Predialysis Nephrologist Care and Access to Kidney Transplantation in the United States

Authors

  • W. C. Winkelmayer,

    Corresponding author
    1. Division of Pharmacoepidemiology and Pharmacoeconomics and the Renal Division, Department of Medicine, Brigham and Women's Hospital, Harvard Medical School, Boston, MA, and the Office of the Chancellor, University of Tennessee Health Sciences Center, Memphis, TN, USA
    Search for more papers by this author
  • J. Mehta,

    1. Division of Pharmacoepidemiology and Pharmacoeconomics and the Renal Division, Department of Medicine, Brigham and Women's Hospital, Harvard Medical School, Boston, MA, and the Office of the Chancellor, University of Tennessee Health Sciences Center, Memphis, TN, USA
    Search for more papers by this author
  • A. Chandraker,

    1. Division of Pharmacoepidemiology and Pharmacoeconomics and the Renal Division, Department of Medicine, Brigham and Women's Hospital, Harvard Medical School, Boston, MA, and the Office of the Chancellor, University of Tennessee Health Sciences Center, Memphis, TN, USA
    Search for more papers by this author
  • W. F. Owen Jr,

    1. Division of Pharmacoepidemiology and Pharmacoeconomics and the Renal Division, Department of Medicine, Brigham and Women's Hospital, Harvard Medical School, Boston, MA, and the Office of the Chancellor, University of Tennessee Health Sciences Center, Memphis, TN, USA
    Search for more papers by this author
  • J. Avorn

    1. Division of Pharmacoepidemiology and Pharmacoeconomics and the Renal Division, Department of Medicine, Brigham and Women's Hospital, Harvard Medical School, Boston, MA, and the Office of the Chancellor, University of Tennessee Health Sciences Center, Memphis, TN, USA
    Search for more papers by this author

*Corresponding author: Wolfgang C. Winkelmayer, wwinkelmayer@partners.org

Abstract

Predialysis nephrologist care is associated with morbidity and mortality in incident dialysis patients, but the relationship with access to kidney transplantation (KT) is unclear. From a national study of incident US dialysis patients, we identified 2253 patients with detailed information about predialysis care, sociodemographic characteristics and comorbidities. We used multivariate Cox proportional hazards models to study associations between predialysis nephrology care and two outcomes: time from first dialysis to the first day on the KT wait-list, and time to first KT. Two-thirds of patients first encountered a nephrologist >3 months prior to dialysis and one-third ≤3 months prior to dialysis (early vs. late nephrologist care; ENC vs. LNC). Overall, 515 patients were added to the KT wait-list and 406 underwent KT during follow-up (2.3 years). In multivariate analyses, ENC was associated with a 41% (95%CI: 15–72%) greater rate of being wait-listed compared to LNC and a 54% (95%CI: 22–96%) greater rate of KT. Similar associations existed with number of predialysis nephrology visits.Earlier and more frequent predialysis nephrologist care were associated with greater access to the KT wait-list as well as a higher rate of KT, indicating that LNC may augment existing inequalities that impair access to KT.

Introduction

While the role of nephrologists in the care of patients on renal replacement therapy (RRT) is clearly defined, the importance of nephrologist care of patients with earlier stages of chronic kidney disease has only recently been evaluated. Previous studies have demonstrated that many US patients approaching end-stage renal disease (ESRD) do not consult a nephrologist until the need for RRT is imminent. This proportion appears to vary substantially across the country: while 20% of ESRD patients without a history of acute renal failure at New England Medical Center in Boston were seen by a nephrologist <3 months prior to RRT (1), 57% of patients initiating RRT in New York City had no prior nephrologist care at all (2) and 24% of patients in West Virginia did not see a nephrologist until <1 month prior to RRT (3). In the entire United States, one-third of new dialysis patients had their first nephrologist consultation <4 months prior to initiation of RRT (4).

Timely initiation of nephrologist care for patients with CKD is imperative for providing these patients with adequate care of the multiple metabolic disorders that typically accompany advanced CKD (5). As these patients approach RRT, a nephrologist can also educate them about treatment options, to help the patient choose the most suitable treatment modality and be better prepared for it. Patients who do not consult a nephrologist sufficiently early face suboptimal preparation for and transition into RRT (6,7) which in turn can be associated with considerable morbidity and mortality, at least during the first few months of RRT (8–10).

