Article

Association of Erythropoietin Dose and Route of

Administration with Clinical Outcomes for
Patients on Hemodialysis in the United States
Daniel G. Wright,* Elizabeth C. Wright,† Andrew S. Narva,‡ Constance T. Noguchi,* and Paul W. Eggers‡

Abstract
*Molecular Medicine
Background and objectives Recombinant human erythropoietin (epoetin) is used routinely to increase blood Branch, †Biostatistics
hemoglobin levels in patients with ESRD and anemia. Although lower doses of epoetin are required to achieve Program, and
‡
equivalent hemoglobin responses when administered subcutaneously rather than intravenously, standard Division of Kidney,
practice has been to administer epoetin to patients on hemodialysis intravenously. Randomized trials of alternative Urologic, and
epoetin treatment regimens in patients with kidney failure have shown that risks of cardiovascular complications Hematologic
Diseases, National
and death are related to the dose levels of epoetin used. Therefore, given the dose-sparing advantages of Institute of Diabetes
subcutaneous epoetin administration, the possibility that treatment of patients on hemodialysis with sub- and Digestive and
cutaneous epoetin might be associated with more favorable outcomes compared with intravenous treatment was Kidney Diseases,
investigated. National Institutes of
Health, Bethesda,
Maryland
Design, setting, participants, & measurements A retrospective cohort study of 62,710 adult patients on
hemodialysis treated with either intravenous or subcutaneous epoetin-a and enrolled in the Centers for Medicare and Correspondence:
Medicaid Services ESRD Clinical Performance Measures Project from 1997 to 2005 was carried out. Risks of death Dr. Daniel G. Wright,
and/or hospitalization for cardiovascular complications (adverse composite event outcomes) during 2 years of Molecular Medicine
follow-up were determined in relationship to epoetin dose and route of administration (intravenous versus sub- Branch, National
Institute of Diabetes
cutaneous) by multivariate Cox proportional hazard modeling adjusted for demographics and clinical parameters. and Digestive and
Kidney Diseases,
Results Epoetin doses used to achieve equivalent hemoglobin responses in study patients were, on average, 25% National Institutes of
higher when epoetin was administered intravenously rather than subcutaneously (as expected). Moreover, Health, Building 10,
adverse composite event outcomes were found to be significantly more likely to occur during follow-up for Room 9N312, 10
Center Drive (MSC-
patients on hemodialysis managed with intravenous rather than subcutaneous epoetin (adjusted hazard ratio for 1822), Bethesda, MD
adverse events within 1 year [intravenous versus subcutaneous] was 1.11 [95% confidence interval, 1.04 to 1.18]). 20892. Email: daniel.
[email protected]
Conclusions This study finds that treatment of patients on hemodialysis with subcutaneous epoetin is associated
with more favorable clinical outcomes than those associated with intravenous epoetin treatment.
Clin J Am Soc Nephrol 10: 1822–1830, 2015. doi: 10.2215/CJN.01590215

Introduction National Kidney Foundation (NKF) in 1997 and 2000

Recombinant human erythropoietin (epoetin-a) treat- (NKF Disease Outcomes Quality Initiative [DOQI]) rec-
ment increases blood hemoglobin levels in almost all ommended that sc rather than iv epoetin be used to
patients with anemia of ESRD and has been a main- treat patients with kidney failure (13–15). This recom-
stay of managing these patients for decades (1–6). mendation also reflected the principle of limiting drug
Both intravenous (iv) and subcutaneous (sc) epoetin use to the minimum required to achieve a therapeutic
effectively ameliorate anemia of kidney failure. How- goal. Nonetheless, standard practice in the United
ever, epoetin pharmacokinetics and activity differ de- States outside of the Veterans Affairs Health Care Sys-
pending on the route of administration. Elevated tem (16) has been to treat patients on hemodialysis with
epoetin blood levels after sc injection are much lower iv epoetin (17). Convenience, patient comfort, and con-
than after iv infusion of equivalent doses but more cerns about rare cases of pure red cell aplasia (PRCA)
prolonged, and lower doses are required to achieve caused by drug–induced antierythropoietin antibodies
the same hemoglobin response when administered sc have been rationales for favoring iv epoetin in hemo-
rather than iv (7–11). dialysis management (17–21). However, clinical care
Epoetin-a is expensive (.$1.7 billion in costs to reimbursement policies also favored iv epoetin use. Be-
Medicare in 2010) (12). Because of substantial cost fore 2011, Medicare, which covers most costs of hemo-
savings that could be achieved without compromis- dialysis care in the United States, paid for epoetin
ing quality of care, clinical guidelines issued by the treatment on a drug cost basis, and reimbursements

