Abstract
Objective
Prevalence of end stage renal disease, modality of treatment and type of hemodialysis vascular access utilized varies widely by race/ethnicity in the United States, but outcomes of hemodialysis vascular access by race/ethnicity are poorly described. The objective of this study is to evaluate variations in outcomes of hemodialysis vascular access in the elderly by race/ethnicity.
Methods
Medicare Outpatient, Inpatient and Carrier files were queried from 2006-2011 for beneficiaries that were age ≥66 and dialysis dependent at time of index fistula/graft creation, qualified for Medicare by age only and were continuously enrolled in Medicare twelve months before and after index fistula/graft creation. Primary outcome measures were early vascular access failure and 12-month failure-free survival, specifically, the variation in the difference between fistula and graft in non-White versus White race/ethnicity groups..
Results
Fistulas comprised a smaller proportion of index procedures performed in Blacks (65.9%, P<.001) and Asians (71.4%, P<.001), compared to Whites (78.0%) with no difference in Hispanics (78.7%, P=.59). Incidence of early failure after graft versus fistula was: Whites, 34.9% versus 43.5% (P<.001), Blacks, 32.9% versus 49.1% (P<.001), Asians, 30.8% versus 40.5% (P=.014) and Hispanics 35.2% versus 43.2% (P=.005). The difference in early failure after fistula versus graft in Blacks was significantly larger than the difference in Whites (P<.001). 12-month failure-free survival after index graft versus fistula was: Whites, 41.9% versus 38.9% (P=.008), Blacks 48.5% versus 37.3% (P<.001), Asians 51.6% versus 45.2% (P=.98) and Hispanics 51.9% versus 42.2% (P<.001). The difference in 12-month failure-free survival after graft versus fistula in Blacks and in Hispanics was larger than the difference in Whites (P<.001 and P=.02, respectively).
Conclusions
Outcomes of fistulas versus grafts in the elderly vary significantly by race/ethnicity. The decreased risk of early failure after graft versus fistula creation is larger in Blacks compared to Whites. The higher failure-free survival at 12 months after graft versus fistula creation is larger in Blacks compared to Whites and trends toward being larger in Hispanics compared to Whites.
Introduction
The prevalence of end stage renal disease (ESRD) in the United States is higher in all non-White racial groups compared to Whites.1 In 2013, the prevalence per million among Blacks was higher than any other racial group at 5,584 compared to 2,196 among Asians, 2,133 among Native Americans, 2,970 among Hispanics and 1,499 among Whites.1 Additionally, among prevalent ESRD patients, the percentage of Blacks that utilize hemodialysis as their treatment modality is the highest, at 76%, compared to 59% of Whites, 70% of Native Americans, 61% of Asians and 69% of Hispanics.1 Furthermore, the type of vascular access utilized varies significantly by race/ethnicity. Among hemodialysis patients in 2013, 64.3% of Whites were using an arteriovenous fistula for vascular access, compared to 56.9% of Blacks, 67.2% of Asians and 67.5% of Hispanics.1
Despite the recognition that these racial/ethnic disparities exist, little work has been done to examine the outcomes of vascular access type by race/ethnicity. Outcomes of hemodialysis vascular access are particularly relevant in the elderly given the 31% per year all-cause mortality in the age 65+ hemodialysis population.1 National guidelines, initiatives and the recently instituted Center for Medicare and Medicaid Services (CMS) End Stage Renal Disease (ESRD) Quality Incentive Program (QIP) all place increasing emphasis on prioritizing the creation of fistulas over grafts for hemodialysis vascular access.2,3 However, there is increasing evidence that fistula maturation and patency may be worse in elderly compared to younger ESRD patients.4,5,6 The combination of elevated one-year mortality rates in elderly ESRD patients and lower fistula maturation/ patency suggests that a significant proportion of elderly hemodialysis patients may not survive to experience the purported long-term benefits that fistulas provide versus grafts, making the short-term results of fistula/graft creation for hemodialysis vascular access an important area of investigation.
The purpose of this study is to determine the variations in 12-month outcomes of fistula versus graft creation among non-White race/ethnicity groups compared to Whites in dialysis dependent patients undergoing first time fistula/graft creation in the Medicare population.
Methods
Medicare Outpatient, Inpatient and Carrier files from 2006 through 2011 were queried for upper extremity vascular access procedures identified by the current procedural terminology (CPT) codes for fistula and graft (Appendix Table 1). Index fistula/graft creation was defined as the first occurrence of the CPT code per beneficiary during the study period. Repeat fistula/graft creation was defined as any new fistula/graft creation, identified by CPT code, performed in the twelve months subsequent to index fistula/graft, regardless of whether the index procedure was a fistula or a graft. Tunneled hemodialysis catheter placement procedures were identified by CPT code (Appendix Table 1).
Inclusion criteria were age ≥66 and dialysis dependent at time of index fistula/graft creation, qualification for Medicare exclusively by virtue of age, complete demographic and co-morbidity data, continuous enrollment in fee-for-service Medicare for twelve months before and after index fistula/graft creation. By excluding patients who qualified for Medicare by virtue of ESRD and limiting the patient population to those who were enrolled in Medicare for twelve months before index fistula/graft creation, we limit the cohort to patients who were undergoing first time dialysis vascular access creation to the best of our ability. Patients who died within twelve months after index fistula/graft creation were included. Co-morbidities were defined by having ever met the claims criteria of the Chronic Condition Data Warehouse as of the date of index fistula/graft creation.7
Dialysis dependence was identified using Berenson-Eggers Type of Service (BETOS) Codes P9A (Dialysis Services Medicare Fee Schedule) and P9B (Dialysis Services Non-Medicare Fee Schedule). BETOS codes are assigned for each Health Care Financing Administration Common Procedure Coding System (HCPCS) procedure code. Each HCPCS code is assigned to a single BETOS code. BETOS code categories are designed to permit objective assignment and be stable over time.8 Chronic dialysis dependence was defined as at least one occurrence of BETOS 9A or 9B during the 12 months preceding index fistula/graft month or within 30 days after creation of index fistula/graft.9
In Medicare claims data, beneficiary race is determined by taking the beneficiary race code that has historically been used by the Social Security Administration (SSA) and applying an algorithm developed by the Research Triangle Institute (RTI). The RTI algorithm incorporates beneficiary language preference, the language that the beneficiary has requested that the SSA use when sending notices, the source of the beneficiary’s Medicare enrollment database race/ethnicity code, the state in which the beneficiary resides and the first and last name of the beneficiary. The RTI algorithm has been shown to significantly improve the accuracy of coding for Hispanics and Asians or Pacific Islanders.10 The race categories designated by Medicare claims data are non-Hispanic White, (hereafter referred to as White), Black, Asian/Pacific Islander (hereafter referred to as Asian), Hispanic, American Indian/Alaska Native and Other. Outcomes were adjusted for using three increasingly comprehensive models. Model 1 co-variates were age and its square, sex, index year, index month, total covered charges from outpatient, inpatient and carrier files in the year preceding index fistula/graft creation date, inpatient setting for the index fistula/graft creation and all co-morbidities available from the Chronic Condition Data Warehouse, (except chronic kidney disease which was used to select the cohort). In Model 2, community contextual factors (obtained by linking residential zip code from Medicare enrollment files to the 2000 Census) were included in addition to all co-variates in Model 1.11,12 In Model 3, binary indicator variables for hospital referral region (HRR) were included in addition to all co-variates in Model 2. HRRs represent regional health care markets and are included in an effort to control for regional variation in access to health care and practice patterns surrounding ESRD care.13
Statistical analysis was conducted using Stata, version 13.1 (StataCorp, College Station, Texas). Two group comparisons for age were assessed by a t-test of equality of means. Two group comparisons for dichotomous outcomes were assessed by chi-squared tests. Prevalence of access procedure type was analyzed by race/ethnicity groups using logistic regression, controlling for the factors described in the various models. Logistic regressions of early failure were performed based on index fistula versus graft, controlling for the factors described in the various models. Adjusted outcomes were calculated based on predictions which varied the access procedure for each patient while holding all other covariates fixed, averaged across the sample.14 The primary outcome measures of this study were early failure and 12-month failure free survival-specifically, the variations in the difference between fistula and graft of these two outcomes among race/ethnicity groups. Early failure was defined by creation of another fistula/graft and/or tunneled hemodialysis catheter placement in the 12 months following index fistula/graft. Calculations of incidence of early failure include patients who expired during the 12 months following index fistula/graft procedure in the denominator. 12-month failure-free survival was defined by the patient surviving for 12 months after index fistula/graft creation without occurrence of another fistula/graft creation or tunneled catheter placement. Survival analysis was performed.15
In order to account for multiple comparisons, a Bonferroni correction can be applied to the P-value for our two primary outcomes, the variation of the difference in fistula and graft for non-White versus White for early failure and 12-month failure free survival. For each primary outcome, there were three hypotheses for equality of Black, Asian/PI and Hispanic, respectively, with Whites. This results in a total of six hypotheses. After applying a Bonferroni correction, the required p-value for statistical significance on each hypothesis is lowered from .05 to .008. As the Bonferroni correction can be a conservative approach, both the corrected and uncorrected significance are presented.
