 2017 (adjusted OR, 1.15 [95% CI, 1.14-1.16]).

Risk-adjusted 1-year mortality rates increased among dually enrolled beneficiaries (from 23.1% [95% CI, 23.05%-23.20%] in 2004 to 26.7% [95% CI, 26.58%-26.84%] in 2017) and among nondually enrolled beneficiaries (from 18.1% [95% CI, 18.11%-18.17%] in 2004 to 20.3% [95% CI, 20.21%-20.31%] in 2017). One-year mortality after hospitalization was higher among dually enrolled beneficiaries compared with nondually enrolled beneficiaries in 2004 (adjusted OR, 1.28 [95% CI, 1.27-1.28]) and remained higher in 2017 (adjusted OR, 1.28 [95% CI, 1.27-1.28]).

Hospitalization-Related Medicare Expenditures, Hospital Length of Stay, and Hospital Discharge Disposition

From 2004 through 2017, Medicare expenditures that were adjusted for the consumer price index increased, whereas hospital length of stay decreased. The median Medicare expenditures per hospitalization for dually enrolled beneficiaries were lower ($6704 [interquartile range, $4799-$10 533] in 2004 and $8406 [interquartile range, $5582-$12 465] in 2017) compared with nondually enrolled beneficiaries ($6892 [interquartile range, $4738-$11 860] in 2004 and $8704 [interquartile range, $5460-$13 019] in 2017).

However, total annual hospitalization-related expenditures per Medicare beneficiary were higher for dually enrolled beneficiaries compared with nondually enrolled beneficiaries and increased during the study period (eFigure 4 in the Supplement). Dually enrolled beneficiaries had a longer hospital length of stay (mean, 6.1 days [SD, 7.0 days] in 2004 and 5.7 days [SD, 6.8 days] in 2017) compared with nondually enrolled beneficiaries (mean, 5.6 days [SD, 6.1 days] in 2004 and 5.0 days [SD, 5.4 days] in 2017). Dually enrolled beneficiaries were more likely to be discharged to a skilled nursing facility or an intermediate care facility compared with nondually enrolled beneficiaries (30.6% vs 16.2%, respectively, in 2004 and 31.3% vs 19.2% in 2017).

Additional Analysis

The analysis of the Medicare Current Beneficiary Survey data examined whether the gap in annual all-cause mortality rates between dually enrolled beneficiaries and nondually enrolled beneficiaries was due to differences in the clinical characteristics of these groups. Confirming our primary analysis, observed annual all-cause mortality rates were higher among dually enrolled beneficiaries compared with among nondually enrolled beneficiaries (eFigure 5 in the Supplement).

Dually enrolled beneficiaries had higher age-, sex-, and race-adjusted all-cause mortality rates compared with nondually enrolled beneficiaries (adjusted OR, 2.49 [95% CI, 2.18-2.84]), a pattern that persisted after additionally accounting for clinical characteristics (comorbidities, health status, and functional status) (adjusted OR, 1.59 [95% CI, 1.39-1.82]) (eTable 2 in the Supplement). There was no significant interaction between dual enrollment status and time (P = .50), suggesting no differential change in annual mortality rates by dual enrollment status over time in this sample.

Discussion

In this study of Medicare beneficiaries aged 65 years or older, annual all-cause mortality rates among dually enrolled beneficiaries were substantially higher compared with nondually enrolled beneficiaries. Although all-cause mortality modestly declined for both groups between 2004 and 2017, the difference in mortality between dually enrolled beneficiaries and nondually enrolled beneficiaries did not decrease. All-cause hospitalizations were higher for dually enrolled beneficiaries compared with nondually enrolled beneficiaries, and steadily declined for both groups during the study period, although the gap between them widened. Hospitalization-related mortality was consistently higher among dually enrolled beneficiaries compared with nondually enrolled beneficiaries, a difference that persisted during the study period.

The persistent gap in all-cause mortality between dually enrolled beneficiaries and nondually enrolled beneficiaries from 2004 to 2017 is consistent with broader trends in income inequality and health. In the United States, there are substantial differences in life expectancy between the richest and poorest individuals, and this inequality has continued to worsen since the early 2000s.2,3 The extent to which these differential gains in health reflect changes in individual risk factors, neighborhood factors, health care delivery, other elements, or a combination of these factors has been debated.

In the current study, although the burden of clinical comorbidities generally increased among dually enrolled beneficiaries, this did not explain the persistent difference in mortality compared with nondually enrolled beneficiaries. There was also substantial geographic variation in all-cause mortality rates among dually enrolled beneficiaries, and the counties with the highest mortality rates in 2004 tended to have the highest mortality rates in 2017. In addition, dually enrolled beneficiaries had higher mortality rates after hospitalization compared with nondually enrolled beneficiaries, a difference that persisted during the study period even after accounting for changes in clinical characteristics.