1822 Copyright © 2015 by the American Society of Nephrology www.cjasn.org Vol 10 October, 2015
Clin J Am Soc Nephrol 10: 1822–1830, October, 2015 Outcomes Associated with Erythropoeitin Dose and Route, Wright et al. 1823

for administering epoetin accounted for up to 25% of dialy- cause of ESRD, dialysis vintage, height and weight, type of
sis center income (22). vascular access, prescribed doses of epoetin, and route of
Furthermore, before 2006, there was no clear evidence of administration. Laboratory measures included blood hemo-
dose–related epoetin toxicities for patients with kidney globin, transferrin saturation, ferritin concentration, and
failure. Since 2006, however, several large randomized tri- serum albumin as well as procedural data used to calculate
als evaluating epoetin and darbopoetin (a long-acting form spKt/V as a measure of dialysis adequacy. Epoetin doses,
of epoetin) treatment regimens in kidney disease found laboratory measurements, and spKt/V values for indi-
that risks of cardiovascular complications and early death vidual patients represented averages of those recorded
were increased significantly when high doses of drug were during the 3-month periods when clinical information
used to maximize hemoglobin responses (23–26). These was obtained for the ESRD CPM Project (27). Character-
findings raised the possibility that sc epoetin might be istics of individual dialysis facilities providing patient
safer than iv epoetin for patients on hemodialysis because data to the CPM Project (e.g., nonprofit versus for profit,
of its dose-sparing advantages. hospital based versus nonhospital based) were obtained
To explore relationships between epoetin dose, route of from USRDS records.
administration, and clinical outcomes in patients with ESRD,
we conducted a retrospective cohort study of .60,000 pa- Determination of Clinical Outcomes and Data Analyses
tients on hemodialysis using data from the Centers for Medi- Adverse composite event outcomes were examined as
care and Medicaid Services (CMS) ESRD Clinical defined previously in randomized studies of patients with
Performance Measures (CPM) Project (www.cms.hhs. kidney disease (23,24). A composite event outcome represen-
gov/cpmproject/) linked to CMS administrative data ted the occurrence of one or more clinical events (death and/
maintained by the US Renal Data System (USRDS; www. or hospitalization for congestive heart failure [CHF], acute
usrds.org). Clinical outcomes of patients on hemodialysis myocardial infarction [AMI], and/or stroke [cerebrovascular
managed with iv versus sc epoetin were compared, and accident (CVA)]) during the 2 years after a patient entered
both dose and route of epoetin administration were evaluated the CPM Project. Patients were classified as having had an
as independent risk factors for adverse clinical outcomes. AMI, CHF, or CVA if hospitalized with primary interna-
tional classification of diseases (ICD-9) diagnosis codes of
410 (AMI), 428 (CHF), or 430–436 (CVA). Patient records
Materials and Methods were also screened for additional ICD-9 codes as described
Study Participants by Collins et al. (28) to detect possible patients with PRCA.
Details of the CMS ESRD CPM have been reported All study patients received epoetin-a and were treated
previously (27). For this project, patients on hemodialysis exclusively with either iv or sc epoetin during periods of
($18 years old) stratified by regional ESRD Networks con- data collection. To control for unmeasured factors related
tracted to monitor quality of care were selected randomly to network and facility differences, the percentage of patients
from a census of all surviving patients undergoing hemodi- treated exclusively with sc epoetin was determined for each
alysis on December 31st of successive years. Data used for of 18 hemodialysis networks defined in the CPM Project,
this study encompass 9 years of the CPM Project (1997–2005) and these were grouped into four categories of sc epoetin
and represent information collected from patient records administration (,4%, 4%–7%, 8%–11%, and $12% of pa-
during 3-month periods immediately preceding each study tients). Bivariate analyses were conducted to compare
year (i.e., 2005 data reflected care and observations recorded groups using chi square statistics for categorical variables
from October 1 to December 31, 2004). If a patient was ran- and the t test or Wilcoxon rank–sum test for continuous
domly selected for the CPM Project more than one time, only variables. Life table cumulative event rates were calculated
data from the first year of selection were included for anal- using Kaplan–Meier estimates and evaluated by the log-rank
ysis. Data from CPM patient records were then linked to test. Multivariate Cox proportional hazard analyses strati-
CMS administrative data maintained by the USRDS. Baseline fied by network group and adjusted for patient character-
clinical data, including epoetin-a treatment parameters, came istics reported to affect clinical outcomes (e.g., nonblack race,
from the CPM Project, whereas outcomes data (i.e., hospital- older age, diabetes as the cause of ESRD, and low serum
izations and mortality) were provided by the USRDS. albumin [27,29,30]) as well as individual dialysis facility
CPM Project data included records for 68,278 patients characteristics were carried out to determine whether varia-
from 1997 to 2005 after elimination of duplicates. Patients bles of epoetin dose and route of administration were asso-
were excluded if they could not be linked to USRDS patient ciated independently with differences in the incidence of
files (471), did not receive epoetin (2858), lacked information adverse composite events. Separate regression analyses
about epoetin dosing or route of administration (1371), or had were done for 12 and 24 months of follow-up. Patients
died or undergone kidney transplantation during the final were censored at the time of transplant if they received a
3 months of the year when selected for enrollment (69). For the kidney transplant during follow-up. All data analyses were
remaining 63,509 patients, 57,602 patients received iv epoetin, conducted using SAS, version 9.3 (SAS, Cary, NC).
5108 patients received sc epoetin, and 799 patients received
both and were excluded from the study. The remaining 62,710
patients constituted the study population for analysis. Results
Clinical Characteristics of Patients on Hemodialysis
Determination of Clinical Variables Managed with iv Versus sc Epoetin-a
Demographic information from the CPM Project included A majority (53%) of 62,710 adults with ESRD included
sex, age, race, and ethnicity. Clinical information included in this study had been on dialysis for .2 years at the
1824 Clinical Journal of the American Society of Nephrology