This research was deemed exempt from review by the Institutional Review Board as the study was retrospective, the Medicare data utilized was de-identified and there was no contact with human subjects for this research. Patient informed consent was unnecessary for this study as only de-identified data was used and there was no contact with human subjects.
Results
During the study period, 16,402 index fistula and graft creations met the inclusion criteria. 10,507 (64.1%) procedures were performed in Whites, 3,331 (20.3%) in Blacks, 685 (4.2%) in Asians, 1,570 (9.6%) in Hispanics, 146 (0.89%) in American Indians / Alaska Natives and 163 (0.99%) in Other. The small sample size of American Indians / Alaska Natives and Other would preclude meaningful statistical analysis and the results of both groups are not further presented. Patient demographics, co-morbidities and zip code community contextual factors varied by graft versus fistula and across racial/ethnic groups (Table I, Appendix Table 2). The mean age of patients who underwent graft was slightly older than patients who underwent fistula in all race/ethnicity groups, and a smaller proportion of females underwent fistula. Patients who underwent graft had higher mean total covered billed charges in the twelve months prior to the fistula/graft creation date in all race/ethnicity groups. There was a trend towards a higher incidence of co-morbidities in the graft group compared to the fistula group across race/ethnicity groups. In most cases across race/ethnicity groups, when there was a difference in zip code community contextual factors between the fistula and graft group, the graft group was more likely to be associated with disadvantaged community contextual factors.
Table 1.
Patient Demographics, Co-Morbidities and Zip Code Community Contextual Factors
| White | Black | Asian | Hispanic | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Index fistula (%) (n=8,397) |
Index graft (%) (n=2,110) |
P-value | Index fistula (%) (n=2,071) |
Index graft (%) (n=1,260) |
P-value | Index fistula (%) (n=467) |
Index graft (%) (n=218) |
P-value | Index fistula (%) (n=1,173) |
Index graft (%) (n=397) |
P-value | |
| Mean age, years | 77.8 | 78.8 | <.001 | 75.6 | 76.8 | <.001 | 76.7 | 79.3 | <.001 | 75.3 | 76.9 | <.001 |
| Female | 3,489 (41.6) |
1,147 (54.4) |
<.001 | 1,095 (52.9) |
836 (66.3) |
<.001 | 225 (48.2) |
124 (56.9) |
.03 | 555 (47.3) |
249 (62.7) |
<.001 |
| Index fistula/graft performed in an inpatient setting |
1,560 (18.6) |
549 (26) | <.001 | 441 (21.3) | 318 (25.2) |
.008 | 115 (24.6) |
60 (27.5) | .42 | 278 (23.6) |
106 (26.7) |
.23 |
| Mean total covered billed charges in 12 months prior to index fistula/graft creation date (in thousands of dollars) |
184.9 | 251.6 | <.001 | 190.3 | 245.8 | <.001 | 208.2 | 256.1 | .03 | 229.2 | 287.1 | <.001 |
| Year of index procedure | *Appendix Table 2 | <.001 | *Appendix Table 2 | .001 | *Appendix Table 2 | .009 | *Appendix Table 2 | .23 | ||||
| Month of index procedure | *Appendix Table 2 | .28 | *Appendix Table 2 | .06 | *Appendix Table 2 | .49 | *Appendix Table 2 | .83 | ||||
| Co-morbidities | ||||||||||||
| Hypertension | 8,324 (99.1) |
2,089 (99) | .58 | 2,043 (98.6) |
1,249 (99.1) |
.21 | - | - | - | - | - | - |
| Diabetes | 5,777 (68.8) |
1,501 (71) | .04 | 1,640 (79.2) |
1037 (82.3) |
.03 | 389 (83.3) |
183 (83.9) |
.83 | 1029 (87.7) |
362 (91.2) |
.06 |
| Atrial Fibrillation | 2,900 (24.5) |
835 (39.6) |
<.001 | 363 (17.5) |
233 (18.5) |
.48 | 102 (21.8) |
52 (23.9) |
.56 | 184 (15.7) |
81 (20.4) |
.03 |
| Heart Failure | 6,606 (78.7) |
1,790 (84.8) |
<.001 | 1,616 (78) |
1028 (81.6) |
.01 | 345 (73.9) |
188 (86.2) |
<.001 | 925 (78.9) |
334 (84.1) |
.02 |
| Stroke/Transient Ischemic Attack |
2,076 (24.7) |
640 (30.3) | <.001 | 538 (26) | 444 (35.2) |
<.001 | 104 (22.3) |
67 (30.7) |
.02 | 309 (26.3) |
119 (30) | .16 |
| Ischemic Heart Disease | 6,881 (81.9) |
1,785 (84.6) |
.004 | 1,539 (74.3) |
1,001 (79.4) |
.001 | 355 (76) |
187 (85.8) |
.003 | 909 (77.5) |
341 (85.9) |
<.001 |
| Acute Myocardial Infarction | 1,421 (16.9) |
374 (17.1) | .38 | 212 (10.2) | 173 (13.7) |
.002 | 61 (14.3) |
31 (14.2) |
.68 | 143 (12.2) |
52 (13.1) |
.64 |
| Chronic Obstructive Pulmonary Disease |
4,003 (47.7) |
1,075 (50.9) |
.007 | 732 (35.3) | 505 (40) | .006 | 154 (33) | 95 (43.6) |
.007 | 394 (33.6) |
187 (47) | <.001 |
| Hyperlipdemia | 7,385 (87.9) |
1,815 (86) | .02 | 1,655 (79.9) |
1,015 (80.6) |
.65 | 402 (86) | 197 (90.4) |
.12 | 1,003 (85.5) |
346 (87.2) |
.42 |
| Anemia | 8,211 (97.8) |
2,082 (98.7) |
.01 | NAa | NAa | - | NAa | NAa | - | NAa | NAa | - |
| Endometrial cancer | 83 (0.99) | 26 (1.2) | .32 | 15 (0.72) | 12 (0.95) | .48 | NAa | NAa | - | NAa | NAa | - |
| Prostate cancer | 789 (9.4) | 169 (8) | .05 | 222 (10.7) | 109 (8.7) | .05 | 23 (4.9) | 16 (7.3) | .2 | 51 (4.3) | 25 (6.3) | .12 |
| Lung cancer | 170 (2) | 36 (1.7) | .35 | 26 (1.3) | 20 (1.6) | .43 | NAa | NAa | - | NAa | NAa | - |