Since the early 2000s, the Medicare program has undergone significant changes to improve health outcomes for older adults. For instance, the Medicare prescription drug insurance benefit (Part D) began in 2006 and substantially increased access to prescription drugs for nondually enrolled beneficiaries (dually enrolled beneficiaries already received prescription coverage through Medicaid). Recent evidence suggests that Part D implementation led to a reduction in hospitalization rates and mortality, which may have differentially benefited the population of nondually enrolled beneficiaries.12,13

As part of the Affordable Care Act in 2010, the Centers for Medicare & Medicaid Services implemented several national pay-for-performance programs that aimed to incentivize the delivery of higher-quality care. Evidence suggests that these programs have not improved outcomes,14-19 but have disproportionately levied financial penalties on safety net practices and hospitals that tend to care for poor and vulnerable populations.20,21 Dually enrolled beneficiaries face unique challenges, such as poverty, residence in more deprived neighborhoods, and housing instability, and future Medicare policy efforts may need to directly address the social determinants of health and provide support for safety net health care systems to improve health equity in this population.15,22,23

Dually enrolled beneficiaries had consistently higher hospitalization rates and associated Medicare expenditures compared with nondually enrolled beneficiaries during the study period. One potential explanation for these patterns is that the financial structure of concomitant Medicare and Medicaid coverage creates conflicting incentives among clinicians, health systems, and payers. This misalignment may lead to inefficiencies in the delivery of care, such as poor care coordination and greater fragmentation within the systems of care for dually enrolled beneficiaries,24 and may also explain why the high costs found among dually enrolled beneficiaries tend to remain during subsequent years.25,26

Policy makers have implemented initiatives that attempt to address these issues, such as the Program of All-Inclusive Care for the Elderly, the Minnesota Senior Health Options program, and dually eligible Medicare Advantage special needs plans.27,28 More recently, the Medicare-Medicaid Financial Alignment Initiative allows states to test models that align financing of both programs and better integrate primary, acute, and behavior health care for dually enrolled beneficiaries.27,29 Whether initiatives like the Financial Alignment Initiative reduce acute care use and improve quality of care and outcomes for this population is an important area of ongoing research.

Limitations

This study has several limitations. First, Medicaid eligibility criteria (eg, income) vary among states and also changed after the expansion of Medicaid in 2014 under the Affordable Care Act. The poverty standard for Medicaid eligibility for Medicare patients aged 65 years or older is set federally, and thus is much more comparable across states. However, a number of states have waivers to this standard, and it is likely that states also vary in the ease of enrollment and the ease in remaining enrolled in Medicaid.

Second, although there is movement in and out of the Medicaid program, recent data suggest that rates of continuous enrollment in Medicaid tend to be high (>85%) among Medicare patients aged 65 years or older.30,31

Third, this study focused on the population aged 65 years or older enrolled in Medicare fee-for-service because hospitalization-related information about the Medicare managed care (Medicare Advantage) population is not available. Although increased Medicare Advantage enrollment could shift the clinical profile of the Medicare fee-for-service population, the current analysis accounted for changes in clinical risk over time for both dually enrolled beneficiaries and nondually enrolled beneficiaries. In addition, the annual rate at which dually enrolled beneficiaries switch from Medicare fee-for-service to Medicare Advantage is low (approximately 4%) and is similar to the rate for nondually enrolled beneficiaries.32 However, if the Centers for Medicare & Medicaid Services makes Medicare Advantage data available for this period, examining outcomes for these beneficiaries could be an important area for future research.

Fourth, because Medicare does not cover long-term care but Medicaid does, Medicare only beneficiaries who need such care are sometimes forced to spend down their assets later in life to gain Medicaid coverage. This process could lead to a steady pool of older and sicker dually enrolled beneficiaries, which would suggest that to some degree, gaps in outcomes between these populations are inevitable.

Fifth, linked Medicaid data on dually enrolled beneficiaries’ expenditures are not available for research use and so we likely underestimated their total costs of care.

Conclusions

Among Medicare fee-for-service beneficiaries aged 65 years or older, dually enrolled beneficiaries had higher annual all-cause mortality, all-cause hospitalizations, and hospitalization-related mortality compared with nondually enrolled beneficiaries. Between 2004 and 2017, these differences did not decrease.

