- Email: [email protected]
Association of Race and Poverty With Mortality Among Nursing Home Residents on Maintenance Dialysis
JAMDA 20 (2019) 904e910
JAMDA journal homepage: www.jamda.com
Original Study
Association of Race and Poverty With Mortality Among Nursing Home Residents on Maintenance Dialysis Robert Nee MD a, b, *, John S. Thurlow MD a, b, Keith C. Norris MD, PhD c, Christina Yuan MD a, b, Maura A. Watson DO, MPH a, b, Lawrence Y. Agodoa MD d, Kevin C. Abbott MD, MPH d a
Nephrology Service, Walter Reed National Military Medical Center, Bethesda, MD Department of Medicine, Uniformed Services University of the Health Sciences, Bethesda, MD Department of Medicine, David Geffen School of Medicine at University of CaliforniaeLos Angeles, Los Angeles, CA d National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, MD b c
a b s t r a c t Keywords: Nursing home racial disparities poverty mortality end-stage renal disease dialysis USRDS
Objectives: The association of race, ethnicity, and socioeconomic factors with survival rates of nursing home (NH) residents with treated end-stage renal disease (ESRD) is unclear. We examined whether race/ ethnicity, ZIP codeelevel, and individual-level indicators of poverty relate to mortality of NH residents on dialysis. Design: Retrospective cohort study. Participants/Setting: Using the United States Renal Data System database, we identified 56,194 nursing home residents initiated on maintenance dialysis from January 1, 2007 through December 31, 2013, followed until May 31, 2014. Measurements: We evaluated baseline characteristics of the NH cohort on dialysis, including race and ethnicity. We assessed the Medicare-Medicaid dual eligibility status as an indicator of individual-level poverty and ZIP codeelevel median household income (MHI) data. We conducted Cox regression analyses with all-cause mortality as the outcome variable, adjusted for clinical and sociodemographic factors including end-of-life preferences. Results: Adjusted Cox analysis showed a significantly lower risk of death among black vs nonblack NH residents [adjusted hazard ratio (AHR) 0.91, 95% confidence interval (CI) 0.89, 0.94]. Dual-eligibility status was significantly associated with lower risk of death compared to those with Medicare alone (AHR 0.80, 95% CI 0.78, 0.82). Compared to those in higher MHI quintile levels, NH ESRD patients in the lowest quintile were significantly associated with higher risk of death (AHR 1.09, 95% CI 1.06, 1.13). Conclusions/Implications: Black and Hispanic NH residents on dialysis had an apparent survival advantage. This “survival paradox” occurs despite well-documented racial/ethnic disparities in ESRD and NH care and warrants further exploration that could generate new insights into means of improving survival of all NH residents on dialysis. Area-level indicator of poverty was independently associated with mortality, whereas dual-eligibility status for Medicare and Medicaid was associated with lower risk of death, which could be partly explained by improved access to care. Published by Elsevier Inc. on behalf of AMDA e The Society for Post-Acute and Long-Term Care Medicine.
The views expressed in this article are those of the authors and do not reflect the official policy of the Department of Army/Navy/Air Force, the Department of Defense, National Institutes of Health or the United States government. Part of this article was presented at the American Society of Nephrology Kidney Week 2016 Annual Meeting, Chicago, IL November 17-20, 2016. This research did not receive any funding from agencies in the public, commercial, or not-for-profit sectors. Keith Norris is supported by NIH grants UL1TR000124 and P30AG021684.
The authors declare no conflicts of interest. * Address correspondence to Robert Nee, MD, Nephrology Service, Walter Reed National Military Medical Center, 8901 Wisconsin Ave, Bethesda, MD 20889. E-mail address: [email protected] (R. Nee).
https://doi.org/10.1016/j.jamda.2019.02.013 1525-8610/Published by Elsevier Inc. on behalf of AMDA e The Society for Post-Acute and Long-Term Care Medicine.
R. Nee et al. / JAMDA 20 (2019) 904e910
Racial, ethnic, and class disparities in health care exist among nursing home (NH) residents, a population that contains many people with chronic conditions that cause functional and/or cognitive limitations.1,2 Smith et al3 found that NH care remained segregated, with two-thirds of all black residents living in just 10% of all nursing homes, and these residents were significantly more likely to be served by those in the bottom quartile of many measures of quality. Mor et al4 also reported that black NH residents were overrepresented in lowrevenue, understaffed, and poor-quality nursing homes. Further, the socioeconomic status composition of NH residents is highly correlated with quality.2 The lower tier of NH care consists of those housing a high proportion of Medicaid and black residents with very limited resources and has more health-related regulatory deficiencies.4 Within the end-stage renal disease (ESRD) population, it has long been recognized that disparities in health care access and delivery, disease burden, and clinical outcomes exist among racial/ethnic and socially disadvantaged groups.5,6 Paradoxically, however, black patients on dialysis have better survival rates than their white counterparts.7 In a seminal study, Kucirka et al8 demonstrated that the commonly cited survival advantage for black dialysis patients applies only to those older than 50 years. Nonetheless, data on racial/ethnic differences in survival in the NH dialysis population remains underreported, as outcome data in the literature for this population is limited.9,10 In a cohort of long-term stay NH residents (>65 years) initiated on dialysis from 2004 to 2006 (n ¼ 3748), blacks had a lower risk of death compared to whites over 12 months [adjusted hazard ratio (AHR) 0.73, 95% confidence interval (CI) 0.68, 0.79].11 Although it is known that older black patients on chronic dialysis survive longer than their white counterparts, there remains knowledge gaps in the literature of this “survival paradox” specific to the nursing home population. The important questions of whether racial/ ethnic gaps in end-of-life preferences and socioeconomic status have a role in survival differences have not been reported and remain unanswered. Therefore, we aim to examine the relationship of race, ethnicity, and socioeconomic factors to survival rates in a large, contemporary cohort of NH residents on dialysis, accounting for patient-level factors and end-of-life care. We hypothesized that minority NH residents on dialysis would have a survival advantage, despite inequities in health care, and that indicators of poverty would be associated with mortality in this population. Methods This study used the United States Renal Data System (USRDS), which incorporates baseline and follow-up demographic and clinical data on all patients accessing the Medicare ESRD program in the United States. We conducted a retrospective cohort study of nursing home residents initiated on either hemodialysis or peritoneal dialysis from January 1, 2007 through December 31, 2013, followed until May 31, 2014. We identified a cohort of 56,194 patients who were nursing home residents at the time of dialysis initiation from the overall dialysis population of 782,161 (7.2%). This NH cohort was identified through the Centers for Medicare & Medicaid Services (CMS) Medical Evidence Form 2728 which is used to register each new patient at the onset of ESRD.12 We then merged the USRDS database with the 2010 US Census for ZIP codeelevel median household income (MHI) data. There were 1233 (2.2%) out of 56,194 NH patients who had missing MHI data. The primary outcome was all-cause mortality. This study was approved as exempt from review by the Walter Reed National Military Medical Center Institutional Review Board. Patients and Sources The demographics of the dialysis population in this study have been described in the USRDS Annual Data Reports for the years
905
studied.12 Variables included in the USRDS standard analysis files (SAFs), as well as data collection methods and validation studies, are listed on the USRDS website (www.usrds.org). The files SAF.PATIENTS were used as the primary data set and SAF.MEDEVID for additional information coded in the CMS Medical Evidence Form 2728. We obtained data on hospice care status from the SAF.DEATH file, which is derived from the death notification form (CMS-2746). Data on feeding tube use (via percutaneous endoscopic gastrostomy), as a proxy indicator of preference for life-prolonging measures, was obtained from inpatient procedure Medicare claims. We examined data on hospice care and feeding tube use because these factors could be associated with racial differences in observed survival rates. Files from the 2010 US Census (http://factfinder.census.gov/faces/nav/jsf/pages/index. xhtml) were used to merge by ZIP code with USRDS files. Because such data are area-based and thus ecological, we also assessed Medicare-Medicaid dual eligibility status from the USRDS files as an indicator of individual-level poverty13e15 and its association with mortality. Although eligibility varies by state, means testing is stricter than for either Medicare or Medicaid alone, and usually includes the poorest patients receiving care, at most <135% poverty and generally lower than 100% of poverty.14,16 Independent Variables We incorporated covariables in our models that represented potential confounders in the analysis of all-cause mortality in the ESRD population. These variables included age at initiation of dialysis, year at first ESRD service, gender, race, Hispanic ethnicity (a non-mutually exclusive category that could overlap with race), estimated glomerular filtration rate at dialysis initiation [by CKDEPI (Chronic Kidney Disease Epidemiology Collaboration) equation17], dialysis modality, vascular access used at start of dialysis, primary cause of ESRD, diabetes mellitus, hypertension, other comorbid conditions from the CMS Medical Evidence Form 2728 (Table 1), tobacco use, alcohol dependence, body mass index, serum albumin, hemoglobin, amputation, ability to ambulate, ability to transfer, requirement for assistance with daily activities, and socioeconomic factors (quintiles of ZIP codeelevel MHI, individual employment status, and insurance status including dual eligibility for Medicare and Medicaid as a surrogate for individual-level poverty). We also assessed patient-level variables related to endof-life care available in the USRDS, including withdrawal from dialysis prior to death as well as hospice care and feeding tube use as noted above. Statistical Analysis Analyses were performed using Stata 14 SE (Stata Corp, College Station, TX). Univariate analyses were performed with chi-square testing for categorical variables and Student t-test for continuous variables. P values <.05 were considered statistically significant for univariate comparisons. Survival curve was estimated using the Kaplan-Meier method and log-rank test to assess significance between survival distributions. We also conducted Cox regression analyses to model factors associated with all-cause mortality. As a sensitivity analysis, we further adjusted for covariables in the Cox model related to end-of-life care as noted above (withdrawal from dialysis, hospice, and feeding tube use). Administrative censoring occurs at the end of study (May 31, 2014). Our model development used the forced entry method in which all independent variables were entered into the model at the same time, without any specific sequence in which variables were entered. Proportional hazards assumptions were examined by graphing log [elog (survival function)] vs log (time) for the comparison groups.