time of entry into the ESRD CPM Project (dialysis Differences in iv Versus sc Epoetin Doses
vintage .2.0 years) (Table 1). Overall, 8% (5108) were From 1997 to 2000, the use of sc epoetin increased slightly
managed with sc epoetin, whereas 92% (57,602) received from 6.5% to 11.0% of patients but then declined to 4.3% by
iv epoetin. The sc epoetin cohort was found to in- 2005 (Figure 1A). For patients managed with sc epoetin,
clude greater proportions of patients with clinical char- average doses prescribed did not change substantially
acteristics associated with a poor prognosis (e.g., from 1997 to 2005. However, average iv epoetin doses
nonblack race, older age, diabetes as the cause of overall were greater (P,0.001) than those prescribed for
ESRD, and low serum albumin [29,30]) than the iv cohort sc administration, which was expected (7,8), and they also
(Table 1). increased progressively over time (P,0.001) (Figure 1B).

Table 1. Characteristics of patients treated with intravenous versus subcutaneous epoetin

Route of Administration
Demographic and Clinical
Variables of Study Patients Total Intravenous Subcutaneous
P Value
(n=62,710) (n=57,602) (n=5108)

Men (%) 52.4 52.2 54.7 ,0.001

Black (%) 37.1 37.6 30.9 ,0.001
Hispanic (%) 12.6 12.6 12.6 0.97
Age, yr
Mean (SD) 61.4 (15.3) 61.3 (15.3) 62.0 (15.2) 0.004
Median (interquartile range) 63.3 (50.7–73.3) 63.2 (50.7–73.3) 64.1 (51.9–73.8)
,65 yr old (%) 46.0 45.8 47.9 0.004
DM as the cause of ESRDa (%) 42.2 42.0 44.7 ,0.001
History of CHF, AMI, or stroke (%) 25.9 26.1 23.7 ,0.001
Dialysis vintage, yr (%)
,0.5 12.2 12.0 14.4 ,0.001
0.5–0.9 14.2 14.1 15.2
1.0–1.9 20.5 20.4 21.5
$2.0 53.1 53.4 48.9
Postdialysis BMI, kga
Mean (SD) 26.5 (7.00) 26.5 (7.0) 26.7 (6.8) 0.07
Median (interquartile range) 25.2 (21.9–29.7) 25.4 (21.9–29.7) 25.4 (22.1–29.8)
$25 (%) 51.5 51.4 52.6 0.09
Hgb
Mean (SD) 11.45 (1.23) 11.45 (1.23) 11.37 (1.22) ,0.001
Median (interquartile range) 11.53 (10.7–12.2) 11.53 (10.8–12.2) 11.47 (10.6–12.1)
#11.5 (%) 49.7 49.5 51.6 0.004
TSAT,20a (%) 23.6 23.4 26.2 ,0.001
Serum ferritin ,100 ng/ml (%) 11.6 11.5 13.0 0.002
Iron deficienta,b (%) 5.2 5.1 6.8 ,0.001
Prescribed intravenous irona (%) 63.5 63.5 63.6 0.95
Serum albumina $3.5/3.8 g/dlc (%) 75.6 76.2 67.9 ,0.001
Epoetin dose, units/kg per week
Mean (SD) 223 (158) 227 (159) 182 (133) ,0.001
Median (interquartile range) 183 (101–308) 187 (103–314) 146 (82–245) ,0.001
,150 (%) 40.7 39.8 51.0 ,0.001
150–299 (%) 33.0 33.1 32.2
$300 (%) 26.3 27.1 16.7
Facility (%)
Nonprofit 25.0 23.4 42.7 ,0.001
Hospital based 16.0 14.8 29.6 ,0.001
Chain affiliated 61.2 63.4 36.5 ,0.001
,20 stations 49.9 49.8 51.3 0.04