| Breast cancer | 305 (3.6) | 114 (5.4) | <.001 | 80 (3.9) | 86 (6.8) | <.001 | NAa | NAa | - | NAa | NAa | - |
| Colorectal cancer | 410 (4.9) | 139 (6.6) | .002 | 83 (4) | 65 (5.2) | .12 | 16 (3.4) | 13 (6) | .13 | 31 (2.6) | 16 (4) | .16 |
| Alzheimer’s Disease | 332 (4) | 117 (5.5) | .001 | 117 (5.6) | 116 (9.2) | <.001 | 23 (4.9) | 13 (6) | .57 | 66 (5.6) | 41 (10.3) |
.001 |
| Dementia | 1,234 (14.7) |
448 (21.2) | <.001 | 396 (19.1) | 366 (29) | <.001 | 79 (16.9) |
51 (23.4) |
.04 | 208 (17.7) |
110 (27.7) |
<.001 |
| Cataract | 5,971 (71.1) |
1,513 (71.7) |
.59 | 1,193 (57.6) |
759 (60) | .14 | 297 (63.6) |
154 (70.6) |
.07 | 725 (61.8) |
284 (71.5) |
<.001 |
| Glaucoma | 1,817 (21.6) |
477 (22.6) | .34 | 646 (31.2) | 426 (33.8) |
.12 | 131 (28.1) |
67 (30.7) |
.47 | 305 (26) | 140 (35.3) |
<.001 |
| Hip/Pelvic Fracture | 432 (5.1) | 180 (8.5) | <.001 | 54 (2.6) | 71 (5.6) | <.001 | 14 (3) | 14 (6.4) | .04 | 46 (3.9) | 21 (5.3) | .24 |
| Depression | 2,495 (29.7) |
762 (36.1) | <.001 | 459 (22) | 352 (27.9) |
<.001 | 86 (18.4) |
55 (25.2) |
.04 | 333 (28.4) |
153 (38.5) |
<.001 |
| Osteoporosis | 1,199 (14.3) |
393 (18.6) | <.001 | 156 (7.5) | 129 (10.2) |
.007 | 84 (18) | 64 (29.4) |
.001 | 187 (15.9) |
95 (23.9) |
<.001 |
| Rheumatoid Arthritis/Osteoarthritis |
4,452 (53) | 1,191 (56.4) |
.005 | 1,031 (49.8) |
687 (54.5) |
.008 | 216 (46.3) |
115 (52.8) |
.11 | 537 (45.8) |
210 (52.9) |
.01 |
| Asthma | 1,210 (14.4) |
333 (15.8) | .11 | 300 (14.5) | 225 (17.9) |
.01 | 79 (16.9) |
34 (15.6) |
.67 | 145 (12.4) |
68 (17.1) |
.02 |
| Acquired hypothyroidism | 2,164 (25.8) |
665 (31.5) | <.001 | 296 (14.3) | 239 (19) | <.001 | 72 (15.4) |
42 (19.3) |
.21 | 250 (21.3) |
112 (28.2) |
.005 |
| Benign prostatic hyperplasia |
2,521 (30) | 489 (23.2) | <.001 | 417 (20.1) | 187 (14.8) |
<.001 | 111 (23.8) |
49 (22.5) |
.71 | 271 (23.1) |
67 (16.9) |
.009 |
| Patient Zip Code Community Contextual Factors | ||||||||||||
| Urban | 6,033 (71.8) |
1,568 (74.3) |
.004 | 1,718 (83) | 1,057 (83.9) |
.39 | 446 (95.5) |
211 (96.6) |
.32 | 1,061 (90.5) |
364 (91.7) |
.32 |
| Single | 3,425 (40.8) |
871 (41.3) | .007 | 1,051 (50.8) |
652 (51.7) |
.01 | 210 (45.3) |
100 (45.9) |
.16 | 522 (44.5) |
179 (45) | .29 |
| Less than high school education |
1,516 (18.1) |
389 (18.4) | .11 | 541 (26.1) | 331 (26.2) |
.76 | 100 (21.5) |
52 (23.9) |
.04 | 390 (33.2) |
150 (37.7) |
<.001 |
| Employed | 5,017 (59.8) |
1,257 (59.6) |
.42 | 1,133 (54.7) |
689 (54.6) |
.85 | 277 (59.4) |
126 (57.8) |
.03 | 637 (54.3) |
206 (52) |
<.001 |
| Poor | 872 (10.4) | 223 (10.6) | .24 | 387 (18.7) | 244 (19.4) |
.07 | 61 (13.1) |
30 (13.7) |
.43 | 225 (19.2) |
87 (21.8) |
<.001 |
| Sensory disability among elderly |
1,190 (14.2) |
300 (14.2) | .61 | 303 (14.6) | 184 (14.7) |
.92 | 66 (14.1) |
31 (14.3) |
.45 | 183 (15.6) |
62 (15.6) |
.93 |
| Non-institutionalized elderly with physical disability |
2,370 (28.2) |
605 (28.6) | .03 | 684 (33) | 418 (33.2) |
.44 | 133 (28.5) |
60 (27.7) |
.09 | 371 (31.7) |
127 (32.1) |
.25 |
| Mental Disability | 864 (10.3) | 221 (10.5) | .10 | 274 (13.3) | 170 (13.5) |
.12 | 59 (12.6) |
29 (13.3) |
.07 | 157 (13.4) |
56 (14.2) |
.004 |
| Self-Care Disability | 770 (9.2) | 198 (9.4) | .03 | 256 (12.4) | 156 (12.4) |
.68 | 48 (10.2) |
22 (10.2) | .93 | 135 (11.5) |
49 (12.4) |
<.001 |
| Difficulty going outside the home disability |
1,646 (19.6) |
422 (20) | .004 | 509 (24.6) | 313 (24.9) |
.16 | 105 (22.5) |
49 (22.5) |
.94 | 282 (24) | 102 (25.6) |
<.001 |
| Social security income (in thousands of dollars) |
11.6 | 11.6 | .72 | 10.3 | 10.2 | .24 | 11.1 | 11.0 | .14 | 10.3 | 10.0 | <.001 |
| Elderly in an institution | 387 (4.6) | 98 (4.6) | .81 | 89 (4.3) | 59 (4.7) | .03 | 18 (3.8) | 8 (3.8) | .98 | 42 (3.6) | 13 (3.2) | .08 |
| Median Household income (in thousands of dollars) |
45.0 | 45.4 | .40 | 36.3 | 36.0 | .42 | 49.8 | 49.6 | .94 | 38.2 | 36.1 | .01 |
These values are not reported as either the group with the co-morbidity or the group without the co-morbidity represents less than 12 individuals and Medicare restricts the reporting of data such instance
Unadjusted, fistulas accounted for 79.9% of the vascular access procedures performed in Whites. A smaller proportion of procedures performed in Blacks, Asians and Hispanics were fistulas (Table II). The intermediate models did not result in major changes in absolute or relative proportions of fistulas except in Hispanics. (Appendix Table 3) Upon adjusting for the variables in model 2, the difference in proportion of fistulas performed in Hispanics versus Whites became negligible. After further adjusting for hospital referral region (model 3), 78.0% of the index vascular access procedures performed in Whites were fistulas. Compared to Whites, a smaller proportion of the index procedures performed in Blacks and in Asians were fistulas (Table II). There was no difference between the proportion of index fistulas performed in Whites versus Hispanics.