906
R. Nee et al. / JAMDA 20 (2019) 904e910
Table 1 Baseline Demographic and Comorbidity Characteristics of Nursing Home Residents on Chronic Dialysis, 2007-2013, Blacks vs Nonblacks Variables Mean age (y) at start of dialysis (SD) Gender Male Female Hispanic ethnicity Dialysis modality Hemodialysis Peritoneal dialysis Vascular access used at start of dialysis Arteriovenous fistula Graft Catheter Cause of ESRD Diabetes mellitus Hypertension Glomerulonephritis Cystic kidney disease Other renal disorders Comorbid conditions Diabetes mellitus Hypertension Atherosclerotic heart disease Congestive heart failure Peripheral vascular disease Cerebrovascular disease (CVA, TIA) COPD Cancer Tobacco use Alcohol dependence Amputation Inability to ambulate Needs assistance with daily activities Inability to transfer Estimated GFR (CKD-EPI), mL/min/1.73 m2 Mean body mass index (SD) Serum albumin, g/dL (SD) Hemoglobin, g/dL (SD) Insurance status Employer group health Medicaid alone Medicare alone Dual eligibility for Medicare and Medicaid* Employed (full- or part-time) MHI quintile levelsy Bottom fifth quintile (range $6993-$46,209/y) Lower middle quintile (range $46,212-$54,989/y) Middle quintile (range $54,992-$64,534/y) Upper middle quintile (range $64,541-$80,793/y) Top fifth quintile (range $80,800-$329,112/y) Mean ZIP codeelevel MHI ($/y) (SD)y Variables related to end-of-life care Withdrawal from dialysis prior to death Hospice care status Feeding tube use
Blacks (n ¼ 14,983) 68.34 (12.7)
Nonblacks (n ¼ 41,211) 72.12 (11.8)
7003 (46.7) 7979 (53.3) 243 (1.6)
20,793 (50.5) 20,417 (49.5) 4732 (11.5)
14,511 (99.7) 41 (0.3)
40,214 (99.4) 237 (0.6)
772 (5.2) 545 (3.7) 13,537 (91.1)
2341 (5.8) 790 (1.9) 37,534 (92.3)
P Value <.001 <.001
<.001 <.001
<.001
<.001 7225 5239 350 46 1487 9743 13,316 3198 6562 2750 3915 2161 1228 830 487 1187 7299 9271 5085 10.84 29.06 2.84 9.82 405 1506 4158 7990 79 4649 3213 2564 2270 1818 58,377
(50.4) (36.5) (2.4) (0.3) (10.4)
19,242 11,489 1356 139 6423
(49.8) (29.7) (3.5) (0.4) (16.6)
(65.0) (88.9) (21.3) (43.8) (18.4) (26.1) (14.4) (8.2) (5.5) (3.2) (7.9) (48.7) (61.9) (33.9) (7.2) (9.1) (1.4) (11.6)
25,814 34,354 12,150 20,062 9098 7401 8245 4163 1945 995 2760 18,732 24,813 12,051 12.38 29.52 2.88 10.19
(62.6) (83.4) (29.5) (48.7) (22.1) (18.0) (20.0) (10.1) (4.7) (2.4) (6.7) (45.4) (60.2) (29.2) (7.6) (8.8) (2.4) (17.5)
(2.9) (10.7) (29.6) (53.3) (0.5)
1170 2220 19,013 15,589 315
(3.1) (5.8) (50.0) (37.8) (0.8)
(32.0) (22.1) (17.7) (15.6) (12.5) (23,060)
4989 8537 9440 8833 8648 67,985
(12.3) (21.1) (23.3) (21.8) (21.4) (25,727)
1980 (13.2) 1736 (11.6) 1885 (12.6)
9641 (23.4) 7557 (18.3) 2860 (6.9)
<.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001 .09 .024 <.001
<.001 .003 <.001
<.001 <.001 <.001 <.001
CKD-EPI, Chronic Kidney Disease Epidemiology Collaboration; COPD, chronic obstructive pulmonary disease; CVA, cerebrovascular accident; GFR, glomerular filtration rate; SD, standard deviation; TIA, transient ischemic attack. Data are n (%) or mean (standard deviation). Univariate analyses were performed with chi-square testing for categorical variables and Student t test for continuous variables. *Mean age of dual-eligible vs nonedual eligible patients (69.2 vs 72.5 years, respectively, P < .001). y Based on ZIP code from the 2010 US Census.
Results From a total of 782,161 incident dialysis patients in the United States from 2007 to 2013, we identified 56,194 nursing home residents. Table 1 shows the baseline demographics and characteristics of the NH cohort, of whom 14,983 (27%) were black and 41,211 (73%) were nonblack. Compared to nonblack NH residents, blacks were younger at dialysis initiation, and more likely to be female and have hypertension as the primary cause of ESRD. Further, slightly higher
proportions of blacks had indications of impaired physical function, including inability to ambulate, requirement for assistance with daily activities, and inability to transfer. Compared to nonblack NH residents, blacks were more likely to have dual-eligibility status for both Medicare and Medicaid (53.3% vs 37.8%) and in the lowest MHI quintile level (32.0% vs 12.3%). The mean ZIP codeelevel MHI was lower among black vs nonblack NH residents ($58,377 vs $67,985, P < .001). Nonblacks were more likely to be of Hispanic ethnicity and have atherosclerotic heart disease, congestive heart failure, peripheral
R. Nee et al. / JAMDA 20 (2019) 904e910
907
residents was attenuated but remained significant (AHR 0.95, 95% CI 0.92, 0.98, P < .001) after adjusting for additional variables related to end-of life care (withdrawal from dialysis, hospice, and feeding tube use). Dual-eligibility status and employer group health insurance were significantly associated with a lower risk of death compared to those with Medicare alone (AHR 0.80 and 0.87, respectively). On the other hand, higher ZIP codeelevel MHI was associated with a lower risk of death, in a near graded fashion. In a separate Cox model, we found that NH residents on dialysis in the lowest quintile had an AHR 1.09 (95% CI 1.06, 1.13) compared to those in higher area-level MHI quintile levels. There were no significant interactions between race (P ¼ .80), Hispanic ethnicity (P ¼ .76), and area-level MHI. Discussion
Fig. 1. Patient survival among nursing home residents on chronic dialysis, by race.
vascular disease, chronic obstructive pulmonary disease, and cancer. In regard to end-of-life preferences, black NH residents were less likely to withdraw from dialysis and enroll in hospice but more likely to undergo feeding tube placement than their nonblack counterparts. The overall crude mortality rate in NH residents on dialysis was 480 per 1000 patient-years [PYs] (95% CI 475-484 per 1000 PYs), with a mean follow-up duration of 2.6 years. Black NH residents had a lower crude mortality rate than nonblacks (406 vs 511 per 1000 PYs, respectively, P < .001) (Figure 1). Similarly, Hispanic NH residents had a lower crude mortality rate than non-Hispanics (407 vs 488 per 1000 PYs, respectively, P < .001) (Figure 2). On sensitivity analysis, nonHispanic blacks had a lower crude mortality rate than non-Hispanic whites (405 vs 533 per 1000 PYs, respectively, P < .001). As shown in Table 2, in the adjusted Cox regression model, black NH residents had a significantly lower mortality rate compared to nonblacks (AHR 0.91). Similarly, Hispanic NH residents had a significantly lower mortality rate compared to non-Hispanics (AHR 0.89). Thus, black race and Hispanic ethnicity among NH dialysis patients were independently associated with greater longevity, accounting for multiple covariates including demographics, dialysis vascular access, comorbid conditions, body mass index, serum albumin, hemoglobin, indicators of impaired physical function, insurance status, and individual-level and ZIP codeelevel measures of poverty. On sensitivity analysis, non-Hispanic blacks also had a significantly lower mortality rate than non-Hispanic whites (AHR 0.90, 95% CI 0.87, 0.92, P < .001). Further, the survival advantage among black vs nonblack NH
Fig. 2. Patient survival among nursing home residents on chronic dialysis, by ethnicity.