Another aspect of timely access to nephrologist care prior to RRT has not sufficiently been studied: access to kidney transplantation. Transplantation confers greater survival and quality of life than hemodialysis or peritoneal dialysis (11,12), and from a payor's perspective, it is the most cost-effective treatment for these patients (13). Donor kidneys, however, are scarce and wait-lists for kidney transplantation from deceased donors have soared (14). Thus, delayed nephrologist care may prevent a patient from being adequately informed about the transplantation option, and precious time may elapse until a patient can start the necessary work-up and be added to a transplant wait-list if medically eligible.

Since time on the wait-list is an important criterion for the allocation of organs from deceased donors (15,16), failure to see a nephrologist prior to RRT may thus directly translate into additional months or years on the wait-list. Time on wait-list, in turn, is an important predictor of outcomes after kidney transplantation (17). Pre-emptive kidney transplantation prior to the need for dialysis has also been shown to have a survival benefit. Early referral to a nephrologist allows for work-up of potential living donors and in some parts of the United States allows listing of patients for a zero-mismatched deceased kidney in patients approaching ESRD. Whether delayed nephrologist care is associated with reduced access to kidney transplantation in the United States has not sufficiently been studied. A small case–control study of predominantly elderly patients in New Jersey has suggested that such an association may exist (18). We sought to study this aspect of predialysis nephrologist care using a large and nationally representative sample of incident dialysis patients.

Methods

Study population and follow-up

The Dialysis Morbidity and Mortality Study (DMMS) Wave 2 was a special study conducted by the US Renal Data System (USRDS) (19). During a predefined time window in 1996/97, all patients who began peritoneal dialysis and a random 20% sample of all incident hemodialysis patients were prospectively enrolled for study (total N = 4024). Data on medical conditions was obtained through a questionnaire at each dialysis facility. Data on prior medical history, demographics and medications were obtained from patient records. Baseline data were collected at approximately 60 days after first RRT. Follow-up questionnaires were administered 9–12 months after enrollment, and data collection was completed in 1999. Patient follow-up is available through the USRDS core dataset; all records can be linked to any given patient using a unique identifier assigned by USRDS. For this study, we used year 2000 USRDS files. No personally identifiable data were used in any analyses.

From the overall DMMS Wave 2 sample, we excluded patients who did not have a USRDS-assigned identifier, patients for whom information was not available on the main exposure variable (timing of first nephrologist visit), and patients who were not truly incident as evident from receipt of maintenance dialysis or kidney transplantation prior to the DMMS Wave 2 enrollment period. We also excluded patients who were positive for human immunodeficiency virus (HIV) or had been diagnosed with the Acquired Immunodeficiency Syndrome (AIDS) at initiation of maintenance dialysis.

Main exposures

The main exposure variables were obtained from the patient questionnaire portion of DMMS Wave 2. Two questions were relevant for this study: ‘Prior to starting regular dialysis, when did you first receive medical attention from a kidney specialist (nephrologist)?’ and ‘In the year prior to dialysis, about how many visits did you make to a kidney specialist (nephrologist)?’ The response categories for the former were: >12 months, between 12 and 4 months, between 3 and 2 months, and less than one month. Similar to our previous work, we termed a first nephrologist visit ≤3 months prior to RRT a ‘late referral’ (20); this was the main study variable. A second set of analyses used all four response categories to this question, with patients who saw a nephrologist <1 month prior to RRT as reference group. The third set of analyses categorized patients into their number of pre-dialysis nephrologist visits: 0–1 (reference group), 2–4, 5 or more.

Other patient characteristics

For each patient, we identified several demographic and socioeconomic characteristics: age at first dialysis (continuous), gender (female, male), race (white, black, other), Hispanic ethnicity (yes/no), socioeconomic characteristics (education, marital status, living alone) and presence of several comorbidities (all yes/no). Comorbid conditions were abstracted from patient records and included diabetes, hypertension, coronary heart disease, congestive heart failure, peripheral artery disease or amputation, cerebrovascular disease, chronic obstructive lung disease, history of cardiac arrest and any malignancy. We also ascertained the initial dialysis modality chosen (center hemodialysis, continuous ambulatory peritoneal dialysis, other).

Outcomes

This study considered two different outcomes: a patient's first date of being wait-listed for kidney transplantation and the actual date of the first transplantation. The former was ascertained from the FIRST_SE variable in the PATIENTS file of the USRDS. The latter was ascertained using the TDATE variable in the TX (transplantation) file.