Table 2.
Fistula prevalence (Unadjusted and Fully Adjusted)
| Unadjusted | Fully Adjusted (Model 3)a | |||
|---|---|---|---|---|
| Fistula Prevalence |
P-value compared to White |
Fistula Prevalenceb |
P-value compared to White |
|
| White | 79.9% | ref | 78.0% | ref |
| Black | 62.4% | <.001 | 65.9% | <.001 |
| Asian/PI | 67.7% | <.001 | 71.4% | <.001 |
| Hispanic | 74.3% | <.001 | 78.7% | .59 |
Adjusted for age and its square, sex, index year, index month, total covered charges from outpatient, inpatient and carrier files in the year preceding index fistula/graft creation date, inpatient setting for the index fistula/graft creation, all co-morbidities available from the Chronic Condition Data Warehouse (except chronic kidney disease), community contextual factors and hospital referral region.
Predictive margin for prevalence of fistula created vs graft.
Overall, 4,707 (28.7%) patients expired during the 12 months post index vascular access procedure. The unadjusted 12-month mortality rates were: White 31.3%, Black 25.0%, Asian 23.3% and Hispanic 23.3%. The unadjusted 12-month mortality rate was higher in grafts compared to fistulas for Whites and Blacks, with no difference for Asians or Hispanics (Table III). Upon adjusting for demographics and co-morbidities, the difference in mortality between graft and fistula became no longer significant for Blacks. This pattern persisted with further adjustment for community contextual factors and hospital referral region. In the fully adjusted model, the mortality in grafts was higher than fistulas for Whites (32.4% versus 30.0%, P=.03) (Table III). In Blacks, Asians and Hispanics, the mortality in the there was no difference between the mortality in the grafts and the fistulas. The difference in mortality between grafts and fistulas within Blacks, Asians and Hispanics was not significantly larger or smaller than the difference between procedures for Whites in any of the models (Table III).
Table 3.
Mortality (Unadjusted and Fully Adjusted)
| Unadjusted | Fully Adjusted (Model 3) | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| 12-Month Mortality | |||||||||
| Access Type |
Rate | P-value for fistula vs graft within race/ ethnicity |
Difference of graft- fistula within race/ ethnicity |
P-value for difference compared to White1 |
Rate2 | P-value for fistula vs graft within race/ ethnicity |
Difference of graft- fistula within race/ ethnicity |
P-value for difference compared to White1 |
|
| White | Fistula | 29.7% | <.001 | 7.6% | ref | 30.0% | .03 | 2.4% | ref |
| Graft | 37.3% | 32.4% | |||||||
| Black | Fistula | 22.8% | <.001 | 5.8% | 0.70 | 25.8% | .28 | 1.7% | .77 |
| Graft | 28.6% | 27.5% | |||||||
| Asian/PI | Fistula | 21.3% | .09 | 5.8% | 0.89 | 22.7% | .73 | 1.2% | .79 |
| Graft | 27.1% | 23.9% | |||||||
| Hispanic | Fistula | 22.4% | .14 | 3.5% | 0.26 | 23.7% | .77 | −0.6% | .28 |
| Graft | 25.9% | 23.1% | |||||||
P-value for difference in incidence of mortality for fistula vs graft within race/ethnicity group compared to difference in incidence of mortality for fistula vs graft within Whites
Predictive margin adjusted for age and its square, sex, index year, index month, total covered charges from outpatient, inpatient and carrier files in the year preceding index fistula/graft creation date, inpatient setting for the index fistula/graft creation and all co-morbidities available from the Chronic Condition Data Warehouse (except chronic kidney disease), community contextual factors and hospital referral region.
In all race/ethnicity groups, the unadjusted incidence of early failure (repeat fistula or graft, or tunneled catheter placement) was lower in patients who underwent graft than fistula (Table IV). The intermediate models did not result in major changes in the absolute or relative incidence of early failure (Appendix Table 4). After adjustment for all covariates (model 3), the incidence of early failure was lower after graft than fistula in all race/ethnicity categories (Table IV). In Whites, the incidence of early failure was 34.9% for graft versus 43.5% for fistula (P<.001), resulting in an incidence of early failure in Whites that was 8.6% lower after graft versus fistula (Table IV). In Blacks, grafts had a 16.2% lower incidence of early failure than fistulas, significantly larger than the 8.6% difference in Whites (P<.001) (Table IV). In Asians and Hispanics, the difference in early failure of graft versus fistula was not significantly larger or smaller compared to Whites.
Table 4.