We found in a national, multiyear study of nursing home residents on maintenance dialysis that blacks and Hispanics had improved survival compared with their nonblack and non-Hispanic counterparts. A similar survival advantage occurred among non-Hispanic blacks vs non-Hispanic whites on dialysis. These findings persisted even after accounting for clinical, sociodemographic, and individual preferences for life-prolonging measures. Our findings are consistent with the “survival paradox” phenomenon in the general ESRD population in which racial- and ethnic-minority groups survive longer than whites, despite their poor intermediate health outcomes and worse clinical performance measures.5,7,12,18,19 Our findings validate and extend the findings of Kucirka et al, who demonstrated that the survival advantage for black patients undergoing dialysis applies only to older adults.8 In a subgroup of patients between 61 and 70 years of age in their study, the investigators demonstrated that blacks had significantly lower mortality than whites (AHR 0.87, 95% CI 0.86, 0.88) and those <50 years had a higher risk of death. The survival paradox is even more surprising given welldocumented racial and ethnic disparities in outcomes and delivery of ESRD care.6,20e22 Furthermore, minority groups do not receive nursing home care of comparable quality to whites.1,2,23 Mor et al4 found that black NH residents were nearly 4 times more likely than whites to reside in nursing homes with limited resources and poor performance. Smith et al24 reported that blacks were more likely to be in nursing homes that were understaffed, cited for deficiencies and terminated from the Medicare and Medicaid program. The underlying mechanisms for reduced mortality in older black dialysis patients remain elusive. Potential explanations include survival bias (black patients who survive long enough to reach ESRD may be healthier than white patients), cultural differences in adaptation to chronic illness, favorable nutritional and anthropometric characteristics, differential response to inflammation, lower prevalence of cardiovascular disease, and a differing cardiovascular risk profile including higher blood pressure, which paradoxically may have a beneficial association with survival in ESRD patients, referred to as “reverse epidemiology” of cardiovascular disease.19,25e27 Furthermore, the paradoxical survival advantage may in part be due to comparatively lower rates of kidney transplantation among blacks vs whites.28 To the extent that black patients are less likely to receive a transplant, those who initiate or remain on dialysis may be relatively healthier than their white counterparts, thus inflating the black survival advantage in older adults.29 In the nursing home population, the apparent survival advantage among black residents may be partly due to the lower likelihood of enrolling in hospice care compared to whites.30e33 Eneanya et al34 demonstrated that blacks with advanced chronic kidney disease were less likely to communicate end-of-life preferences with family members and more likely to prefer life-extending treatments compared to whites. Prior studies have also shown that blacks may be less likely to withdraw
908
R. Nee et al. / JAMDA 20 (2019) 904e910
Table 2 Multivariable Cox Regression Analyses of Baseline Patient Characteristics for Death Among Nursing Home Residents on Chronic Dialysis Variables
Adjusted HR
95% CI
P Value
Age at start of dialysis Year at first ESRD service Gender (male vs female) Race (black vs nonblack) Ethnicity (Hispanic vs non-Hispanic) Dialysis modality (hemodialysis vs peritoneal dialysis) Vascular access used at start of dialysis Catheter Graft Arteriovenous fistula Cause of ESRD Diabetes mellitus Hypertension Glomerulonephritis Cystic kidney disease Other renal disorders Comorbid conditions Diabetes mellitus Hypertension Atherosclerotic heart disease Congestive heart failure Peripheral vascular disease Cerebrovascular disease (CVA, TIA) COPD Cancer Tobacco use Alcohol dependence Amputation Inability to ambulate Needs assistance with daily activities Inability to transfer Estimated GFR (<10 mL/min/1.73 m2 vs >10 mL/min/1.73 m2) Mean body mass index (>30 vs <30) Serum albumin (<3.0 vs >3.0 g/dL) Hemoglobin (<9 vs >9 g/dL) Insurance status Medicare alone Medicaid alone Employer group health Dual eligibility for Medicare and Medicaid Employed MHI quintile levels Bottom fifth quintile (range $6993-$46,209/y) Lower middle quintile (range $46,212-$54,989/y) Middle quintile (range $54,992-$64,534/y) Upper middle quintile (range $64,541-$80,793/y) Top fifth quintile (range $80,800-$329,112/y)
1.02 0.96 1.02 0.91 0.89 1.19
1.02, 0.95, 1.00, 0.89, 0.85, 0.57,
1.03 0.97 1.05 0.94 0.93 2.49
<.001 <.001 .045 <.001 <.001 .65
1.0 (referent) 0.79 0.78
0.73, 0.85 0.75, 0.82
<.001 <.001
1.0 (referent) 1.00 0.87 0.90 0.99
0.97, 0.81, 0.74, 0.95,
1.03 0.93 1.09 1.02
.99 <.001 .27 .50
0.97 0.88 0.96 1.16 1.04 1.04 1.16 1.15 1.06 1.07 1.04 1.07 1.02 1.24 0.87 0.90 1.12 0.96
0.94, 0.85, 0.94, 1.13, 1.01, 1.01, 1.13, 1.11, 1.00, 1.00, 0.99, 1.04, 0.99, 1.21, 0.85, 0.88, 1.10, 0.93,
1.00 0.90 0.99 1.18 1.07 1.07 1.19 1.20 1.12 1.15 1.08 1.10 1.04 1.28 0.89 0.93 1.15 0.98
.035 <.001 .003 <.001 .007 .005 <.001 <.001 .035 .06 .13 <.001 .13 <.001 <.001 <.001 <.001 <.001
1.0 (referent) 1.04 0.87 0.80 0.79
0.99, 0.82, 0.78, 0.67,
1.09 0.94 0.82 0.92
.12 <.001 <.001 .002
1.0 (referent) 0.96 0.90 0.90 0.88
0.93, 0.87, 0.87, 0.84,
0.99 0.93 0.93 0.91
.049 <.001 <.001 <.001
COPD, chronic obstructive pulmonary disease; CVA, cerebrovascular accident; GFR, glomerular filtration rate; TIA, transient ischemic attack.
from dialysis.35e37 In a recent analysis of a large cohort of dialysis patients dying between 2000 and 2014 (not specific to nursing home residents), non-Hispanic blacks were less likely to discontinue dialysis and less likely to receive hospice care than non-Hispanic whites [adjusted odds ratios 0.51 (95% CI 0.50, 0.52) and 0.58 (95% CI 0.57, 0.59), respectively].38 The findings from these studies are consistent with our analyses that suggest greater preferences for life-prolonging measures among black NH residents on maintenance dialysis, based on USRDS data on withdrawal from dialysis, hospice, and feeding tube use. Although these end-of-life care factors partly accounted for the survival advantage in black vs nonblack NH residents (AHR for mortality increased from 0.91 to 0.95), the racial difference in mortality remained significant. Given the interdependency of race, ethnicity, and socioeconomic factors on adverse clinical outcomes in patients with ESRD,39,40 we assessed whether poverty attenuates the survival advantage among black NH residents on dialysis. We demonstrated that NH residents on dialysis in the lowest area-level MHI quintile had a higher risk of death compared to those in higher quintile levels, independent of race/ethnicity. However, there was no significant interaction between MHI and race/ethnicity in predicting
mortality. In other words, area-level MHI did not significantly modify the survival advantage among black NH residents on dialysis. These findings corroborate with a USRDS study by Kimmel et al41 who showed that higher ZIP codeelevel MHI was associated with longer survival in both black and white patients in the general hemodialysis population and that controlling for income did not modify black patients’ survival advantages. Our findings of an inverse relationship between income and mortality risk is not surprising given that poor and minority NH residents served by lower-tier nursing homes with limited resources are more likely to receive substandard care.4 Dual-eligibility status for both Medicare and Medicaid, as a proxy measure of individual-level poverty, was significantly associated with lower mortality. This association is incongruent with the ecologic measure of income using ZIP codes. Dual-eligible beneficiaries represent a disadvantaged subgroup of older Americans who are generally impoverished and have higher prevalence of physical and cognitive impairments, less education, and lower levels of social support than their Medicare-only counterparts.42,43 Kimmel et al44 reported that dual-eligibility status, as a proxy measure of poverty, was a major driver for racial mortality differences in the
R. Nee et al. / JAMDA 20 (2019) 904e910
aged US Medicare population. Thus, it remains unclear why dualeligible NH residents on dialysis had better survival. It is possible that poorer NH residents may gain Medicaid eligibility earlier than those with more resources and assets; thus, there may be a subset of dual-eligible residents who have a shorter NH “vintage” which favors their chance of survival. Kurella et al45 reported that initiation of dialysis among NH residents with ESRD was associated with a marked decline in function status and progressively higher mortality over time. Also, it is worth noting that dual-eligible NH residents in our cohort were younger at the time of dialysis initiation than their nonedual eligible counterparts (69.2 vs 72.5 years, respectively). We adjusted for age in the multivariable Cox model, but there could be residual confounders related to the younger age that may partly explain the improved survival in dual-eligible beneficiaries. Perhaps the NH ESRD population with dual eligibility represents a unique impoverished group of patients who benefit from joint Medicare and Medicaid coverage. The dualeligible program was designed to supplement patients who have financial and other disadvantages and thus may improve access to care in the ESRD population, with minimal to no cost-sharing. Although a survey indicated an unmet need for assistance with activities of daily living among dual-eligible beneficiaries, they encountered few problems obtaining medical care.46 Only 3% reported not having a regular physician or other regular source of care, and only 1% were unable to arrange specialist care. Also, the Medicare Part D program provides special protections for dualeligible NH residents that may ameliorate the health impact from coverage limitations of prescription drug plans,47 however less wealthy patients who do not qualify for Medicaid but have Medicare Part D may be unable to afford the out-of-pocket expenses when they fall into a coverage gap, often known as the donut hole.48 Our study has certain limitations. First, we cannot make conclusions about causality given the retrospective nature of our study. Second, there could be ascertainment bias related to providers’ responses on CMS Medical Evidence Form 2728, as demonstrated by Bowling et al49 who reported underestimation of NH utilization among older patients (75 years) with incident ESRD; however, those identified by the CMS Form 2728 as being in an NH have a high probability of actually being in an NH at the time of dialysis initiation (positive predictive value 80%). Third, duration of time spent in the NH pre- and post-dialysis initiation could not be ascertained from the CMS Form 2728 and may bias results as noted above. Fourth, in the absence of individual-level income data, we used ZIP codeebased MHI as a surrogate for patient income. We acknowledge potential biases associated with ZIP code as a proxy measure of individual-level socioeconomic status since an association observed with income on an aggregate level may not represent the association that exists at an individual income level. Also, residual confounders may exist between residents of higher vs lower MHI ZIP codes, such as disparities in health care access or quality. Fifth, dual eligibility is by definition binary and provides no additional information for income above poverty level. Lastly, there were significant missing data in the USRDS on advance directives and were therefore not included in our analyses. The strengths of the current study include the large sample size as reflected by broad representativeness of the USRDS registry, with nearly complete inclusion of all ESRD patients in the United States.50 Comparisons of mortality rates have high validity given the accurate and complete ascertainment of death data in the USRDS.51 Also, as noted above, the high positive predictive value of the NH indicator on CMS Form 2728 ensures that the majority of patients have the condition of interest in our study cohort.48 Importantly, we have also incorporated in our analyses patient-level factors that are particularly relevant in the nursing home setting and end-of-life preferences.