Continuous variables (with the exception of age and dialysis vintage) represent the mean of measurements recorded during 3-month
study periods. DM, diabetes mellitus; CHF, congestive heart failure; AMI, acute myocardial infarction; BMI, body mass index; Hgb,
hemoglobin; TSAT, transferrin saturation.
a
Numbers of patients for whom data were available (all, intravenous, and subcutaneous): DM as the cause of ESRD (62,698, 57,591, and
5107), BMI (61,994, 56,995, and 4999), TSAT (59,082, 54,442, and 4640), serum ferritin (59,890, 55,105, and 4785), iron deficient (57,489,
52,979, and 4510), prescribed iron (62,494, 57,404, and 5090), and serum albumin (62,394, 57,324, and 5070).
b
Mean TSAT ,20% and mean serum ferritin ,100 ng/ml.
c
Bromcresol green/bromcresol purple laboratory methods.
Clin J Am Soc Nephrol 10: 1822–1830, October, 2015 Outcomes Associated with Erythropoeitin Dose and Route, Wright et al. 1825

Consequently, a greater proportion of patients managed (11.3761.22 versus 11.4561.23 g/dl; mean 6 SD), and
with sc epoetin (51%) received relatively low doses proportions of patients with blood hemoglobin levels
(,150 units/kg per week) than those treated with iv epoetin maintained at each of varying levels from ,10 to $13
(40%), and a greater proportion of patients treated with iv g/dl were very similar for the iv and sc epoetin cohorts
epoetin (27%) received high doses (.300 units/kg per (Figure 2A). However, median epoetin doses (as well as
week; representing the top quartile of epoetin doses) than the 25% and 75% quartile doses) at all levels of hemoglo-
patients treated with sc epoetin (17%; P,0.001). bin response were substantially lower (by 20%–28%)
Average blood hemoglobin levels maintained in patients when epoetin was administered sc rather than iv
managed with sc versus iv epoetin were nearly identical (P,0.001) (Figure 2B).

Figure 1. | Route of epoetin (Epo) administration and Epo dosing changed from 1997 to 2005. (A) Percentage of patients on hemodialysis
managed with subcutaneous (SC) versus intravenous (IV) Epo from study year 1997 to 2005. (B) Mean doses of Epo (6SD) used to
manage study subjects from study year 1997 to 2005. Slopes calculated by linear regression indicate that the mean doses of IV epoetin
increased by 7.0 units/kg per week per year (P,0.001), whereas mean doses of SC epoetin increased by 1.5 units/kg per week per year
(P=0.06).
1826 Clinical Journal of the American Society of Nephrology

Association of Epoetin Dose and Route of Administration composite event within 1 year of follow-up, and 40% expe-
with Adverse Clinical Outcomes rienced an adverse composite event within 2 years. Death
Study cohorts were compared with respect to incidence was the most frequent adverse event (78% of all events
of adverse composite events, which as defined in the occurring within 2 years and the only event in 58% of pa-
Correction of Hemoglobin and Outcomes in Renal In- tients). Also, as observed in the CHOIR Study (23,24), CHF
sufficiency (CHOIR) Study (23,24), included death and/or leading to hospitalization was more frequent (23%) than
hospitalization for CHF, AMI, or CVA. High risks of cardio- AMI (8%) or CVA (11%) as an initial adverse event.
vascular events and death in patients with ESRD are well Kaplan–Meier life table analysis showed that the cumu-
documented (29–36) and evident in this study. Overall, 28% lative rate of adverse composite events was significantly
of 62,710 study participants experienced an adverse greater overall for patients managed with iv versus sc