Statistical analyses

All statistical analyses were preformed using SAS version 9.1 (SAS Institute, Cary, NC). We compared the characteristics of patients who were referred late versus early, using paired t-tests and χ2-tests. The two study outcomes were evaluated using time-to-event analyses. The index date was the date of first dialysis, and the outcome date was the first date on the wait-list or the date of receipt of a first kidney transplant, respectively. Patients were censored at death and the end of follow-up. For the former analyses, we excluded patients who were electively added to the wait-list prior to first RRT. We first described the relationship between the exposure variables and the study outcomes using Kaplan–Meier plots. We then used univariate and multivariate Cox proportional hazards models to study the crude and adjusted associations between the variables of interest and these outcomes. Multivariate models included all covariates regardless of any significance threshold. We tested for interactions between the main exposure variables and all other variables and tested for violations of the proportionality assumption using interactions with follow-up time. As measures of association, we estimated hazards ratios (HR) and their 95% confidence intervals.

Results

Cohort selection and characterization

Of the 4024 patients who initiated RRT in 1996/97, 1518 (37.7%) patients did not respond to the question about the timing of their first predialysis nephrologist care. Patients were also excluded if other USRDS data indicated that they had dialysis or transplant before the date of first dialysis listed in the DMMS Wave 2 file (149; 3.7%), patients for whom a USRDS-assigned identifier was not available (83; 2.1%), and patients who were known to have HIV or AIDS (20; 0.5%). The final study cohort contained 2253 incident dialysis patients, or 56% of the original file. Of these, 1018 (45.2%) had received nephrologist care for >1 year prior to RRT, 483 (21.4%) had a first nephrologist visit between 12 and 4 months prior to RRT, whereas 293 (13.0%) first saw a nephrologist between 1 and 3 months and 459 (20.4%) <1 month prior to initiation of chronic dialysis treatment (Table 1). Thus, two-thirds (66.6%), of these patients received timely (early) nephrologist care and one-third (33.4%) had late nephrologist referral. The proportions of patients with 0–1, 2–4 and ≥5 nephrologist visits were 30.1%, 26.2% and 43.7%, respectively (Table 1; this question was unanswered or the patient not sure in 6% of all patients).

Table 1.  Predialysis nephrologist care in the Dialysis Morbidity and Mortality Study Wave 2 (N = 2253)
Timing of first nephrologist visitN%
  1. Patients were categorized by timing of first nephrologist visit as having had an early (>3 months prior to first dialysis) or late (≤3 months prior to first dialysis) first nephrologist visit. Of all 2253 patients, 136 (6%) had no response recorded for number of predialysis nephrologist visits, or they indicated that they could not remember.

>12 months prior to dialysis101845.2
≥4–12 months prior to dialysis48321.4
≥1–3 months prior to dialysis29313.0
<1 month prior to dialysis45920.4
Number of nephrologist
visits prior to dialysisN%
None or 163730.1
2–455526.2
≥592543.7

Late referrals were more likely to be Hispanic, had less education, were less likely to be married, and less likely to have been diagnosed with diabetes and hypertension (Table 2). These patients also had a different pattern of underlying renal disease and they used hemodialysis more frequently as their first treatment modality.

Table 2.  Demographic characteristics and prior comorbidities, by timing of first nephrologist visit
N, % or Mean ± SDFirst nephrologist visit
Early (>3 months; N = 1501)Late (≤3 months; N = 752)p
  1. 1These proportions are not representative of the US population since peritoneal dialysis patients were oversampled by a 5: 1 ratio compared to hemodialysis patients in this study.