909
Conclusions and Implications Black and Hispanic NH residents on dialysis demonstrate an apparent survival advantage, despite well-documented inequities in health care access and quality. This “survival paradox” warrants further exploration that could generate new insights into means of improving survival of all NH residents on dialysis. Area-level indicator of poverty was independently associated with mortality but did not modify racial differences in survival. Dual-eligibility status for both Medicare and Medicaid was associated with lower risk of death, which could be partly explained by improved access to care. These findings highlight the important role of social determinants of health in longevity among NH residents on dialysis.
References 1. Grabowski DC, McGuire TG. Black-white disparities in care in nursing homes. Atl Econ J 2009;37:299e314. 2. Konetzka RT, Werner RM. Disparities in long-term care: Building equity into market-based reforms. Med Care Res Rev 2009;66:491e521. 3. Smith DB, Feng Z, Fennell ML, et al. Separate and unequal: Racial segregation and disparities in quality across U.S. nursing homes. Health Aff (Millwood) 2007;26:1448e1458. 4. Mor V, Zinn J, Angelelli J, et al. Driven to tiers: Socioeconomic and racial disparities in the quality of nursing home care. Milbank Q 2004;82:227e256. 5. Norris KC, Agodoa LY. Unraveling the racial disparities associated with kidney disease. Kidney Int 2005;68:914e924. 6. Reddan DN, Szczech LA, Klassen PS, Owen WF Jr. Racial inequity in America’s ESRD program. Semin Dial 2000;13:399e403. 7. Agodoa L, Eggers P. Racial and ethnic disparities in end-stage kidney failuresurvival paradoxes in African-Americans. Semin Dial 2007;20:577e585. 8. Kucirka LM, Grams ME, Lessler J, et al. Association of race and age with survival among patients undergoing dialysis. JAMA 2011;306:620e626. 9. Hall RK, O’Hare AM, Anderson RA, Colón-Emeric CS. End-stage renal disease in nursing homes: A systematic review. J Am Med Dir Assoc 2013;14:242e247. 10. Anderson JE, Kraus J, Sturgeon D. Incidence, prevalence, and outcomes of endstage renal disease patients placed in nursing homes. Am J Kidney Dis 1993;21: 619e627. 11. United States Renal Data System. 2010 USRDS Annual Data Report: Epidemiology of Kidney Disease in the United States, vol. 1. Bethesda, MD: National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases; 2010. p. 100. 12. United States Renal Data System. 2017 USRDS Annual Data Report: Epidemiology of Kidney Disease in the United States. Bethesda, MD: National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases; 2017. 13. McMillan A, Pine PL, Gornick M, Prihoda R. A study of the “crossover population”: Aged persons entitled to both Medicare and Medicaid. Health Care Financ Rev 1983;4:19e46. 14. Hagland M. Medicaid/Medicare dual eligibles. Healthplan 1997;38:52e56. 58, 60e61. 15. Nemore P. Dual eligibles. Issue Brief Cent Medicare Educ 2004;5:1e6. 16. United States Census Bureau: Poverty. Available at: https://www.census.gov/ topics/income-poverty/poverty/guidance/poverty-measures.html. Accessed February 15, 2018. 17. Levey AS, Stevens LA, Schmid CH, et al. CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration). A new equation to estimate glomerular filtration rate. Ann Intern Med 2009;150:604e612. 18. Kalantar-Zadeh K, Kovesdy CP, Derose SF, et al. Racial and survival paradoxes in chronic kidney disease. Nat Clin Pract Nephrol 2007;3:493e506. 19. Buckalew VM Jr, Freedman BI. Reappraisal of the impact of race on survival in patients on dialysis. Am J Kidney Dis 2010;55:1102e1110. 20. Powe NR, Melamed ML. Racial disparities in the optimal delivery of chronic kidney disease care. Med Clin North Am 2005;89:475e488. 21. Prakash S, Rodriguez RA, Austin PC, et al. Racial composition of residential areas associates with access to pre-ESRD nephrology care. J Am Soc Nephrol 2010;21: 1192e1199. 22. Ifudu O, Dawood M, Iofel Y, et al. Delayed referral of black, Hispanic, and older patients with chronic renal failure. Am J Kidney Dis 1999;33:728e733. 23. Fennell ML, Miller SC, Mor V. Facility effects on racial differences in nursing home quality of care. Am J Med Qual 2000;15:174e181. 24. Smith DB, Feng Z, Fennell ML, et al. Racial disparities in access to long-term care: The illusive pursuit of equity. J Health Polit Policy Law 2008;33:861e881. 25. Yan G, Norris KC, Yu AJ, et al. The relationship of age, race, and ethnicity with survival in dialysis patients. Clin J Am Soc Nephrol 2013;8:953e961. 26. Nee R, Agodoa LY. Racial differences in kidney disease and end-stage kidney disease in the USA. In: Garcia-Garcia G, Agodoa LY, Norris KC, editors. Chronic Kidney Disease in Disadvantaged Populations. San Diego, CA: Academic Press; 2017. p. 65e72.
910
R. Nee et al. / JAMDA 20 (2019) 904e910
27. Kalantar-Zadeh K, Block G, Humphreys MH, Kopple JD. Reverse epidemiology of cardiovascular risk factors in maintenance dialysis patients. Kidney Int 2003; 63:793e808. 28. Rhee CM, Lertdumrongluk P, Streja E, et al. Impact of age, race and ethnicity on dialysis patient survival and kidney transplantation disparities. Am J Nephrol 2014;39:183e194. 29. Kalantar-Zadeh K, Kovesdy CP, Norris KC. Racial survival paradox of dialysis patients: Robust and resilient. Am J Kidney Dis 2012;60:182e185. 30. Zheng NT, Mukamel DB, Caprio T, et al. Racial disparities in in-hospital death and hospice use among nursing home residents at the end of life. Med Care 2011;49:992e998. 31. Kwak J, Haley WE, Chiriboga DA. Racial differences in hospice use and inhospital death among Medicare and Medicaid dual-eligible nursing home residents. Gerontologist 2008;48:32e41. 32. Murray AM, Arko C, Chen SC, et al. Use of hospice in the United States dialysis population. Clin J Am Soc Nephrol 2006;1:1248e1255. 33. Thomas BA, Rodriguez RA, Boyko EJ, et al. Geographic variation in black-white differences in end-of-life care for patients with ESRD. Clin J Am Soc Nephrol 2013;8:1171e1178. 34. Eneanya ND, Wenger JB, Waite K, et al. Racial disparities in end-of-life communication and preferences among chronic kidney disease patients. Am J Nephrol 2016;44:46e53. 35. Kurella Tamura M, Goldstein MK, Pérez-Stable EJ. Preferences for dialysis withdrawal and engagement in advance care planning within a diverse sample of dialysis patients. Nephrol Dial Transplant 2010;25:237e242. 36. Nelson CB, Port FK, Wolfe RA, Guire KE. The association of diabetic status, age, and race to withdrawal from dialysis. J Am Soc Nephrol 1994;4:1608e1614. 37. Port FK, Wolfe RA, Hawthorne VM, Ferguson CW. Discontinuation of dialysis therapy as a cause of death. Am J Nephrol 1989;9:145e149. 38. Foley RN, Sexton DJ, Drawz P, et al. Race, ethnicity, and end-of-life care in dialysis patients in the United States. J Am Soc Nephrol 2018;29:2387e2399.
39. Nicholas SB, Kalantar-Zadeh K, Norris KC. Socioeconomic disparities in chronic kidney disease. Adv Chronic Kidney Dis 2015;22:6e15. 40. Hossain MP, Goyder EC, Rigby JE. El Nahas M. CKD and poverty: A growing global challenge. Am J Kidney Dis 2009;53:166e174. 41. Kimmel PL, Fwu CW, Eggers PW. Segregation, income disparities, and survival in hemodialysis patients. J Am Soc Nephrol 2013;24:293e301. 42. Congressional Budget Office. Dual-eligible beneficiaries of Medicare and Medicaid: Characteristics, health care spending, and evolving policies. Available at: https://www.cbo.gov/publication/44308. Accessed May 18, 2017. 43. Allen SM, Piette ER, Mor V. The adverse consequences of unmet need among older persons living in the community: Dual-eligible versus Medicare-only beneficiaries. J Gerontol B Psychol Sci Soc Sci 2014;69:S51eS58. 44. Kimmel PL, Fwu CW, Abbott KC, et al. Racial disparities in poverty account for mortality differences in US Medicare beneficiaries. SSM Popul Health 2016;2: 123e129. 45. Kurella Tamura M, Covinsky KE, Chertow GM, et al. Functional status of elderly adults before and after initiation of dialysis. N Engl J Med 2009;361:1539e1547. 46. Komisar HL, Feder J, Kasper JD. Unmet long-term care needs: An analysis of Medicare-Medicaid dual eligibles. Inquiry 2005;42:171e182. 47. Huskamp HA, Stevenson DG, O’Malley AJ, et al. Medicare Part D plan generosity and medication use among dual-eligible nursing home residents. Med Care 2013;51:894e900. 48. Jamison JJ, Wang J, Surbhi S, et al. Impact of Medicare Part D on racial and ethnic minorities. Divers Equal Health Care 2016;13:326e333. 49. Bowling CB, Zhang R, Franch H, et al. Underreporting of nursing home utilization on the CMS-2728 in older incident dialysis patients and implications for assessing mortality risk. BMC Nephrol 2015;16:32. 50. Foley RN, Collins AJ. The USRDS: What you need to know about what it can and can’t tell us about ESRD. Clin J Am Soc Nephrol 2013;8:845e851. 51. Eggers PW. CMS 2728: What good is it? Clin J Am Soc Nephrol 2010;5: 1908e1909.