Early Failure and 12-month Failure Free Survival (Unadjusted and Fully Adjusted)
| Unadjusted | Fully Adjusted (Model 3) | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Early Failure | |||||||||
| Access Type |
Rate | P-value for fistula vs graft within race/ ethnicity |
Difference of fistula- graft within race/ ethnicity |
P-value for difference compared to White1 |
Rate2 | P-value for fistula vs graft within race/ ethnicity |
Difference of fistula- graft within race/ ethnicity |
P-value for difference compared to White3 |
|
| White | Fistula | 41.7% | <.001 | 6.6% | ref | 43.5% | <.001 | 8.6% | ref |
| Graft | 35.1% | 34.9% | |||||||
| Black | Fistula | 51.3% | <.001 | 16% | <.001 | 49.1% | <.001 | 16.2% | <.001 |
| Graft | 35.3% | 32.9% | |||||||
| Asian/PI | Fistula | 41.3% | .02 | 9% | .55 | 40.5% | .014 | 9.7% | 0.74 |
| Graft | 32.3% | 30.8% | |||||||
| Hispanic | Fistula | 44.9% | .01 | 6.9% | .98 | 43.2% | .005 | 8.0% | 0.82 |
| Graft | 38.0% | 35.2% | |||||||
| 12-month Failure-Free Survival | |||||||||
| Access Type |
Survival | P-value for fistula vs graft within race/ ethnicity |
Difference of graft- fistula within race/ ethnicity |
P-value for difference compared to White3 |
12-month Failure- Free Survival2 |
P-value for fistula vs graft within race/ ethnicity |
Difference of graft- fistula within race/ ethnicity |
P-value for difference compared to White3 |
|
| White | Fistula | 40.3% | .33 | −1.1% | ref | 38.9% | .008 | 3.0% | ref |
| Graft | 39.2% | 41.9% | |||||||
| Black | Fistula | 37.4% | <.001 | 9% | <.001 | 37.3% | <.001 | 11.2% | <.001 |
| Graft | 46.4% | 48.5% | |||||||
| Asian/PI | Fistula | 45.5% | .55 | 2.3% | .41 | 45.2% | .98 | 6.4% | .38 |
| Graft | 47.8% | 51.6% | |||||||
| Hispanic | Fistula | 41.2% | .05 | 5.4% | .03 | 42.2% | <.001 | 9.7% | .024 |
| Graft | 46.6% | 51.9% | |||||||
P-value for difference in incidence of early failure for fistula vs graft within race/ethnicity group compared to difference in incidence of early failure for fistula vs graft within Whites
Predictive margin adjusted for age and its square, sex, index year, index month, total covered charges from outpatient, inpatient and carrier files in the year preceding index fistula/graft creation date, inpatient setting for the index fistula/graft creation and all co-morbidities available from the Chronic Condition Data Warehouse (except chronic kidney disease), community contextual factors and hospital referral region.
P-value for difference 12 month failure free survival for fistula vs graft within race/ethnicity group compared to difference 12 month failure free survival for fistula vs graft within White
After applying the Bonferroni correction, this value becomes no longer statistically significant.
In Whites, Blacks and Hispanics, the unadjusted 12-month failure-free survival rate was significantly higher for grafts than for fistulas (Table IV). The intermediate models did not result in major changes in the absolute or relative 12-month failure-free survival (Appendix Table 4). In Whites, Blacks and Hispanics, the fully adjusted 12-month failure-free survival was higher after graft versus fistula (Table IV). In Asians, there was no difference in 12-month failure-free survival for graft versus fistula. In Whites, the 12-month failure-free survival was 41.9% for graft compared to 38.9% for fistula (P=.008) (Table IV). Restated, the 12-month failure-free survival in Whites was 3.0% higher for grafts compared to fistulas. In Blacks, the 12-month failure-free survival was 11.2% higher for grafts compared to fistulas, significantly larger than the 3.0% difference seen in Whites (P<.001) (Table IV). In Hispanics, the 12-month failure-free survival was 9.7% higher for grafts compared to fistulas, which is also larger than the difference seen in Whites (P=.02) (Table IV). However, upon applying the Bonferroni correction, this P-value becomes no longer statistically significant. In Asians, the difference in 12-month failure-free survival of graft versus fistula was not significantly larger or smaller compared to Whites.
Discussion
It is well-documented that the prevalence of ESRD in the United States varies widely by race/ethnicity.1 While the modality of ESRD treatment varies by race/ethnicity, hemodialysis is still the most prevalent mode of renal replacement therapy utilized by all race/ethnicity groups.1 It is also well-documented that the prevalence of hemodialysis vascular access type varies widely by race/ethnicity.1 In the face of such disparities in care, a well-known Institute of Medicine report “Confronting Racial and Ethnic Disparities in Health Care” called for further research to better understand how race and ethnicity are associated with disparities in outcomes from care.16 Despite this, little work has been done to examine the outcomes of hemodialysis vascular access by race/ethnicity.
Among prevalent hemodialysis patients in 2013, 64.3% of Whites were using a fistula for vascular access, compared to 56.9% of Blacks1 a difference that is reflected in our study where there was a 12% higher fully-adjusted incidence of fistula creation in Whites compared to Blacks. This could suggest that there is something inherently related to being Black that makes a patient less likely to undergo fistula creation compared to a White patient. There is some evidence to indicate that Blacks have smaller diameter upper extremity veins than non-Black patients which may contribute to Blacks having a lower prevalence of fistula creation, a trend that has been reported by numerous studies.17,18,19 However, the reported data regarding racial/ethnic variations in upper extremity vein diameter is very limited.
Our study also found that Asians had a significantly lower incidence of fistula creation compared to Whites. However, USRDS data demonstrates a slightly higher prevalence of Asians compared to Whites using a fistula for vascular access (67.2% versus 64.3%), although USRDS does not report whether this difference is significant.1 It is possible that this difference in fistula use and fistula creation is attributed to a higher incidence of fistula creation in younger Asians, a group not included in our study. In addition, we found that there was no difference in the fully adjusted incidence of fistula created between Hispanics and Whites. Interestingly, the unadjusted model demonstrated a significantly lower incidence of fistula creation in Hispanics versus Whites. However, once community contextual factors were adjusted for, this difference was reduced to 1% and became non-significant, suggesting that the unadjusted difference is related to community characteristics of where the Hispanic patients live, and not to being Hispanic itself.
The overall mortality in our study is consistent with the USRDS reported all-cause mortality of 31% per year in the elderly ESRD population.1 The unadjusted mortality rate in Whites was higher than all other race/ethnicity groups. This has been demonstrated in numerous previous reports documenting increased mortality risk among White ESRD patients, which cannot be explained by differences in co-morbidities or transplantation rates, or vascular access type.18,20,21 Within Whites, after full adjustment, grafts were associated with a 2.4% higher mortality than fistulas. Although this difference in mortality is statistically significant, its magnitude is so small that may not be considered clinically significant. Similarly, other authors have demonstrated in a decision and cost-utility analysis that fistula use was associated with a minimal improvement in survival over grafts of only 2.6 months.22 Whites were the only race/ethnicity group in our study that experienced a lower mortality with fistula compared to graft, further calling into question the purported mortality benefit of fistula over graft.
Within all race/ethnicity groups, the incidence of early failure was significantly lower after index graft versus fistula creation, consistent with the growing body of literature indicating that fistula maturation failure is high in the elderly.4,5 The fully adjusted difference between graft and fistula for Blacks was nearly twice that of Whites and highly statistically significant. In the unadjusted model, the difference was even larger, becoming smaller with each model, suggesting that variations in demographics, co-morbidities, community contextual factors and hospital referral region account for a portion of this difference in early failure between Blacks and Whites. The remaining difference is unexplained by the factors included in model 3, suggesting that another factor associated with being Black causes the significantly larger difference between graft and fistula early failure compared to Whites.