Figure 2. | Proportions of patients managed with intravenous (IV) or subcutaneous (SC) epoetin (Epo), with blood hemoglobin levels
maintained at varying levels, were very similar, but median doses of Epo used to achieve these hemoglobin levels were different. (A) Per-
centage of patients managed with SC or IV Epo whose blood hemoglobin levels were maintained at varying levels from ,10 to $13 g/dl.
Overall distributions of patients at varying hemoglobin levels were very similar but statistically different (P,0.01) for the cohorts managed with
IV versus SC epoetin. (B) Medians and 25th and 75th percentiles (indicated by lines) of Epo doses (units per kilogram per week) used in patients
managed with IV or SC Epo to achieve blood hemoglobin levels of varying levels from ,10 to $13 g/dl. Median doses are lower for SC
compared with IV within each hemoglobin category by 20%–28% (P,0.001 by the Wilcoxon rank–sum test).
Clin J Am Soc Nephrol 10: 1822–1830, October, 2015 Outcomes Associated with Erythropoeitin Dose and Route, Wright et al. 1827

epoetin (Figure 3A, P,0.001) as well as patient subgroups hazard ratio, 1.12; 95% CI, 1.03 to 1.23; n=9305). Moreover,
with hemoglobin levels either above or below the median higher adverse event rates and elevated AHRs for these
of 11.5 g/dl (Figure 3, B, P,0.001 and C, P,0.001), repre- events were observed consistently with iv versus sc epoetin
senting patients who were relatively epoetin sensitive or independent of patient age, history of a prior event, albu-
resistant, respectively. Cumulative adverse event rates min level, hemoglobin response, and the type of dialysis
(Supplemental Figure 1) were also significantly greater facility managing care (Supplemental Table 1).
for patients managed with iv versus sc epoetin at relatively Concerns about patients with rare drug–associated
low doses of epoetin (,150 units/kg per week; P=0.02) PRCA have led to recommendations that patients on he-
and intermediate doses (150–299 units/kg per week; modialysis be treated with iv rather than sc epoetin
P=0.01) but not those treated with the highest doses (17,19,21). We, therefore, examined CMS records for occur-
($300 units/kg per week; P=0.44). rences of PRCA among 62,710 study participants within
Multivariate Cox proportional hazard modeling adjusted 5 years after their entry into the ESRD CPM Project. Before
for both patient characteristics known to affect clinical out- 2008, no specific ICD-9 code was defined for PRCA; how-
comes (e.g., nonblack race, older age, diabetes as the cause of ever, this diagnosis was included among other forms of
ESRD, and low serum albumin [27,29,30]) and various dialy- acquired aplastic anemia (284.8). We, therefore, screened
sis facility characteristics (e.g., nonprofit versus for profit, etc.) records for this ICD-9 code and then used other diagnosis,
also indicated that the risks of adverse events were higher for procedure, and treatment codes to detect possible cases of
patients treated with iv versus sc epoetin (Table 2, all sub- PRCA, using the criteria described by Collins et al. (28) (i.e.,
jects). Adjusted hazard ratios (AHR) were highest and sig- records of bone marrow examination, persistent anemia,
nificantly elevated (95% confidence interval [95% CI], 1.01 and repeated red cell transfusions but not other causes of
to 1.26) for patients treated with low and intermediate marrow failure). Of 57,602 study patients treated with iv
doses of epoetin (,300 units/kg per week) representing epoetin, 92 (0.16%) had a 284.8 code recorded at some
approximately 75% of the study population (Table 2, time within 5 years of entry into the CPM Project com-
AHR for iv versus sc epoetin by dose level) and for patients pared with 3 (0.06%) of 5108 patients treated with sc epoetin.
with hemoglobin levels maintained above 11.5 g/dl who However, of all patients, during 201,655 patient-years of
were relatively epoetin sensitive (Table 2, AHR for iv ver- follow-up, only 1 patient with possible PRCA (treated
sus sc epoetin by hemoglobin level); however, AHRs (iv with iv epoetin) fulfilled the criteria of Collins et al. (28).
versus sc epoetin) were elevated for all subgroups (AHR,
1.06 to 1.15) but not statistically different between epoetin
dose and hemoglobin subgroups (P=0.38 and P=0.40, re- Discussion
spectively). Higher risks of adverse events associated with This retrospective cohort study of .60,000 adult patients
iv versus sc epoetin were observed in ESRD networks that on hemodialysis confirms that higher doses of epoetin-a are
used sc epoetin both least frequently (,4% of patients; haz- required to achieve equivalent hemoglobin responses in pa-
ard ratio, 1.19; 95% confidence interval [95% CI], 1.04 to tients with kidney failure when administered iv rather than
1.36; n=16,476) and most frequently (.12% of patients; sc. This study also finds that risks of early death and/or