JAMDA journal homepage: www.jamda.com
Original Study
Association of Race and Poverty With Mortality Among Nursing Home Residents on Maintenance Dialysis Robert Nee MD a, b, *, John S. Thurlow MD a, b, Keith C. Norris MD, PhD c, Christina Yuan MD a, b, Maura A. Watson DO, MPH a, b, Lawrence Y. Agodoa MD d, Kevin C. Abbott MD, MPH d a
Nephrology Service, Walter Reed National Military Medical Center, Bethesda, MD Department of Medicine, Uniformed Services University of the Health Sciences, Bethesda, MD Department of Medicine, David Geffen School of Medicine at University of CaliforniaeLos Angeles, Los Angeles, CA d National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, MD b c
a b s t r a c t Keywords: Nursing home racial disparities poverty mortality end-stage renal disease dialysis USRDS
Objectives: The association of race, ethnicity, and socioeconomic factors with survival rates of nursing home (NH) residents with treated end-stage renal disease (ESRD) is unclear. We examined whether race/ ethnicity, ZIP codeelevel, and individual-level indicators of poverty relate to mortality of NH residents on dialysis. Design: Retrospective cohort study. Participants/Setting: Using the United States Renal Data System database, we identified 56,194 nursing home residents initiated on maintenance dialysis from January 1, 2007 through December 31, 2013, followed until May 31, 2014. Measurements: We evaluated baseline characteristics of the NH cohort on dialysis, including race and ethnicity. We assessed the Medicare-Medicaid dual eligibility status as an indicator of individual-level poverty and ZIP codeelevel median household income (MHI) data. We conducted Cox regression analyses with all-cause mortality as the outcome variable, adjusted for clinical and sociodemographic factors including end-of-life preferences. Results: Adjusted Cox analysis showed a significantly lower risk of death among black vs nonblack NH residents [adjusted hazard ratio (AHR) 0.91, 95% confidence interval (CI) 0.89, 0.94]. Dual-eligibility status was significantly associated with lower risk of death compared to those with Medicare alone (AHR 0.80, 95% CI 0.78, 0.82). Compared to those in higher MHI quintile levels, NH ESRD patients in the lowest quintile were significantly associated with higher risk of death (AHR 1.09, 95% CI 1.06, 1.13). Conclusions/Implications: Black and Hispanic NH residents on dialysis had an apparent survival advantage. This “survival paradox” occurs despite well-documented racial/ethnic disparities in ESRD and NH care and warrants further exploration that could generate new insights into means of improving survival of all NH residents on dialysis. Area-level indicator of poverty was independently associated with mortality, whereas dual-eligibility status for Medicare and Medicaid was associated with lower risk of death, which could be partly explained by improved access to care. Published by Elsevier Inc. on behalf of AMDA e The Society for Post-Acute and Long-Term Care Medicine.
The views expressed in this article are those of the authors and do not reflect the official policy of the Department of Army/Navy/Air Force, the Department of Defense, National Institutes of Health or the United States government. Part of this article was presented at the American Society of Nephrology Kidney Week 2016 Annual Meeting, Chicago, IL November 17-20, 2016. This research did not receive any funding from agencies in the public, commercial, or not-for-profit sectors. Keith Norris is supported by NIH grants UL1TR000124 and P30AG021684.
The authors declare no conflicts of interest. * Address correspondence to Robert Nee, MD, Nephrology Service, Walter Reed National Military Medical Center, 8901 Wisconsin Ave, Bethesda, MD 20889. E-mail address: [email protected] (R. Nee).
https://doi.org/10.1016/j.jamda.2019.02.013 1525-8610/Published by Elsevier Inc. on behalf of AMDA e The Society for Post-Acute and Long-Term Care Medicine.
R. Nee et al. / JAMDA 20 (2019) 904e910
Racial, ethnic, and class disparities in health care exist among nursing home (NH) residents, a population that contains many people with chronic conditions that cause functional and/or cognitive limitations.1,2 Smith et al3 found that NH care remained segregated, with two-thirds of all black residents living in just 10% of all nursing homes, and these residents were significantly more likely to be served by those in the bottom quartile of many measures of quality. Mor et al4 also reported that black NH residents were overrepresented in lowrevenue, understaffed, and poor-quality nursing homes. Further, the socioeconomic status composition of NH residents is highly correlated with quality.2 The lower tier of NH care consists of those housing a high proportion of Medicaid and black residents with very limited resources and has more health-related regulatory deficiencies.4 Within the end-stage renal disease (ESRD) population, it has long been recognized that disparities in health care access and delivery, disease burden, and clinical outcomes exist among racial/ethnic and socially disadvantaged groups.5,6 Paradoxically, however, black patients on dialysis have better survival rates than their white counterparts.7 In a seminal study, Kucirka et al8 demonstrated that the commonly cited survival advantage for black dialysis patients applies only to those older than 50 years. Nonetheless, data on racial/ethnic differences in survival in the NH dialysis population remains underreported, as outcome data in the literature for this population is limited.9,10 In a cohort of long-term stay NH residents (>65 years) initiated on dialysis from 2004 to 2006 (n ¼ 3748), blacks had a lower risk of death compared to whites over 12 months [adjusted hazard ratio (AHR) 0.73, 95% confidence interval (CI) 0.68, 0.79].11 Although it is known that older black patients on chronic dialysis survive longer than their white counterparts, there remains knowledge gaps in the literature of this “survival paradox” specific to the nursing home population. The important questions of whether racial/ ethnic gaps in end-of-life preferences and socioeconomic status have a role in survival differences have not been reported and remain unanswered. Therefore, we aim to examine the relationship of race, ethnicity, and socioeconomic factors to survival rates in a large, contemporary cohort of NH residents on dialysis, accounting for patient-level factors and end-of-life care. We hypothesized that minority NH residents on dialysis would have a survival advantage, despite inequities in health care, and that indicators of poverty would be associated with mortality in this population. Methods This study used the United States Renal Data System (USRDS), which incorporates baseline and follow-up demographic and clinical data on all patients accessing the Medicare ESRD program in the United States. We conducted a retrospective cohort study of nursing home residents initiated on either hemodialysis or peritoneal dialysis from January 1, 2007 through December 31, 2013, followed until May 31, 2014. We identified a cohort of 56,194 patients who were nursing home residents at the time of dialysis initiation from the overall dialysis population of 782,161 (7.2%). This NH cohort was identified through the Centers for Medicare & Medicaid Services (CMS) Medical Evidence Form 2728 which is used to register each new patient at the onset of ESRD.12 We then merged the USRDS database with the 2010 US Census for ZIP codeelevel median household income (MHI) data. There were 1233 (2.2%) out of 56,194 NH patients who had missing MHI data. The primary outcome was all-cause mortality. This study was approved as exempt from review by the Walter Reed National Military Medical Center Institutional Review Board. Patients and Sources The demographics of the dialysis population in this study have been described in the USRDS Annual Data Reports for the years
905
studied.12 Variables included in the USRDS standard analysis files (SAFs), as well as data collection methods and validation studies, are listed on the USRDS website (www.usrds.org). The files SAF.PATIENTS were used as the primary data set and SAF.MEDEVID for additional information coded in the CMS Medical Evidence Form 2728. We obtained data on hospice care status from the SAF.DEATH file, which is derived from the death notification form (CMS-2746). Data on feeding tube use (via percutaneous endoscopic gastrostomy), as a proxy indicator of preference for life-prolonging measures, was obtained from inpatient procedure Medicare claims. We examined data on hospice care and feeding tube use because these factors could be associated with racial differences in observed survival rates. Files from the 2010 US Census (http://factfinder.census.gov/faces/nav/jsf/pages/index. xhtml) were used to merge by ZIP code with USRDS files. Because such data are area-based and thus ecological, we also assessed Medicare-Medicaid dual eligibility status from the USRDS files as an indicator of individual-level poverty13e15 and its association with mortality. Although eligibility varies by state, means testing is stricter than for either Medicare or Medicaid alone, and usually includes the poorest patients receiving care, at most <135% poverty and generally lower than 100% of poverty.14,16 Independent Variables We incorporated covariables in our models that represented potential confounders in the analysis of all-cause mortality in the ESRD population. These variables included age at initiation of dialysis, year at first ESRD service, gender, race, Hispanic ethnicity (a non-mutually exclusive category that could overlap with race), estimated glomerular filtration rate at dialysis initiation [by CKDEPI (Chronic Kidney Disease Epidemiology Collaboration) equation17], dialysis modality, vascular access used at start of dialysis, primary cause of ESRD, diabetes mellitus, hypertension, other comorbid conditions from the CMS Medical Evidence Form 2728 (Table 1), tobacco use, alcohol dependence, body mass index, serum albumin, hemoglobin, amputation, ability to ambulate, ability to transfer, requirement for assistance with daily activities, and socioeconomic factors (quintiles of ZIP codeelevel MHI, individual employment status, and insurance status including dual eligibility for Medicare and Medicaid as a surrogate for individual-level poverty). We also assessed patient-level variables related to endof-life care available in the USRDS, including withdrawal from dialysis prior to death as well as hospice care and feeding tube use as noted above. Statistical Analysis Analyses were performed using Stata 14 SE (Stata Corp, College Station, TX). Univariate analyses were performed with chi-square testing for categorical variables and Student t-test for continuous variables. P values <.05 were considered statistically significant for univariate comparisons. Survival curve was estimated using the Kaplan-Meier method and log-rank test to assess significance between survival distributions. We also conducted Cox regression analyses to model factors associated with all-cause mortality. As a sensitivity analysis, we further adjusted for covariables in the Cox model related to end-of-life care as noted above (withdrawal from dialysis, hospice, and feeding tube use). Administrative censoring occurs at the end of study (May 31, 2014). Our model development used the forced entry method in which all independent variables were entered into the model at the same time, without any specific sequence in which variables were entered. Proportional hazards assumptions were examined by graphing log [elog (survival function)] vs log (time) for the comparison groups.