The rate of early fistula thrombosis and failure to mature in Blacks has been demonstrated to be as high as 45%, consistent with our findings.23 Additionally, Black race has been shown to be an independent risk factor for hospitalization due to vascular access-related morbidity.24 Despite Blacks being the highest utilizers of hemodialysis and having nearly twice the prevalence of ESRD compared to any other racial/ethnic group, the underlying causes of these poor vascular access outcomes have not been well studied. A study published in 2002 demonstrated that Lipoprotein (a), a low density lipoprotein associated with increased risk of atherothrombosis, is present in higher in Blacks compared to Whites.25 Blacks with the highest lipoprotein (a) levels had a higher incidence of requiring vascular access interventions; there was no association among Whites. Further studies do not appear to have been performed.
The optimal outcome for any ESRD patient, after vascular access creation, is to survive without requiring another vascular access creation or tunneled catheter placement, measured in our study by failure-free survival at 12 months. Overall, the 12-month failure-free survival was sobering for all groups with a fully-adjusted high of only 51.9% in Hispanics who underwent grafts. Given the poor life expectancy of elderly ESRD patients, it is even more important that quality of life issues associated with failed vascular access procedures be taken into consideration when determining the optimal access type for each individual patient.
Among Whites, there was no significant difference in unadjusted failure-free survival between grafts and fistulas, with fistulas even having a slight 1.1% absolute advantage over grafts. Upon adjusting for demographics and co-morbidities, grafts conferred a failure-free survival advantage over fistula, suggesting that the unadjusted equipoise between fistula and graft failure-free survival for Whites is attributed to variations in demographics and co-morbidities. The failure-free survival advantage of grafts increased slightly with each successive model, to only a high of 3%.
Within Hispanics, the fully adjusted failure-free survival was greater after grafts versus fistulas with a difference of nearly 10%, a larger difference compared to Whites which may or may not be statistically significant. Among Asians, the fully adjusted failure-free survival after graft was over 6% higher than fistula; however, this did not achieve statistical significance, possibly due to small sample size. The difference in failure-free survival between fistula and graft was greatest among Blacks. The fully adjusted advantage of grafts over fistulas Blacks was nearly four times that seen in Whites. Again, this suggests that there is a significant variation in outcomes of vascular access in the elderly among race/ethnicity groups and that particularly in Blacks and Hispanics, grafts may confer improved outcomes compared to fistulas.
This study has a number of limitations, the most significant being that it is an observational study. Although we controlled for all readily available confounders measured in claims data, it is possible that worse outcomes in certain minority populations compared to Whites can be attributed to factors other than race/ethnicity. In fact, we believe that future studies should explore these variations in outcomes by race/ethnicity to determine with greater granularity the contributing factors and whether the factors are modifiable. The RTI algorithm utilized by Medicare claims does not recognize Hispanic as an ethnicity and does not allow for the fact that it is possible to be Hispanic as well as White or Black or any other race. While we acknowledge this important distinction, we are limited by the categories that Medicare claims records. Other access-related complications, such as infection and revision, were not included in this study due to the fact that the CPT codes for these complications are not as clear as codes for fistula/graft creation. As a result, the total incidence of encounters related to vascular access is likely to be significantly underestimated. Additionally, we did not measure the number of catheter dependent days, which may vary significantly by fistula versus graft and would be an important outcome to examine in future studies.
Conclusions
Outcomes of grafts versus fistulas in the elderly vary significantly by race/ethnicity and are not accounted for by the demographics, co-morbidities, community contextual factors or hospital referral region studied. The decreased risk of early failure after graft versus fistula creation is larger in Blacks compared to Whites. The higher failure-free survival at 12 months after graft versus fistula creation is larger in Blacks compared to Whites and trends towards being larger in Hispanics compared to Whites. It is imperative that vascular access guidelines be updated to take into consideration individual patient characteristics, including age, race/ethnicity and life expectancy and that future research focus on identifying potentially modifiable risk factors for these variations in outcomes.
Acknowledgement
This research was supported by the National Institute of Health (USC—RCMAR Grant P30AG043073).
Appendix Table I.
Current Procedural Terminology (CPT) codes and procedure descriptions
| CPT code |
Procedure description |
|---|---|
| Fistula | |
| 36818 | arteriovenous anastomosis, open; by upper arm cephalic vein transposition |
| 36819 | arteriovenous anastomosis, open; by upper arm basilic vein transposition |
| 36820 | arteriovenous anastomosis, open; by forearm vein transposition |
| 36821 | arteriovenous anastomosis, opendirect, any site |
| 36825 | creation of arteriovenous fistula by other than direct arteriovenous anastomosis; autogenous graft |
| Graft | |
| 36830 | creation of arteriovenous fistula by other than direct arteriovenous anastomosis; non autogenous graft |
| Tunneled Catheter | |
| 36565 | Insertion of tunneled centrally inserted central venous access device, requiring 2 catheters via 2 separate venous access sites; without subcutaneous port or pump (eg, Tesio type catheter) |
| 36558 | Insertion of tunneled centrally inserted central venous catheter, without subcutaneous port or pump; age 5 years or older |
Appendix Table II.
Distribution of access procedures by year and month.
| White | Black | Asian | Hispanic | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Index fistula (%) (n=8,397) |
Index graft (%) (n=2,110) |
P-value | Index fistula (%) (n=2,071) |
Index graft (%) (n=1,260) |
P- value |
Index fistula (%) (n=467) |
Index graft (%) (n=218) |
P- value |
Index fistula (%) (n=1,173) |
Index graft (%) (n=397) |
P- value |
|
| Year that index fistula/graft was performed 2007 2008 2009 2010 |
2,130 (25.4) 2,035 (24.2) 2,092 (24.9) 2,140 (25.5) |
651 (30.9) 523 (24.8) 501 (23.7) 435 (20.6) |
<0.001 |
548 (26.5) 504 (24.3) 470 (22.7) 549 (26.5) |
382 (30.3) 283 (22.5) 323 (25.6) 272 (21.6) |
0.001 |
109 (23.3) 104 (22.3) 133 (28.5) 121 (25.9) |
76 (34.9) 49 (22.5) 52 (23.9) 41 (18.8) |
0.009 |
280 (23.9) 279 (23.8) 304 (25.9) 310 (26.4) |
107 (26.9) 102 (25.7) 102 (25.7) 86 (21.6) |
0.23 |
| Month that index fistula/graft was performed January February March April May June July August September October November December |
752 (9) 739 (9) 737 (8.8) 696 (8.3) 689 (8.2) 763 (9.1) 711 (8.5) 715 (8.5) 649 (7.7) 718 (8.6) 639 (7.6) 589 (7) |
211 (10) 194 (9.2) 193 (9.1) 197 (9.3) 182 (8.6) 162 (7.7) 166 (7.9) 173 (8.2) 174 (8.2) 179 (8.5) 138 (6.5) 141 (6.7) |
0.28 |
207 (10) 169(8.2) 187 (9) 186 (9) 176 (8.5) 170(8.2) 124 (6) 189 (9.1) 173 (8.4) 173 (8.4) 172 (8.3) 145 (7) |
138 (11) 123 (9.8) 126 (10) 119 (9.4) 97 (7.7) 96 (7.6) 104 (8.3) 118 (9.4) 96 (7.6) 96 (7.6) 79 (6.3) 68 (5.4) |
0.06 |
41 (8.8) 36 (7.7) 56 (12) 46 (9.9) 37 (7.9) 34 (7.3) 31 (6.6) 36 (7.7) 41 (8.8) 44 (9.4) 33 (7.1) 32 (6.9) |
22 (10.1) 21 (9.6) 18 (8.3) 18 (8.3) 13 (6) 25 (11.5) 21 (9.6) 17 (7.8) 14 (6.4) 17 (7.8) 15 (6.9) 17 (7.8) |
0.49 |
105 (9) 97 (8.3) 123 (10.5) 105 (9) 89 (7.6) 106 (9) 104 (8.9) 91 (7.8) 88 (7.5) 104 (8.9) 80 (6.8) 81 (6.9) |
41 (10.3) 41 (10.3) 36 (9.1) 45 (11.3) 31 (7.8) 37 (9.3) 35 (8.8) 26 (6.5) 28 (7.1) 30 (7.6) 22 (5.5) 25 (6.3) |
0.83 |
Appendix Table III.