Figure 3. | Cumulative adverse event rates are greater for patients treated with intravenous (IV) epoetin (Epo) versus subcutaneous (SC) Epo.
Kaplan–Meier life table analysis of time to an adverse composite event (death or hospitalization for congestive heart failure, acute myocardial
infarcation, or cerebrovascular accident) during 2 years of follow up for patients treated with IV (black lines) versus SC Epo (gray lines). (A) All
patients (n=60,013; P,0.001). (B) Hemoglobin (Hgb) #11.5 g/dl (n=29,669; P,0.001). (C) Hgb.11.5 g/dl (n=30,344; P,0.001).
1828 Clinical Journal of the American Society of Nephrology

Table 2. Multivariate Cox proportional hazard model analysis of risk for 12- and 24-month composite events (hospitalization caused
by acute myocardial infarction, congestive heart failure, or stroke and/or death during the 12-or 14-month follow–up periods)

Demographic and Clinical 12-mo Composite Events 24-mo Composite

Variables of Study Patients AHR (95% CI)d Events AHR (95% CI)d

All subjects
Intravenous versus subcutaneous epoetin 1.11 (1.04 to 1.18) 1.09 (1.04 to 1.14)
Men versus women 1.03 (1.00 to 1.07) 1.03 (1.01 to 1.06)
Black versus nonblack race 0.85 (0.82 to 0.88) 0.86 (0.83 to 0.88)
Hispanic versus non-Hispanic 0.91 (0.86 to 0.95) 0.89 (0.85 to 0.92)
Age, yra 1.03 (1.02 to 1.03) 1.03 (1.03 to 1.03)
Dialysis vintage, yrb 1.01 (1.00 to 1.01) 1.01 (1.00 to 1.01)
Diabetes mellitus as the cause of ESRD 1.21 (1.18 to 1.25) 1.26 (1.22 to 1.29)
Previous AMI, CHF, or stroke 1.89 (1.83 to 1.96) 1.78 (1.73 to 1.82)
BMI$25 versus ,25 0.82 (0.79 to 0.84) 0.85 (0.82 to 0.87)
Hemoglobin #11.5 g/dl 1.21 (1.18 to 1.25) 1.17 (1.14 to 1.20)
Serum albumin $3.5/3.2 g/dlc 0.65 (0.63 to 0.67) 0.69 (0.67 to 0.71)
Epoetin dose 150–299 versus ,150 units/kg per week 1.18 (1.14 to 1.23) 1.15 (1.12 to 1.19)
Epoetin dose $300 versus ,150 units/kg per week 1.51 (1.46 to 1.57) 1.44 (1.40 to 1.49)
Facility: nonprofit 0.93 (0.89 to 0.98) 0.93 (0.89 to 0.96)
Facility: hospital based 0.87 (0.82 to 0.92) 0.87 (0.83 to 0.92)
Facility: chain affiliated 1.04 (1.00 to 1.07) 0.99 (0.96 to 1.02)
Facility: ,20 stations 1.00 (0.97 to 1.03) 1.01 (0.98 to 1.03)
AHR for intravenous versus
subcutaneous epoetin by dose level
Intravenous versus subcutaneous 1.12 (1.02 to 1.22) 1.11 (1.03 to 1.19)
(,150 units/kg per week)
Intravenous versus subcutaneous 1.13 (1.01 to 1.25) 1.09 (1.00 to 1.18)
(150–299 units/kg per week)
Intravenous versus subcutaneous 1.06 (0.94 to 1.20) 1.04 (0.94 to 1.15)
($300 units/kg per week)
AHR for intravenous versus subcutaneous
epoetin by hemoglobin level
Intravenous versus subcutaneous 1.08 (1.00 to 1.17) 1.06 (1.00 to 1.13)
(hemoglobin #11.5 g/dl)
Intravenous versus subcutaneous 1.15 (1.05 to 1.26) 1.12 (1.05 to 1.20)
(hemoglobin .11.5 g/dl)

Continuous variables (with the exception of age and dialysis vintage) represent the mean of measurements recorded during 3-month
study periods; n=60,013 after exclusion of 2697 with missing data. AHR, adjusted hazard ratio; 95% CI, 95% confidence interval; AMI,
acute myocardial infarction; CHF, congestive heart failure; BMI, body mass index.
a
As of entry into the Clinical Performance Measures Project survey.
b
As of December 31st of the study period year.
c
Bromcresol green/bromcresol purple laboratory methods.
d
Stratified by network and grouped according to the frequency of use of subcutaneous epoetin.