906
R. Nee et al. / JAMDA 20 (2019) 904e910
Table 1 Baseline Demographic and Comorbidity Characteristics of Nursing Home Residents on Chronic Dialysis, 2007-2013, Blacks vs Nonblacks Variables Mean age (y) at start of dialysis (SD) Gender Male Female Hispanic ethnicity Dialysis modality Hemodialysis Peritoneal dialysis Vascular access used at start of dialysis Arteriovenous fistula Graft Catheter Cause of ESRD Diabetes mellitus Hypertension Glomerulonephritis Cystic kidney disease Other renal disorders Comorbid conditions Diabetes mellitus Hypertension Atherosclerotic heart disease Congestive heart failure Peripheral vascular disease Cerebrovascular disease (CVA, TIA) COPD Cancer Tobacco use Alcohol dependence Amputation Inability to ambulate Needs assistance with daily activities Inability to transfer Estimated GFR (CKD-EPI), mL/min/1.73 m2 Mean body mass index (SD) Serum albumin, g/dL (SD) Hemoglobin, g/dL (SD) Insurance status Employer group health Medicaid alone Medicare alone Dual eligibility for Medicare and Medicaid* Employed (full- or part-time) MHI quintile levelsy Bottom fifth quintile (range $6993-$46,209/y) Lower middle quintile (range $46,212-$54,989/y) Middle quintile (range $54,992-$64,534/y) Upper middle quintile (range $64,541-$80,793/y) Top fifth quintile (range $80,800-$329,112/y) Mean ZIP codeelevel MHI ($/y) (SD)y Variables related to end-of-life care Withdrawal from dialysis prior to death Hospice care status Feeding tube use
Blacks (n ¼ 14,983) 68.34 (12.7)
Nonblacks (n ¼ 41,211) 72.12 (11.8)
7003 (46.7) 7979 (53.3) 243 (1.6)
20,793 (50.5) 20,417 (49.5) 4732 (11.5)
14,511 (99.7) 41 (0.3)
40,214 (99.4) 237 (0.6)
772 (5.2) 545 (3.7) 13,537 (91.1)
2341 (5.8) 790 (1.9) 37,534 (92.3)
P Value <.001 <.001
<.001 <.001
<.001
<.001 7225 5239 350 46 1487 9743 13,316 3198 6562 2750 3915 2161 1228 830 487 1187 7299 9271 5085 10.84 29.06 2.84 9.82 405 1506 4158 7990 79 4649 3213 2564 2270 1818 58,377
(50.4) (36.5) (2.4) (0.3) (10.4)
19,242 11,489 1356 139 6423
(49.8) (29.7) (3.5) (0.4) (16.6)
(65.0) (88.9) (21.3) (43.8) (18.4) (26.1) (14.4) (8.2) (5.5) (3.2) (7.9) (48.7) (61.9) (33.9) (7.2) (9.1) (1.4) (11.6)
25,814 34,354 12,150 20,062 9098 7401 8245 4163 1945 995 2760 18,732 24,813 12,051 12.38 29.52 2.88 10.19
(62.6) (83.4) (29.5) (48.7) (22.1) (18.0) (20.0) (10.1) (4.7) (2.4) (6.7) (45.4) (60.2) (29.2) (7.6) (8.8) (2.4) (17.5)
(2.9) (10.7) (29.6) (53.3) (0.5)
1170 2220 19,013 15,589 315
(3.1) (5.8) (50.0) (37.8) (0.8)
(32.0) (22.1) (17.7) (15.6) (12.5) (23,060)
4989 8537 9440 8833 8648 67,985
(12.3) (21.1) (23.3) (21.8) (21.4) (25,727)
1980 (13.2) 1736 (11.6) 1885 (12.6)
9641 (23.4) 7557 (18.3) 2860 (6.9)
<.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001 .09 .024 <.001
<.001 .003 <.001
<.001 <.001 <.001 <.001
CKD-EPI, Chronic Kidney Disease Epidemiology Collaboration; COPD, chronic obstructive pulmonary disease; CVA, cerebrovascular accident; GFR, glomerular filtration rate; SD, standard deviation; TIA, transient ischemic attack. Data are n (%) or mean (standard deviation). Univariate analyses were performed with chi-square testing for categorical variables and Student t test for continuous variables. *Mean age of dual-eligible vs nonedual eligible patients (69.2 vs 72.5 years, respectively, P < .001). y Based on ZIP code from the 2010 US Census.
Results From a total of 782,161 incident dialysis patients in the United States from 2007 to 2013, we identified 56,194 nursing home residents. Table 1 shows the baseline demographics and characteristics of the NH cohort, of whom 14,983 (27%) were black and 41,211 (73%) were nonblack. Compared to nonblack NH residents, blacks were younger at dialysis initiation, and more likely to be female and have hypertension as the primary cause of ESRD. Further, slightly higher
proportions of blacks had indications of impaired physical function, including inability to ambulate, requirement for assistance with daily activities, and inability to transfer. Compared to nonblack NH residents, blacks were more likely to have dual-eligibility status for both Medicare and Medicaid (53.3% vs 37.8%) and in the lowest MHI quintile level (32.0% vs 12.3%). The mean ZIP codeelevel MHI was lower among black vs nonblack NH residents ($58,377 vs $67,985, P < .001). Nonblacks were more likely to be of Hispanic ethnicity and have atherosclerotic heart disease, congestive heart failure, peripheral
R. Nee et al. / JAMDA 20 (2019) 904e910
907
residents was attenuated but remained significant (AHR 0.95, 95% CI 0.92, 0.98, P < .001) after adjusting for additional variables related to end-of life care (withdrawal from dialysis, hospice, and feeding tube use). Dual-eligibility status and employer group health insurance were significantly associated with a lower risk of death compared to those with Medicare alone (AHR 0.80 and 0.87, respectively). On the other hand, higher ZIP codeelevel MHI was associated with a lower risk of death, in a near graded fashion. In a separate Cox model, we found that NH residents on dialysis in the lowest quintile had an AHR 1.09 (95% CI 1.06, 1.13) compared to those in higher area-level MHI quintile levels. There were no significant interactions between race (P ¼ .80), Hispanic ethnicity (P ¼ .76), and area-level MHI. Discussion
Fig. 1. Patient survival among nursing home residents on chronic dialysis, by race.
vascular disease, chronic obstructive pulmonary disease, and cancer. In regard to end-of-life preferences, black NH residents were less likely to withdraw from dialysis and enroll in hospice but more likely to undergo feeding tube placement than their nonblack counterparts. The overall crude mortality rate in NH residents on dialysis was 480 per 1000 patient-years [PYs] (95% CI 475-484 per 1000 PYs), with a mean follow-up duration of 2.6 years. Black NH residents had a lower crude mortality rate than nonblacks (406 vs 511 per 1000 PYs, respectively, P < .001) (Figure 1). Similarly, Hispanic NH residents had a lower crude mortality rate than non-Hispanics (407 vs 488 per 1000 PYs, respectively, P < .001) (Figure 2). On sensitivity analysis, nonHispanic blacks had a lower crude mortality rate than non-Hispanic whites (405 vs 533 per 1000 PYs, respectively, P < .001). As shown in Table 2, in the adjusted Cox regression model, black NH residents had a significantly lower mortality rate compared to nonblacks (AHR 0.91). Similarly, Hispanic NH residents had a significantly lower mortality rate compared to non-Hispanics (AHR 0.89). Thus, black race and Hispanic ethnicity among NH dialysis patients were independently associated with greater longevity, accounting for multiple covariates including demographics, dialysis vascular access, comorbid conditions, body mass index, serum albumin, hemoglobin, indicators of impaired physical function, insurance status, and individual-level and ZIP codeelevel measures of poverty. On sensitivity analysis, non-Hispanic blacks also had a significantly lower mortality rate than non-Hispanic whites (AHR 0.90, 95% CI 0.87, 0.92, P < .001). Further, the survival advantage among black vs nonblack NH
Fig. 2. Patient survival among nursing home residents on chronic dialysis, by ethnicity.