Age  58.4±15.0   58.7±16.0  0.73
GenderFemale676(45.0%)356(47.3%)0.30
RaceWhite1005(67.0%)467(62.1%) 
Black377(25.1%)218(29.0%) 
Other119(7.9%) 67 (8.9%) 0.07
EthnicityHispanic156(10.4%)102(13.6%)0.03
Education<12 years of school381(25.4%)244(32.5%) 
High school graduate478(31.9%)234(31.1%) 
Some college260(17.3%)114(15.2%) 
College graduate249(16.6%)84 (11.2%) 
Missing133(8.9%) 76 (10.1%)<0.001  
Marital statusSingle182(12.1%)128(17.0%) 
Married923(61.5%)387(51.5%) 
Divorced/widowed/separated396(26.4%)237(31.5%)<0.001  
Diabetes mellitus 737(49.1%)336(44.7%)0.05
Hypertension 1110(74.0%)508(67.6%)0.002
Coronary artery disease 561(37.4%)281(37.4%)0.99
Congestive heart failure 519(34.6%)283(37.6%)0.15
Peripheral artery disease/amputation 246(16.4%)131(17.4%)0.54
Cerebrovascular disease 123(8.2%) 60 (8.0%) 0.86
Chronic obstructive lung disease 115(7.7%) 52 (6.9%) 0.52
Cancer 119(7.9%) 70 (9.3%) 0.27
Underlying renal disease
 Diabetic nephropathy656(43.7%)300(39.9%) 
Hypertensive nephropathy364(24.3%)235(31.3%) 
Glomerulonephritis133(8.9%) 67 (8.9%)  
Other or unknown348(23.2%)150(20.0%)0.004
Hemodialysis vs. other dialysis1 767(51.1%)491(65.3%)<0.001

Nephrologist care as a determinant of access to the wait-list

For this set of analyses, we excluded 83 patients (3.7%) who had been electively wait-listed prior to their first dialysis. Over a median of 2.3 years of follow-up, 515 patients (23.7%) were added to the wait-list for kidney transplantation. Figure 1 shows the time course of cumulative incidence of this endpoint, which appeared to be greater in patients with longer predialysis nephrologist care. Univariate Cox proportional hazards models confirmed this impression: early referral to a nephrologist was associated with a 35% (HR: 1.35; 95% CI: 1.11–1.64) greater rate of being added to the kidney transplant wait-list compared to patients who were referred late (Table 3). Multivariate adjustment for all other baseline variables confirmed this association. Timely nephrologist care was independently associated with a 41% greater rate of wait-listing compared to delayed nephrologist care (HR: 1.41; 95% CI: 1.15–1.72). Further discrimination regarding the timing of the first nephrologist contact revealed a clear dose response in multivariate analyses: compared to patients who first saw a nephrologist during the month preceding the initiation of chronic RRT, patients who had their first nephrologist visit 1–3, 4–12, or >12 months prior to RRT were 30% (HR: 1.30; 95% CI: 0.93–1.83), 48% (HR: 1.48; 95% CI: 1.10–1.99), and 59% (HR: 1.59; 95% CI: 1.24–2.04) more likely to be added to the wait-list for kidney transplantation, respectively (Table 3). Similarly, the number of predialysis nephrologist visits was also associated with access to the wait-list after multivariate adjustment: compared to patients who saw a nephrologist prior to RRT only once or not at all, having had 2–4, or ≥5 such visits was associated with a 37% (HR: 1.37; 95% CI: 1.07–1.77) and 51% (HR: 1.51; 95% CI: 1.21–1.88) greater rate of being wait-listed, respectively. We found no evidence for effect modification among the model variables, especially for a priori variables such as age, race, gender, and baseline dialysis modality, and the proportionality assumption was not violated.

Figure 1.

Timing of first predialysis nephrologist care and wait-listing for kidney transplantation. Kaplan–Meier survival plot of time from first dialysis to first day on the waiting list for kidney transplantation (in years), stratified by duration of predialysis nephrologist care: <1 month (red), 1–3 months (green), 4–12 months (blue), >12 months (black).

Table 3.  Association between pre-dialysis nephrologist care and wait-listing for kidney transplantation
 UnivariateMultivariate
Relative risk95% confidence intervalp-valueRelative risk95% confidence intervalp-value
First nephrologist visitLate (≤3 months)1.0Referent1.0
Early (>3 months)1.351.11–1.640.0021.411.15–1.72<0.001
First nephrologist visit<1 month1.0Referent1.0Referent
(relative to first RRT)≥1–3 months1.080.78–1.510.631.300.93–1.830.13
≥4–12 months1.130.85–1.510.401.481.10–1.990.009
>12 months1.551.21–1.97<0.001 1.591.24–2.04<0.001
Number of nephrologist visits0–11.0Referent1.0Referent
(prior to first RRT)2–41.140.89–1.460.291.371.07–1.770.01
≥51.371.11–1.690.0041.511.21–1.88<0.001 
Not sure/missing0.460.26–0.820.0080.610.34–1.190.09