hospitalization for cardiovascular complications were sig- injection are substantially lower than those after iv infusion
nificantly greater for patients managed with iv versus sc of identical doses, and it is possible that increased risks of
epoetin. Moreover, higher adverse event rates and signifi- adverse events associated with epoetin treatment occur pri-
cantly elevated AHRs for adverse events were observed marily when epoetin blood levels exceed a certain threshold.
consistently in association with iv versus sc epoetin admin- Given the distinct pharmacokinetics of iv versus sc epoetin
istration for patients independent of age, history of a prior (9), lower doses of epoetin would be more likely to exceed
adverse event, albumin level, hemoglobin response, and such a threshold when administered iv rather than sc. How-
type of dialysis facility managing care. Because patients ever, because epoetin has greater erythropoietic bioactivity
treated with iv epoetin consistently required higher doses when given sc rather than iv, patients treated with high doses
to achieve equivalent hemoglobin responses than those of sc epoetin (.300 units/kg per week) likely included a
given sc epoetin, these findings support the conclusion greater proportion of patients who were drug resistant than
that epoetin exposure per se affected clinical outcomes in a those who received the same doses iv. Because epoetin re-
dose-dependent manner, as was indicated by a secondary sistance is per se a surrogate risk factor for adverse clinical
analysis of the CHOIR Study (24). outcomes, this may explain why higher risks of adverse clin-
Pharmacokinetic studies of iv and sc epoetin administration ical outcomes with iv versus sc epoetin were less evident at
in healthy volunteers and patients with ESRD (9–11) have doses .300 units/kg per week (Table 2, AHR for iv versus
shown that peak epoetin blood levels attained after sc sc epoetin by dose level, Supplemental Figure 1).
Clin J Am Soc Nephrol 10: 1822–1830, October, 2015 Outcomes Associated with Erythropoeitin Dose and Route, Wright et al. 1829