We found in a national, multiyear study of nursing home residents on maintenance dialysis that blacks and Hispanics had improved survival compared with their nonblack and non-Hispanic counterparts. A similar survival advantage occurred among non-Hispanic blacks vs non-Hispanic whites on dialysis. These findings persisted even after accounting for clinical, sociodemographic, and individual preferences for life-prolonging measures. Our findings are consistent with the “survival paradox” phenomenon in the general ESRD population in which racial- and ethnic-minority groups survive longer than whites, despite their poor intermediate health outcomes and worse clinical performance measures.5,7,12,18,19 Our findings validate and extend the findings of Kucirka et al, who demonstrated that the survival advantage for black patients undergoing dialysis applies only to older adults.8 In a subgroup of patients between 61 and 70 years of age in their study, the investigators demonstrated that blacks had significantly lower mortality than whites (AHR 0.87, 95% CI 0.86, 0.88) and those <50 years had a higher risk of death. The survival paradox is even more surprising given welldocumented racial and ethnic disparities in outcomes and delivery of ESRD care.6,20e22 Furthermore, minority groups do not receive nursing home care of comparable quality to whites.1,2,23 Mor et al4 found that black NH residents were nearly 4 times more likely than whites to reside in nursing homes with limited resources and poor performance. Smith et al24 reported that blacks were more likely to be in nursing homes that were understaffed, cited for deficiencies and terminated from the Medicare and Medicaid program. The underlying mechanisms for reduced mortality in older black dialysis patients remain elusive. Potential explanations include survival bias (black patients who survive long enough to reach ESRD may be healthier than white patients), cultural differences in adaptation to chronic illness, favorable nutritional and anthropometric characteristics, differential response to inflammation, lower prevalence of cardiovascular disease, and a differing cardiovascular risk profile including higher blood pressure, which paradoxically may have a beneficial association with survival in ESRD patients, referred to as “reverse epidemiology” of cardiovascular disease.19,25e27 Furthermore, the paradoxical survival advantage may in part be due to comparatively lower rates of kidney transplantation among blacks vs whites.28 To the extent that black patients are less likely to receive a transplant, those who initiate or remain on dialysis may be relatively healthier than their white counterparts, thus inflating the black survival advantage in older adults.29 In the nursing home population, the apparent survival advantage among black residents may be partly due to the lower likelihood of enrolling in hospice care compared to whites.30e33 Eneanya et al34 demonstrated that blacks with advanced chronic kidney disease were less likely to communicate end-of-life preferences with family members and more likely to prefer life-extending treatments compared to whites. Prior studies have also shown that blacks may be less likely to withdraw
908
R. Nee et al. / JAMDA 20 (2019) 904e910
Table 2 Multivariable Cox Regression Analyses of Baseline Patient Characteristics for Death Among Nursing Home Residents on Chronic Dialysis Variables
Adjusted HR
95% CI
P Value
Age at start of dialysis Year at first ESRD service Gender (male vs female) Race (black vs nonblack) Ethnicity (Hispanic vs non-Hispanic) Dialysis modality (hemodialysis vs peritoneal dialysis) Vascular access used at start of dialysis Catheter Graft Arteriovenous fistula Cause of ESRD Diabetes mellitus Hypertension Glomerulonephritis Cystic kidney disease Other renal disorders Comorbid conditions Diabetes mellitus Hypertension Atherosclerotic heart disease Congestive heart failure Peripheral vascular disease Cerebrovascular disease (CVA, TIA) COPD Cancer Tobacco use Alcohol dependence Amputation Inability to ambulate Needs assistance with daily activities Inability to transfer Estimated GFR (<10 mL/min/1.73 m2 vs >10 mL/min/1.73 m2) Mean body mass index (>30 vs <30) Serum albumin (<3.0 vs >3.0 g/dL) Hemoglobin (<9 vs >9 g/dL) Insurance status Medicare alone Medicaid alone Employer group health Dual eligibility for Medicare and Medicaid Employed MHI quintile levels Bottom fifth quintile (range $6993-$46,209/y) Lower middle quintile (range $46,212-$54,989/y) Middle quintile (range $54,992-$64,534/y) Upper middle quintile (range $64,541-$80,793/y) Top fifth quintile (range $80,800-$329,112/y)
1.02 0.96 1.02 0.91 0.89 1.19
1.02, 0.95, 1.00, 0.89, 0.85, 0.57,
1.03 0.97 1.05 0.94 0.93 2.49
<.001 <.001 .045 <.001 <.001 .65
1.0 (referent) 0.79 0.78
0.73, 0.85 0.75, 0.82
<.001 <.001
1.0 (referent) 1.00 0.87 0.90 0.99
0.97, 0.81, 0.74, 0.95,
1.03 0.93 1.09 1.02
.99 <.001 .27 .50
0.97 0.88 0.96 1.16 1.04 1.04 1.16 1.15 1.06 1.07 1.04 1.07 1.02 1.24 0.87 0.90 1.12 0.96
0.94, 0.85, 0.94, 1.13, 1.01, 1.01, 1.13, 1.11, 1.00, 1.00, 0.99, 1.04, 0.99, 1.21, 0.85, 0.88, 1.10, 0.93,
1.00 0.90 0.99 1.18 1.07 1.07 1.19 1.20 1.12 1.15 1.08 1.10 1.04 1.28 0.89 0.93 1.15 0.98
.035 <.001 .003 <.001 .007 .005 <.001 <.001 .035 .06 .13 <.001 .13 <.001 <.001 <.001 <.001 <.001
1.0 (referent) 1.04 0.87 0.80 0.79
0.99, 0.82, 0.78, 0.67,
1.09 0.94 0.82 0.92
.12 <.001 <.001 .002
1.0 (referent) 0.96 0.90 0.90 0.88
0.93, 0.87, 0.87, 0.84,
0.99 0.93 0.93 0.91
.049 <.001 <.001 <.001
COPD, chronic obstructive pulmonary disease; CVA, cerebrovascular accident; GFR, glomerular filtration rate; TIA, transient ischemic attack.
from dialysis.35e37 In a recent analysis of a large cohort of dialysis patients dying between 2000 and 2014 (not specific to nursing home residents), non-Hispanic blacks were less likely to discontinue dialysis and less likely to receive hospice care than non-Hispanic whites [adjusted odds ratios 0.51 (95% CI 0.50, 0.52) and 0.58 (95% CI 0.57, 0.59), respectively].38 The findings from these studies are consistent with our analyses that suggest greater preferences for life-prolonging measures among black NH residents on maintenance dialysis, based on USRDS data on withdrawal from dialysis, hospice, and feeding tube use. Although these end-of-life care factors partly accounted for the survival advantage in black vs nonblack NH residents (AHR for mortality increased from 0.91 to 0.95), the racial difference in mortality remained significant. Given the interdependency of race, ethnicity, and socioeconomic factors on adverse clinical outcomes in patients with ESRD,39,40 we assessed whether poverty attenuates the survival advantage among black NH residents on dialysis. We demonstrated that NH residents on dialysis in the lowest area-level MHI quintile had a higher risk of death compared to those in higher quintile levels, independent of race/ethnicity. However, there was no significant interaction between MHI and race/ethnicity in predicting
mortality. In other words, area-level MHI did not significantly modify the survival advantage among black NH residents on dialysis. These findings corroborate with a USRDS study by Kimmel et al41 who showed that higher ZIP codeelevel MHI was associated with longer survival in both black and white patients in the general hemodialysis population and that controlling for income did not modify black patients’ survival advantages. Our findings of an inverse relationship between income and mortality risk is not surprising given that poor and minority NH residents served by lower-tier nursing homes with limited resources are more likely to receive substandard care.4 Dual-eligibility status for both Medicare and Medicaid, as a proxy measure of individual-level poverty, was significantly associated with lower mortality. This association is incongruent with the ecologic measure of income using ZIP codes. Dual-eligible beneficiaries represent a disadvantaged subgroup of older Americans who are generally impoverished and have higher prevalence of physical and cognitive impairments, less education, and lower levels of social support than their Medicare-only counterparts.42,43 Kimmel et al44 reported that dual-eligibility status, as a proxy measure of poverty, was a major driver for racial mortality differences in the
R. Nee et al. / JAMDA 20 (2019) 904e910
aged US Medicare population. Thus, it remains unclear why dualeligible NH residents on dialysis had better survival. It is possible that poorer NH residents may gain Medicaid eligibility earlier than those with more resources and assets; thus, there may be a subset of dual-eligible residents who have a shorter NH “vintage” which favors their chance of survival. Kurella et al45 reported that initiation of dialysis among NH residents with ESRD was associated with a marked decline in function status and progressively higher mortality over time. Also, it is worth noting that dual-eligible NH residents in our cohort were younger at the time of dialysis initiation than their nonedual eligible counterparts (69.2 vs 72.5 years, respectively). We adjusted for age in the multivariable Cox model, but there could be residual confounders related to the younger age that may partly explain the improved survival in dual-eligible beneficiaries. Perhaps the NH ESRD population with dual eligibility represents a unique impoverished group of patients who benefit from joint Medicare and Medicaid coverage. The dualeligible program was designed to supplement patients who have financial and other disadvantages and thus may improve access to care in the ESRD population, with minimal to no cost-sharing. Although a survey indicated an unmet need for assistance with activities of daily living among dual-eligible beneficiaries, they encountered few problems obtaining medical care.46 Only 3% reported not having a regular physician or other regular source of care, and only 1% were unable to arrange specialist care. Also, the Medicare Part D program provides special protections for dualeligible NH residents that may ameliorate the health impact from coverage limitations of prescription drug plans,47 however less wealthy patients who do not qualify for Medicaid but have Medicare Part D may be unable to afford the out-of-pocket expenses when they fall into a coverage gap, often known as the donut hole.48 Our study has certain limitations. First, we cannot make conclusions about causality given the retrospective nature of our study. Second, there could be ascertainment bias related to providers’ responses on CMS Medical Evidence Form 2728, as demonstrated by Bowling et al49 who reported underestimation of NH utilization among older patients (75 years) with incident ESRD; however, those identified by the CMS Form 2728 as being in an NH have a high probability of actually being in an NH at the time of dialysis initiation (positive predictive value 80%). Third, duration of time spent in the NH pre- and post-dialysis initiation could not be ascertained from the CMS Form 2728 and may bias results as noted above. Fourth, in the absence of individual-level income data, we used ZIP codeebased MHI as a surrogate for patient income. We acknowledge potential biases associated with ZIP code as a proxy measure of individual-level socioeconomic status since an association observed with income on an aggregate level may not represent the association that exists at an individual income level. Also, residual confounders may exist between residents of higher vs lower MHI ZIP codes, such as disparities in health care access or quality. Fifth, dual eligibility is by definition binary and provides no additional information for income above poverty level. Lastly, there were significant missing data in the USRDS on advance directives and were therefore not included in our analyses. The strengths of the current study include the large sample size as reflected by broad representativeness of the USRDS registry, with nearly complete inclusion of all ESRD patients in the United States.50 Comparisons of mortality rates have high validity given the accurate and complete ascertainment of death data in the USRDS.51 Also, as noted above, the high positive predictive value of the NH indicator on CMS Form 2728 ensures that the majority of patients have the condition of interest in our study cohort.48 Importantly, we have also incorporated in our analyses patient-level factors that are particularly relevant in the nursing home setting and end-of-life preferences.