Back to top
Article Information

Corresponding Author: Rishi K. Wadhera, MD, MPP, MPhil, Richard A. and Susan F. Smith Center for Outcomes Research in Cardiology, Beth Israel Deaconess Medical Center, Harvard Medical School, 375 Longwood Ave, Boston, MA 02215 (rwadhera@bidmc.harvard.edu).

Accepted for Publication: January 27, 2020.

Author Contributions: Drs Wadhera and Wang had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Wadhera, Yeh, Joynt Maddox.

Acquisition, analysis, or interpretation of data: Wadhera, Wang, Figueroa, Dominici, Joynt Maddox.

Drafting of the manuscript: Wadhera, Figueroa, Joynt Maddox.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Wang, Dominici.

Administrative, technical, or material support: Figueroa.

Supervision: Yeh, Joynt Maddox.

Other: Wadhera.

Conflict of Interest Disclosures: Dr Wadhera reported receiving support from the National Heart, Lung, and Blood Institute and has served as a consultant to Regeneron. Dr Figueroa reported receiving support from the Commonwealth Fund and the Harvard Center for AIDS Research. Dr Dominici reported receiving support from the National Institute of Environmental Health Sciences, the Health Effects Institute, the National Institutes of Health, and the Environmental Protection Agency; and reported receiving personal fees from Colgate and Johnson & Johnson. Dr Yeh reported receiving support from the National Heart, Lung, and Blood Institute and the Richard A. and Susan F. Smith Center for Outcomes Research in Cardiology; and receiving grants and personal fees from Abbott Vascular, AstraZeneca, Boston Scientific, and Medtronic. Dr Joynt Maddox reported receiving support from the National Heart, Lung, and Blood Institute, the National Institute on Aging, the Commonwealth Fund, and the US Department of Health and Human Services. No other disclosures were reported.

Funding/Support: This work was supported by the Richard A. and Susan F. Smith Center for Outcomes Research in Cardiology.

Role of the Funder/Sponsor: The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: Dr Joynt Maddox is an Associate Editor of JAMA, but she was not involved in any of the decisions regarding review of the manuscript or its acceptance.