AVF prevalence (Intermediate Models)
| Model 1a | Model 2b | |||
|---|---|---|---|---|
| AVF Prevalencec |
P-value compared to White |
AVF Prevalencec |
P-value compared to White |
|
| white | 79.5% | ref | 78.7% | ref |
| black | 63.5% | <0.001 | 64.1% | <0.001 |
| asian/PI | 68.6% | <0.001 | 70.5% | <0.001 |
| hispanic | 74.8% | <0.001 | 78.4% | 0.80 |
Adjusted for age and its square, sex, index year, index month, total covered charges from outpatient, inpatient and carrier files in the year preceding index fistula/graft creation date, inpatient setting for the index fistula/graft creation and all co-morbidities available from the Chronic Condition Data Warehouse (except chronic kidney disease)
Adjusted for age and its square, sex, index year, index month, total covered charges from outpatient, inpatient and carrier files in the year preceding index fistula/graft creation date, inpatient setting for the index fistula/graft creation, all co-morbidities available from the Chronic Condition Data Warehouse (except chronic kidney disease) and community contextual factors
Predictive margin for prevalence of fistula created vs graft.
Appendix Table IV.
Mortality, Early Failure and 12-month Failure Free Survival (Intermediate Models)
| Model 1 | Model 2 | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Mortality | |||||||||
| Access Type |
Rate1 | P-value for fistula vs graft within race/ ethnicity |
Difference of graft- fistula within race/ ethnicity |
P-value for difference compared to White2 |
Rate3 | P-value for fistula vs graft within race/ ethnicity |
Difference of graft- fistula within race/ ethnicity |
P-value for difference compared to White4 |
|
| White | Fistula | 29.9% | 0.02 | 2.5% | ref | 30.0% | 0.02 | 2.6% | ref |
| Graft | 32.4% | 32.6% | |||||||
| Black | Fistula | 25.7% | 0.21 | 1.9% | 0.85 | 25.7% | 0.28 | 1.6% | 0.71 |
| Graft | 27.6% | 27.3% | |||||||
| Asian/PI | Fistula | 22.9% | 0.79 | 0.8% | 0.70 | 22.8% | 0.57 | 1.9% | 0.93 |
| Graft | 23.7% | 24.7% | |||||||
| Hispanic | Fistula | 24.8% | 0.76 | −0.7% | 0.25 | 24.1% | 0.64 | −1.1% | 0.19 |
| Graft | 24.1% | 23.0% | |||||||
| Early Failure | |||||||||
| Access Type |
Rate1 | P-value for fistula vs graft within race/ ethnicity |
Difference of fistula- graft within race/ ethnicity |
P-value for difference compared to White4 |
Rate3 | P-value for fistula vs graft within race/ ethnicity |
Difference of fistula- graft within race/ ethnicity |
P-value for difference compared to White4 |
|
| White | Fistula | 42.4% | <0.01 | 7.6% | ref | 43.1% | <0.01 | 8% | ref |
| Graft | 34.8% | 35.1% | |||||||
| Black | Fistula | 50.5% | <0.01 | 16.7% | <0.01 | 49.8% | <0.01 | 16.7% | <0.01 |
| Graft | 33.8% | 33.1% | |||||||
| Asian/PI | Fistula | 41.4% | 0.021 | 9.1% | 0.68 | 40.9% | 0.01 | 9.8% | 0.62 |
| Graft | 32.3% | 31.1% | |||||||
| Hispanic | Fistula | 44.6% | <0.01 | 8.3% | 0.85 | 43.1% | <0.01 | 7.8% | 0.95 |
| Graft | 36.3% | 35.3% | |||||||
| 12-month Failure Free Survival | |||||||||
| Access Type |
Survival1 | P-value for fistula vs graft within race/ ethnicity |
Difference of graft- fistula within race/ ethnicity |
P-value for difference compared to White5 |
Survival3 | P-value for fistula vs graft within race/ ethnicity |
Difference of graft- fistula within race/ ethnicity |
P-value for difference compared to White5 |
|
| White | Fistula | 39.7% | 0.03 | 2.4% | ref | 39.1% | <0.01 | 2.5% | ref |
| Graft | 42.1% | 41.6% | |||||||
| Black | Fistula | 36.4% | <0.01 | 11.6% | <0.01 | 37.0% | <0.01 | 11.5% | <0.01 |
| Graft | 48.0% | 48.5% | |||||||
| Asian/PI | Fistula | 44.4% | 0.16 | 5.3% | 0.45 | 44.9% | 0.014 | 5.7% | 0.44 |
| Graft | 49.7% | 50.6% | |||||||
| Hispanic | Fistula | 40.1% | <0.01 | 8.8% | 0.02 | 42.3% | <0.01 | 9.3% | 0.02 |
| Graft | 48.9% | 51.6% | |||||||
Predictive margin adjusted for age and its square, sex, index year, index month, total covered charges from outpatient, inpatient and carrier files in the year preceding index fistula/graft creation date, inpatient setting for the index fistula/graft creation and all co-morbidities available from the Chronic Condition Data Warehouse (except chronic kidney disease)
P-value for difference in incidence of mortality for fistula vs graft within race/ethnicity group compared to difference in incidence of mortality for fistula vs graft within Whites
Predictive margin adjusted for age and its square, sex, index year, index month, total covered charges from outpatient, inpatient and carrier files in the year preceding index fistula/graft creation date, inpatient setting for the index fistula/graft creation, all co-morbidities available from the Chronic Condition Data Warehouse (except chronic kidney disease) and community contextual factors
P-value for difference in incidence of early failure for fistula vs graft within race/ethnicity group compared to difference in incidence of early failure for fistula vs graft within Whites
P-value for difference 12 month failure free survival for fistula vs graft within race/ethnicity group compared to difference 12 month failure free survival for fistula vs graft within White
| JVS-D-16-00857 |
|
Variations in Outcomes of
Hemodialysis Vascular Access by Race/Ethnicity in the Elderly |
| Race and ethnicity based variations in the elderly on outcomes after hemodialysis access placement are not well known. |
| Retrospective multi-center study using the United States Medicare Outpatient, Inpatient and Carrier files |
| The 12 month outcomes in 16,402 patients undergoing placement of av fistula and av graft were inferior in black and hispanic patients compared to white patients. |
| This study suggests that further investigation is necessary to understand why blacks and hispanics have inferior outcomes compared to whites after av access placement. |
| 1.Strong |
| B.Medium |
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- 1.United States Renal Data System . National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases. Bethesda, MD: 2015. 2015 USRDS annual data report: Epidemiology of kidney disease in the United States. [Google Scholar]
- 2.Neumann ME. “Fistula first” initiative pushes for new standards in access care. Nephrol News Issues. 2004;18(9):43. [PubMed] [Google Scholar]
- 3.NKF-K/DOQI Clinical practice guidelines for vascular access: update 2006. Am J Kidney Dis. 2006;48(Suppl 1):S176–273. doi: 10.1053/j.ajkd.2006.04.029. [DOI] [PubMed] [Google Scholar]
- 4.Lok CE, Oliver MJ, Su J, Bhola C, Hannigan N, Jassal SV. Arteriovenous fistula outcomes in the era of the elderly dialysis population. Kidney Int. 2005;67(6):2462–2469. doi: 10.1111/j.1523-1755.2005.00355.x. [DOI] [PubMed] [Google Scholar]
- 5.Lazarides MK, Georgiadis GS, Antoniou GA, Staramos DN. A meta-analysis of dialysis access outcome in elderly patients. J Vasc Surg. 2007;45(2):420–426. doi: 10.1016/j.jvs.2006.10.035. [DOI] [PubMed] [Google Scholar]
- 6.Woo K, Goldman DP, Romley JA. Early Failure of Dialysis Access among the Elderly in the Era of Fistula First. Clin J Am Soc Nephrol. 2015;10(10):1791–1798. doi: 10.2215/CJN.09040914. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Chronic Conditions Data Warehouse Chronic Condition Warehouse Categories. https://www.ccwdata.org/web/guest/condition-categories. Accessed 2 April 2016.