Nephrologist care as a determinant of kidney transplantation

Of the 2253 patients in the cohort, 406 (18.0%) received a kidney transplant during follow-up. Kaplan–Meier curves indicated that earlier predialysis nephrologist care was also associated with higher rates of kidney transplantation (Figure 2). Univariate Cox models confirmed the impression from these graphs: patients seen by a nephrologist in a timely fashion (ER) had a 59% greater chance of being transplanted, and these univariate findings were practically unconfounded by other factors in the multivariate model (HR: 1.54; 95% CI: 1.22–1.96; Table 4). Multivariate analyses using more graded information on timing of first predialysis nephrologist care confirmed the importance of timely nephrologist care, but revealed that only patients who first saw a nephrologist >12 months prior to RRT had a greater rate of kidney transplantation (HR: 1.69; 95% CI: 1.27–2.01; Table 4). Number of predialysis nephrologist visits was also independently associated with access to transplantation. Compared to patients who had 0–1 nephrologist visits, patients who saw a nephrologist 2–4 times or ≥5 times had rates of kidney transplantation that were 32% (HR: 1.32; 95% CI: 0.99–1.76) and 42% (HR: 1.42; 95% CI: 1.11–1.82) greater, respectively (Table 4). No effect modification for any other baseline variable or interaction with follow-up time was detected in these models.

Figure 2.

Timing of first predialysis nephrologist care and kidney transplantation. Kaplan–Meier survival plot of time from first dialysis to date of first kidney transplantation (in years), stratified by duration of predialysis nephrologist care: <1 month (red), 1–3 months (green), 4–12 months (blue), >12 months (black).

Table 4.  Association between pre-dialysis nephrologist care and rate of kidney transplantation
 UnivariateMultivariate
Relative risk95% confidence intervalp-valueRelative risk95% confidence intervalp-value
First nephrologist visitLate (≤3 months)1.0 Referent1.0 
Early (>3 months)1.591.25–1.96<0.0011.541.22–1.96<0.001
First nephrologist visit<1 month1.0 Referent1.0 Referent
(relative to first RRT)≥1–3 months0.920.61–1.390.681.070.70–1.630.75
≥4–12 months1.070.76–1.510.691.330.94–1.890.11
>12 months1.751.32–2.31<0.0011.691.27–2.24<0.001
Number of nephrologist visits0–11.0 Referent1.0 Referent
(prior to first RRT)2–41.230.93–1.630.141.320.99–1.760.06
≥51.321.03–1.680.021.421.11–1.820.006
Missing0.680.39–1.200.181.130.64–2.010.67

Discussion

This is the first study to evaluate the association between predialysis nephrologist care and access to transplantation in a representative sample of incident US dialysis patients. We found that both earlier timing and greater intensity of nephrologist care prior to RRT were associated with increased access to the wait-list for kidney transplantation and greater rates of transplantation. The only other study of this issue in US patients found that early rather than late referral (>3 vs. ≤3 months prior to RRT) was associated with a nearly five-fold increased rate of kidney transplantation in elderly patients in New Jersey, but the study sample was very small and as a consequence, the confidence intervals were wide (odds ratio: 4.5; 95% CI: 1.03–20) (18). The present study confirms these hypothesis-generating results in a large and nationally representative sample, and provides a more precise estimate of this association (HR = 1.54; 95% CI: 1.22–1.96). In addition, our study gives further detail on these relationships due to more refined information on timing and additional data on the intensity of predialysis nephrologist care. More research is necessary on the extent to which earlier and more consistent nephrologist intervention is the actual cause of improved access, or is a marker for other patient or system characteristics that produce more favorable outcomes.

Only one other study outside the United States has studied this question. Using data from the joint ESRD registry of Australia and New Zealand, Cass et al. found that early referral was associated with twice the rate of being added to the wait-list (HR = 2.04; 95% CI: 1.69–2.44) and 1.5 times the rate of kidney transplantation (HR = 1.54; 95% CI: 1.30–1.82) (21). While their results are strikingly similar to our findings, it was uncertain whether these results could be generalized to the US situation for several reasons: a different organ allocation system, fundamentally different health care systems in Australia and New Zealand, exclusion of patients aged ≥65 years, and differences in the composition of the population and the distribution of several patient characteristics between patients who received early versus late nephrologist referral (21).