References
1.
US Census Bureau. Gini index of money income and equivalence-adjusted income: 1967 to 2014. https://www.census.gov/content/dam/Census/library/visualizations/2015/demo/gini-index-equil-adj-inc.pdf. Accessed November 21, 2019.
2.
Chetty  R, Stepner  M, Abraham  S,  et al.  The association between income and life expectancy in the United States, 2001-2014.  JAMA. 2016;315(16):1750-1766.PubMedGoogle ScholarCrossref
3.
Bor  J, Cohen  GH, Galea  S.  Population health in an era of rising income inequality: USA, 1980-2015.  Lancet. 2017;389(10077):1475-1490.PubMedGoogle ScholarCrossref
5.
Office of the Assistant Secretary for Planning and Evaluation; Department of Health and Human Services. Report to Congress: social risk factors and performance under Medicare’s value-based purchasing programs. https://aspe.hhs.gov/pdf-report/report-congress-social-risk-factors-and-performance-under-medicares-value-based-purchasing-programs. Accessed April 11, 2019.
6.
Congressional Budget Office. Dual-eligible beneficiaries of Medicare and Medicaid. https://www.cbo.gov/publication/44308. Accessed April 11, 2019.
7.
Elixhauser  A, Steiner  C, Harris  DR,  et al.  Comorbidity measures for use with administrative data.  Med Care. 1998;36(1):8-27.PubMedGoogle ScholarCrossref
8.
Tsugawa  Y, Figueroa  JF, Papanicolas  I,  et al.  Assessment of strategies for managing expansion of diagnosis coding using risk-adjustment methods for Medicare data.  JAMA Intern Med. 2019;179(9):1287-1290.PubMedGoogle ScholarCrossref
9.
Downing  NS, Wang  C, Gupta  A,  et al.  Association of racial and socioeconomic disparities with outcomes among patients hospitalized with acute myocardial infarction, heart failure, and pneumonia.  JAMA Netw Open. 2018;1(5):e182044.PubMedGoogle Scholar
10.
Zhang  J, Yu  KF.  What’s the relative risk?  JAMA. 1998;280(19):1690-1691.PubMedGoogle ScholarCrossref
11.
Centers for Medicare & Medicaid Services. Technical documentation for the Medicare Current Beneficiary Survey. https://www.cms.gov/Research-Statistics-Data-and-Systems/Research/MCBS/Downloads/OLD/HHC2003appendixA.pdf. Accessed November 29, 2019.
12.
Huh  J, Reif  J.  Did Medicare Part D reduce mortality?  J Health Econ. 2017;53:17-37.PubMedGoogle ScholarCrossref
13.
Afendulis  CC, He  Y, Zaslavsky  AM,  et al.  The impact of Medicare Part D on hospitalization rates.  Health Serv Res. 2011;46(4):1022-1038.PubMedGoogle ScholarCrossref
14.
Figueroa  JF, Tsugawa  Y, Zheng  J,  et al.  Association between the value-based purchasing pay for performance program and patient mortality in US hospitals.  BMJ. 2016;353:i2214.PubMedGoogle ScholarCrossref
15.
Wadhera  RK, Yeh  RW, Joynt Maddox  KE.  The hospital readmissions reduction program.  N Engl J Med. 2019;380(24):2289-2291.PubMedGoogle ScholarCrossref
16.
Ryan  AM, Krinsky  S, Maurer  KA,  et al.  Changes in hospital quality associated with hospital value-based purchasing.  N Engl J Med. 2017;376(24):2358-2366.PubMedGoogle ScholarCrossref
17.
Sankaran  R, Sukul  D, Nuliyalu  U,  et al.  Changes in hospital safety following penalties in the US Hospital Acquired Condition Reduction Program.  BMJ. 2019;366:l4109.PubMedGoogle ScholarCrossref
18.
Roberts  ET, Zaslavsky  AM, McWilliams  JM.  The value-based payment modifier.  Ann Intern Med. 2018;168(4):255-265.PubMedGoogle ScholarCrossref
19.
Wadhera  RK, Joynt Maddox  KE, Wasfy  JH,  et al.  Association of the hospital readmissions reduction program with mortality among Medicare beneficiaries hospitalized for heart failure, acute myocardial infarction, and pneumonia.  JAMA. 2018;320(24):2542-2552.PubMedGoogle ScholarCrossref
20.
Joynt  KE, Jha  AK.  Characteristics of hospitals receiving penalties under the Hospital Readmissions Reduction Program.  JAMA. 2013;309(4):342-343.PubMedGoogle ScholarCrossref
21.
Chen  LM, Epstein  AM, Orav  EJ,  et al.  Association of practice-level social and medical risk with performance in the Medicare physician value-based payment modifier program.  JAMA. 2017;318(5):453-461.PubMedGoogle ScholarCrossref
22.
Joynt Maddox  KE.  Financial incentives and vulnerable populations.  N Engl J Med. 2018;378(11):977-979.PubMedGoogle ScholarCrossref
23.
Joynt  KE, De Lew  N, Sheingold  SH,  et al.  Should Medicare value-based purchasing take social risk into account?  N Engl J Med. 2017;376(6):510-513.PubMedGoogle ScholarCrossref
24.
Verma  S. Better care for people dually eligible for Medicare and Medicaid. https://www.healthaffairs.org/do/10.1377/hblog20190423.701475/full/. Accessed June 24, 2019.
25.
Figueroa  JF, Zhou  X, Jha  AK.  Characteristics and spending patterns of persistently high-cost Medicare patients.  Health Aff (Millwood). 2019;38(1):107-114.PubMedGoogle ScholarCrossref
26.
Figueroa  JF, Lyon  Z, Zhou  X,  et al.  Persistence and drivers of high-cost status among dual-eligible Medicare and Medicaid beneficiaries.  Ann Intern Med. 2018;169(8):528-534.PubMedGoogle ScholarCrossref
27.
Medpac. Managed care plans for dual-eligible beneficiaries: report to the Congress: Medicare and the health care delivery system. http://medpac.gov/docs/default-source/reports/jun18_ch9_medpacreport_sec.pdf. Accessed June 23, 2019.
28.
Grabowski  DC.  Special needs plans and the coordination of benefits and services for dual eligibles.  Health Aff (Millwood). 2009;28(1):136-146.PubMedGoogle ScholarCrossref
30.
Ku  L, Steinmetz  E. Bridging the gap: continuity and quality of coverage in Medicaid association for community affilitated plans. https://ccf.georgetown.edu/wp-content/uploads/2013/09/GW-Continuity-Report-9-10-13.pdf. Accessed June 24, 2019.
31.
Riley  GF, Zhao  L, Tilahun  N.  Understanding factors associated with loss of medicaid coverage among dual eligibles can help identify vulnerable enrollees.  Health Aff (Millwood). 2014;33(1):147-152.PubMedGoogle ScholarCrossref
32.
Jacobson  GA, Neuman  P, Damico  A.  At least half of new Medicare advantage enrollees had switched from traditional Medicare during 2006-11.  Health Aff (Millwood). 2015;34(1):48-55.PubMedGoogle ScholarCrossref
×