- 8.Center for Medicare and Medicaid Services Berenson-Eggers Type of Service (BETOS) http://www.cms.gov/Medicare/Coding/HCPCSReleaseCodeSets/BETOS.html. Accessed 2 April 2016.
- 9.Clement FM, James MT, Chin R, Klarenbach SW, Manns BJ, Quinn RR, et al. Validation of a case definition to define chronic dialysis using outpatient administrative data. BMC Med Res Methodol. 2011;11(1 Mar):25–31. doi: 10.1186/1471-2288-11-25. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Eicheldinger C, Bonito A. More accurate racial and ethnic codes for Medicare administrative data. Health Care Financ Rev. 2008;29(3):27–42. [PMC free article] [PubMed] [Google Scholar]
- 11.Fisher ES, Wennberg DE, Stukel TA, Gottlieb DJ, Lucas FL, Pinder EL. The implications of regional variations in Medicare spending. Part 2: health outcomes and satisfaction with care. Ann Intern Med. 2003;138(4):288–298. doi: 10.7326/0003-4819-138-4-200302180-00007. [DOI] [PubMed] [Google Scholar]
- 12.Romley JA, Jena AB, Goldman DP. Hospital Spending and Inpatient Mortality: Evidence From California: An Observational Study. Ann Intern Med. 2011;154(3):160–167. doi: 10.7326/0003-4819-154-3-201102010-00005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Dartmouth Atlas Hospital referral regions. http://www.dartmouthatlas.org/data/region/ Accessed 2 April 2016.
- 14.Graubard BI, Korn EL. Predictive margins with survey data. Biometrics. 1999;55(2):652–659. doi: 10.1111/j.0006-341x.1999.00652.x. [DOI] [PubMed] [Google Scholar]
- 15.Cox DR. Regression models and life-tables (with discussion) J Roy Statist Soc Ser B. 1972;34(2):187–220. [Google Scholar]
- 16.Unequal Treatment: Confronting Racial and Ethnic Disparities in Health Care . In: Institute of Medicine (US) Committee on Understanding and Eliminating Racial and Ethnic Disparities in Health Care. Smedley BD, Stith AY, Nelson AR, editors. National Academies Press (US); Washington (DC): 2003. [PubMed] [Google Scholar]
- 17.Ishaque B, Zayed MA, Miller J, Nguyen D, Kaji AH, Lee JT, et al. Ethnic differences in arm vein diameter and arteriovenous fistula creation rates in men undergoing hemodialysis access. J Vasc Surg. 2012;56(2):424–431. doi: 10.1016/j.jvs.2012.01.029. [DOI] [PubMed] [Google Scholar]
- 18.Woo K, Yao JF, Selevan DC, Hye RJ. Influence of Vascular Access Type on Gender and Ethnicity related Mortality in Hemodialysis Patients. Perm J. 2012;16(2):4–9. doi: 10.7812/tpp/12-005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Allon M, Ornt DB, Schwab SJ, Rasmussen C, Delmez JA, Greene T, et al. Factors associated with the prevalence of arteriovenous fistulas in hemodialysis patients in the HEMO study. Hemodialysis (HEMO) Study Group. Kidney Int. 2000;58(5):2178–2185. doi: 10.1111/j.1523-1755.2000.00391.x. [DOI] [PubMed] [Google Scholar]
- 20.Bleyer AJ, Tell GS, Evans GW, Ettinger WH, Jr, Burkart JM. Survival of patients undergoing renal replacement therapy in one center with special emphasis on racial differences. Am J Kidney Dis. 1996;28(1):72–81. doi: 10.1016/s0272-6386(96)90133-x. [DOI] [PubMed] [Google Scholar]
- 21.Bloembergen WE, Port FK, Mauger EA, Wolfe RA. Causes of death in dialysis patients: racial and gender differences. J Am Soc Nephrol. 1994;5(5):1231–1242. doi: 10.1681/ASN.V551231. [DOI] [PubMed] [Google Scholar]
- 22.Xue H, Lacson E, Jr, Wang W, Curhan GC, Brunelli SM. Choice of vascular access among incident hemodialysis patients: A decision and cost-utility analysis. Clin J Am Soc Nephrol. 2010;5(12):2289–2296. doi: 10.2215/CJN.03210410. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Obialo CI, Tagoe AT, Martin PC, Asche-Crowe PE. Adequacy and survival of autogenous arteriovenous fistula in African American hemodialysis patients. ASAIO Journal. 2003;49(4):435–439. [PubMed] [Google Scholar]
- 24.Feldman HI, Held PJ, Hutchinson JT, Stoiber E, Hartigan MF, Berlin JA. Hemodialysis vascular access morbidity in the United States. Kidney Int. 1993;43(5):1091–1096. doi: 10.1038/ki.1993.153. [DOI] [PubMed] [Google Scholar]
- 25.Astor BC, Eustace JA, Klag MJ, Powe NF, Longenecker JC, Fink NE, et al. Race-specific association of lipoprotein(a) with vascular access interventions in hemodialysis patients: The CHOICE Study. Kidney Int. 2002;61(3):1115–1123. doi: 10.1046/j.1523-1755.2002.00194.x. [DOI] [PubMed] [Google Scholar]