While earlier studies have found associations between earlier nephrologist care and greater quality of care regarding the transition of CKD patients to RRT as well as with short-term outcomes on RRT (greater persistence with the initial dialysis modality, greater patient survival) (1–4,6–10), the present study indicates that there is also an important equity issue at stake. From a public policy perspective, debate is needed on whether patients should be penalized twice for delayed referral to a nephrologist, in that they experience suboptimal preparation for RRT and greater morbidity and mortality on dialysis, and also have reduced access to kidney transplantation. Transplantation is associated with better survival compared to similar patients on dialysis (11), and greater duration of pre-transplantation RRT is also a determinant of greater mortality after transplantation (17). Since delayed nephrologist care is more prevalent in racial minorities and socioeconomically underprivileged patients (20,22), these findings indicate that the current US allocation system may actually exacerbate the socioeconomic and racial disparities that exist in access to transplantation. Other organ procurement systems (e.g. the Eurotransplant Foundation, and the joint organ procurement system in Australia and New Zealand) have recognized this issue and removed this inequality by standardizing the waiting-time used for organ allocation by using the first day of dialysis as the index date for calculation of wait-list-relevant waiting time (23). Such an allocation system could reduce the likelihood of these patients experiencing such a double-hit from inadequate access to care. Indeed, a national conference on the wait-list for kidney transplantation (4–5 March 2002, Philadelphia, PA) recommended in their report that ‘Time waiting for a transplant should be calculated from the point at which a patient begins maintenance dialysis, not upon completion of transplant evaluation’. (24). Several organ procurement regions in the United States have since discussed and experimented with changing the wait-list date to date of first dialysis or based on a creatinine clearance threshold, but to our knowledge, no formal evaluations of these policy changes have been published.

There are certain limitations to this study. Patients were not randomized to receiving a certain timing or intensity of predialysis nephrologist care, but their natural course was recorded and analyzed. Hence, there might be unobserved imbalances in other prognostic parameters that we were unable to capture using this dataset. Thus, residual confounding is a possibility, despite availability of a very rich dataset. We had access to data on the presence of several comorbidities, but not on their severity. Since comorbidity was ascertained from medical records rather than diagnoses associated from health care claims, there is less likelihood that these data were biased by lower

access to care in general. Information on patterns of non-nephrologist care prior to RRT was not available for study. Further, it is possible that patients who received less nephrologist care also had a lower preference for kidney transplantation. Different patient preferences for this procedure have previously been shown for patients of different races, but these differences only explained a small fraction of the substantial racial differences in access to kidney transplantation (25). Our analyses did not discriminate between transplantations from living versus deceased donor organs, but the objective was to specifically evaluate the differences in access to transplantation independent of donor type. A number of patients may have reached ESRD via an acute course of kidney injury and so their relatively late nephrologist care may have been inevitable (26). Information on predialysis nephrologist care was missing in roughly a third of the patients, and non-white race, lower educational level, having initiated renal replacement therapy using peritoneal dialysis, and homelessness, nursing home or institution residence were each independently associated with such information missing. However, both study outcomes did not differ between those patients who provided information on their predialysis nephrologist care compared to those who did not, thus affirming generalizability of the findings from the studied subset of the overall study sample. While more recent data would be desirable to test the study hypotheses in more recent years, such updated data is currently not available. Nevertheless, the strengths of this study are its use of a cohort that is completely representative of all US dialysis patients, its prospective design, and inclusion of several patient characteristics that are usually not available for study with medical claims data.

We conclude from this study that substantial differences in the access to kidney transplantation exist as a function of timing and intensity of predialysis nephrologist care. Inequalities in access to nephrologist care appear to augment existing inequalities in the access to transplantation caused by other socioeconomic and demographic factors. More attention should be paid to ensuring earlier access to nephrologist care for patients with advancing CKD, regardless of their racial, socioeconomic or health insurance status. Since organs from deceased donors should be regarded a common good, changes in organ allocation algorithms such as standardizing time on wait-list to the first day of maintenance dialysis should be considered as a means to provide more equitable access to this life-saving procedure.

Acknowledgments

Dr. Winkelmayer was a 2004–2006 T. Franklin Williams Scholar in Geriatric Nephrology (American Society of Nephrology—Association of Subspecialty Professors Junior Development Award in Geriatric Nephrology, jointly sponsored by the Atlantic Philanthropies, the American Society of Nephrology, the John A. Hartford Foundation, and the Association of Subspecialty Professors).

The data reported here have been supplied by the United States Renal Data System (USRDS). The interpretation and reporting of these data are the responsibility of the authors and in no way should be seen as an official policy or interpretation of the US government. This work was reviewed and approved by the Institutional Review Board of Brigham and Women's Hospital.

Ancillary