909
Conclusions and Implications Black and Hispanic NH residents on dialysis demonstrate an apparent survival advantage, despite well-documented inequities in health care access and quality. This “survival paradox” warrants further exploration that could generate new insights into means of improving survival of all NH residents on dialysis. Area-level indicator of poverty was independently associated with mortality but did not modify racial differences in survival. Dual-eligibility status for both Medicare and Medicaid was associated with lower risk of death, which could be partly explained by improved access to care. These findings highlight the important role of social determinants of health in longevity among NH residents on dialysis.
References 1. Grabowski DC, McGuire TG. Black-white disparities in care in nursing homes. Atl Econ J 2009;37:299e314. 2. Konetzka RT, Werner RM. Disparities in long-term care: Building equity into market-based reforms. Med Care Res Rev 2009;66:491e521. 3. Smith DB, Feng Z, Fennell ML, et al. Separate and unequal: Racial segregation and disparities in quality across U.S. nursing homes. Health Aff (Millwood) 2007;26:1448e1458. 4. Mor V, Zinn J, Angelelli J, et al. Driven to tiers: Socioeconomic and racial disparities in the quality of nursing home care. Milbank Q 2004;82:227e256. 5. Norris KC, Agodoa LY. Unraveling the racial disparities associated with kidney disease. Kidney Int 2005;68:914e924. 6. Reddan DN, Szczech LA, Klassen PS, Owen WF Jr. Racial inequity in America’s ESRD program. Semin Dial 2000;13:399e403. 7. Agodoa L, Eggers P. Racial and ethnic disparities in end-stage kidney failuresurvival paradoxes in African-Americans. Semin Dial 2007;20:577e585. 8. Kucirka LM, Grams ME, Lessler J, et al. Association of race and age with survival among patients undergoing dialysis. JAMA 2011;306:620e626. 9. Hall RK, O’Hare AM, Anderson RA, Colón-Emeric CS. End-stage renal disease in nursing homes: A systematic review. J Am Med Dir Assoc 2013;14:242e247. 10. Anderson JE, Kraus J, Sturgeon D. Incidence, prevalence, and outcomes of endstage renal disease patients placed in nursing homes. Am J Kidney Dis 1993;21: 619e627. 11. United States Renal Data System. 2010 USRDS Annual Data Report: Epidemiology of Kidney Disease in the United States, vol. 1. Bethesda, MD: National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases; 2010. p. 100. 12. United States Renal Data System. 2017 USRDS Annual Data Report: Epidemiology of Kidney Disease in the United States. Bethesda, MD: National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases; 2017. 13. McMillan A, Pine PL, Gornick M, Prihoda R. A study of the “crossover population”: Aged persons entitled to both Medicare and Medicaid. Health Care Financ Rev 1983;4:19e46. 14. Hagland M. Medicaid/Medicare dual eligibles. Healthplan 1997;38:52e56. 58, 60e61. 15. Nemore P. Dual eligibles. Issue Brief Cent Medicare Educ 2004;5:1e6. 16. United States Census Bureau: Poverty. Available at: https://www.census.gov/ topics/income-poverty/poverty/guidance/poverty-measures.html. Accessed February 15, 2018. 17. Levey AS, Stevens LA, Schmid CH, et al. CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration). A new equation to estimate glomerular filtration rate. Ann Intern Med 2009;150:604e612. 18. Kalantar-Zadeh K, Kovesdy CP, Derose SF, et al. Racial and survival paradoxes in chronic kidney disease. Nat Clin Pract Nephrol 2007;3:493e506. 19. Buckalew VM Jr, Freedman BI. Reappraisal of the impact of race on survival in patients on dialysis. Am J Kidney Dis 2010;55:1102e1110. 20. Powe NR, Melamed ML. Racial disparities in the optimal delivery of chronic kidney disease care. Med Clin North Am 2005;89:475e488. 21. Prakash S, Rodriguez RA, Austin PC, et al. Racial composition of residential areas associates with access to pre-ESRD nephrology care. J Am Soc Nephrol 2010;21: 1192e1199. 22. Ifudu O, Dawood M, Iofel Y, et al. Delayed referral of black, Hispanic, and older patients with chronic renal failure. Am J Kidney Dis 1999;33:728e733. 23. Fennell ML, Miller SC, Mor V. Facility effects on racial differences in nursing home quality of care. Am J Med Qual 2000;15:174e181. 24. Smith DB, Feng Z, Fennell ML, et al. Racial disparities in access to long-term care: The illusive pursuit of equity. J Health Polit Policy Law 2008;33:861e881. 25. Yan G, Norris KC, Yu AJ, et al. The relationship of age, race, and ethnicity with survival in dialysis patients. Clin J Am Soc Nephrol 2013;8:953e961. 26. Nee R, Agodoa LY. Racial differences in kidney disease and end-stage kidney disease in the USA. In: Garcia-Garcia G, Agodoa LY, Norris KC, editors. Chronic Kidney Disease in Disadvantaged Populations. San Diego, CA: Academic Press; 2017. p. 65e72.
910
R. Nee et al. / JAMDA 20 (2019) 904e910
27. Kalantar-Zadeh K, Block G, Humphreys MH, Kopple JD. Reverse epidemiology of cardiovascular risk factors in maintenance dialysis patients. Kidney Int 2003; 63:793e808. 28. Rhee CM, Lertdumrongluk P, Streja E, et al. Impact of age, race and ethnicity on dialysis patient survival and kidney transplantation disparities. Am J Nephrol 2014;39:183e194. 29. Kalantar-Zadeh K, Kovesdy CP, Norris KC. Racial survival paradox of dialysis patients: Robust and resilient. Am J Kidney Dis 2012;60:182e185. 30. Zheng NT, Mukamel DB, Caprio T, et al. Racial disparities in in-hospital death and hospice use among nursing home residents at the end of life. Med Care 2011;49:992e998. 31. Kwak J, Haley WE, Chiriboga DA. Racial differences in hospice use and inhospital death among Medicare and Medicaid dual-eligible nursing home residents. Gerontologist 2008;48:32e41. 32. Murray AM, Arko C, Chen SC, et al. Use of hospice in the United States dialysis population. Clin J Am Soc Nephrol 2006;1:1248e1255. 33. Thomas BA, Rodriguez RA, Boyko EJ, et al. Geographic variation in black-white differences in end-of-life care for patients with ESRD. Clin J Am Soc Nephrol 2013;8:1171e1178. 34. Eneanya ND, Wenger JB, Waite K, et al. Racial disparities in end-of-life communication and preferences among chronic kidney disease patients. Am J Nephrol 2016;44:46e53. 35. Kurella Tamura M, Goldstein MK, Pérez-Stable EJ. Preferences for dialysis withdrawal and engagement in advance care planning within a diverse sample of dialysis patients. Nephrol Dial Transplant 2010;25:237e242. 36. Nelson CB, Port FK, Wolfe RA, Guire KE. The association of diabetic status, age, and race to withdrawal from dialysis. J Am Soc Nephrol 1994;4:1608e1614. 37. Port FK, Wolfe RA, Hawthorne VM, Ferguson CW. Discontinuation of dialysis therapy as a cause of death. Am J Nephrol 1989;9:145e149. 38. Foley RN, Sexton DJ, Drawz P, et al. Race, ethnicity, and end-of-life care in dialysis patients in the United States. J Am Soc Nephrol 2018;29:2387e2399.
39. Nicholas SB, Kalantar-Zadeh K, Norris KC. Socioeconomic disparities in chronic kidney disease. Adv Chronic Kidney Dis 2015;22:6e15. 40. Hossain MP, Goyder EC, Rigby JE. El Nahas M. CKD and poverty: A growing global challenge. Am J Kidney Dis 2009;53:166e174. 41. Kimmel PL, Fwu CW, Eggers PW. Segregation, income disparities, and survival in hemodialysis patients. J Am Soc Nephrol 2013;24:293e301. 42. Congressional Budget Office. Dual-eligible beneficiaries of Medicare and Medicaid: Characteristics, health care spending, and evolving policies. Available at: https://www.cbo.gov/publication/44308. Accessed May 18, 2017. 43. Allen SM, Piette ER, Mor V. The adverse consequences of unmet need among older persons living in the community: Dual-eligible versus Medicare-only beneficiaries. J Gerontol B Psychol Sci Soc Sci 2014;69:S51eS58. 44. Kimmel PL, Fwu CW, Abbott KC, et al. Racial disparities in poverty account for mortality differences in US Medicare beneficiaries. SSM Popul Health 2016;2: 123e129. 45. Kurella Tamura M, Covinsky KE, Chertow GM, et al. Functional status of elderly adults before and after initiation of dialysis. N Engl J Med 2009;361:1539e1547. 46. Komisar HL, Feder J, Kasper JD. Unmet long-term care needs: An analysis of Medicare-Medicaid dual eligibles. Inquiry 2005;42:171e182. 47. Huskamp HA, Stevenson DG, O’Malley AJ, et al. Medicare Part D plan generosity and medication use among dual-eligible nursing home residents. Med Care 2013;51:894e900. 48. Jamison JJ, Wang J, Surbhi S, et al. Impact of Medicare Part D on racial and ethnic minorities. Divers Equal Health Care 2016;13:326e333. 49. Bowling CB, Zhang R, Franch H, et al. Underreporting of nursing home utilization on the CMS-2728 in older incident dialysis patients and implications for assessing mortality risk. BMC Nephrol 2015;16:32. 50. Foley RN, Collins AJ. The USRDS: What you need to know about what it can and can’t tell us about ESRD. Clin J Am Soc Nephrol 2013;8:845e851. 51. Eggers PW. CMS 2728: What good is it? Clin J Am Soc Nephrol 2010;5: 1908e1909.