There is long-standing evidence that expression of eryth- hospitalizations for cardiovascular complications by mini-
ropoietin receptors is not restricted to erythroid precursors in mizing epoetin dosing. Assuming that findings from this
the marrow but is also detectable in nonerythroid tissues, study accurately represent all patients with ESRD in the
including neurons, adipocytes, and vascular endothelium United States during the years 1997–2005, it can be estimated
(37). There is also evidence that erythropoietin affects vas- (47) that nearly 30,000 early deaths and/or hospitalizations
cular endothelial function and influences angiogenesis (38– for cardiovascular complications (approximately 3300 per
42). Hence, nonerythropoietic effects of epoetin, particularly year) might have been avoided had all patients been man-
at high pharmacologic blood levels, could have a role in aged with sc epoetin. (This estimate is made on the basis of
promoting adverse clinical events associated with epoetin a number needed to treat of 27, calculated from observed
treatment in patients with kidney disease. adverse composite event rates within 12 months of 0.282 for
Doses of epoetin and other erythropoiesis–stimulating patients treated with iv epoetin versus 0.249 for patients
agents (ESAs) used to manage patients with kidney dis- treated with sc epoetin, an AHR of 1.11 for adverse events
ease in the United States have declined in recent years [iv versus sc], and a total of 872,702 new patients on hemo-
(USRDS 2013 information) (12,43,44). Evidence that ag- dialysis [USRDS] from 1997 to 2005, of whom 92% [802,886]
gressive ESA treatment regimens may be hazardous for were treated with iv epoetin.) However, because this is a ret-
these patients, reported since 2006, has likely prompted rospective, observational study, conclusions drawn from it
this change. However, financial considerations may also must be qualified, because unmeasured hemodialysis facility
have had an influence. Medicare, which pays for most and patient management variables not accounted for by
costs of hemodialysis care in the United States, discontin- multivariate analysis may have affected the results.
ued separate reimbursements for ESA treatment on a drug
cost basis in 2011 (12,43,44), including them instead in Acknowledgments
bundled payment formulae for dialysis care, thereby re- The authors thank Dr. Diane Frankenfield for her contributions to
moving financial disincentives to limit ESA use. We were the initial data analysis that led to the study before her retirement
unable to determine whether the use of iv versus sc epoetin from the Office of Research, Development, and Information at the
at dialysis centers has also changed in recent years, Centers for Medicare and Medicaid Services.
because information about the route of epoetin adminis- This study was supported by intramural programs of the National
tration became unavailable through the CMS after 2006. Institute of Diabetes and Digestive and Kidney Diseases of the
However, the new Consolidated Renal Operations in National Institutes of Health.
a Web-enabled Network data reporting system for track-
ing ESRD management (http://projectcrownweb.org/ Disclosures
assets/release_notes/Kidney_Data_Dictionary_Search.html) None.
should provide this information in the future.
The choice of iv versus sc epoetin to treat patients on References
hemodialysis in the United States has been influenced by 1. Eschbach JW, Abdulhadi MH, Browne JK, Delano BG, Downing
concerns about the immunogenic potential of epoetin and MR, Egrie JC, Evans RW, Friedman EA, Graber SE, Haley NR,
patients with rare cases of drug-associated PRCA. These Korbet S, Krantz SB, Lundin AP, Nissenson AR, Ogden DA,
concerns were cited as a basis for changes in NKF DOQI Paganini EP, Rader B, Rutsky EA, Stivelman J, Stone WJ, Teschan P,
Van Stone JC, Van Wyck DB, Kenneth Zuckerman K, Adamson
guidelines in 2006 and led to recommendations by Amgen JW: Recombinant human erythropoietin in anemic patients with
and the Food and Drug Administration in 2005 and 2006 end-stage renal disease. Results of a phase III multicenter clinical
(17,19,21,45) that iv rather than sc epoetin be used to treat trial. Ann Intern Med 111: 992–1000, 1989
patients on hemodialysis, despite the fact that patients with 2. Eschbach JW, Kelly MR, Haley NR, Abels RI, Adamson JW: Treat-
ment of the anemia of progressive renal failure with recombinant
rare cases of PRCA were observed primarily outside the human erythropoietin. N Engl J Med 321: 158–163, 1989
United States and associated with Eprex, a form of epoetin 3. Valderrábano F: Erythropoietin in chronic renal failure. Kidney
not marketed in the United States (17–19). Although cases Int 50: 1373–1391, 1996
of PRCA have been documented in individuals treated with 4. Jones M, Ibels L, Schenkel B, Zagari M: Impact of epoetin alfa on
Epogen and Procrit (epoetin-a used in the United States), clinical end points in patients with chronic renal failure: A meta-
analysis. Kidney Int 65: 757–767, 2004
these cases have been extremely rare (incidence estimated 5. Ross SD, Fahrbach K, Frame D, Scheye R, Connelly JE, Glaspy J: The
at 0.2/100,000 patient-years) (18) and not exclusively asso- effect of anemia treatment on selected health-related quality-of-life
ciated with sc administration (18,19), consistent with our domains: A systematic review. Clin Ther 25: 1786–1805, 2003
finding of only one possible patient with PRCA per 6. Pisoni RL, Bragg-Gresham JL, Young EW, Akizawa T, Asano Y,
Locatelli F, Bommer J, Cruz JM, Kerr PG, Mendelssohn DC, Held PJ,
201,655 patient-years (associated with iv epoetin). Port FK: Anemia management and outcomes from 12 countries
Several different ESAs are now available in the United in the Dialysis Outcomes and Practice Patterns Study (DOPPS).
States to treat anemia of ESRD (46). However, epoetin-a Am J Kidney Dis 44: 94–111, 2004
(Epogen) continues to be used widely in the United States. 7. Kaufman JS, Reda DJ, Fye CL, Goldfarb DS, Henderson WG,
Although doses of epoetin and other ESAs used in hemodi- Kleinman JG, Vaamonde CA; Department of Veterans Affairs
Cooperative Study Group on Erythropoietin in Hemodialysis
alysis management have declined, our study’s findings re- Patients: Subcutaneous compared with intravenous epoetin in
main relevant to current practice, because observed patients receiving hemodialysis. N Engl J Med 339: 578–583, 1998
differences in clinical outcomes associated with iv versus 8. Besarab A, Reyes CM, Hornberger J: Meta-analysis of sub-
sc epoetin treatment were found to be most evident at lower cutaneous versus intravenous epoetin in maintenance treatment
of anemia in hemodialysis patients. Am J Kidney Dis 40: 439–
dose levels. In conclusion, this study provides evidence that 446, 2002
routine use of sc rather than iv epoetin to manage anemia in 9. Brockmöller J, Köchling J, Weber W, Looby M, Roots I, Neumayer
patients on hemodialysis could enhance survival and reduce HH: The pharmacokinetics and pharmacodynamics of
1830 Clinical Journal of the American Society of Nephrology

recombinant human erythropoietin in haemodialysis patients. Br 30. Price DA, Owen WF Jr.: African-Americans on maintenance di-
J Clin Pharmacol 34: 499–508, 1992 alysis: A review of racial differences in incidence, treatment, and
10. McMahon FG, Vargas R, Ryan M, Jain AK, Abels RI, Perry B, survival. Adv Ren Replace Ther 4: 3–12, 1997
Smith IL: Pharmacokinetics and effects of recombinant human 31. Owen WF Jr., Chertow GM, Lazarus JM, Lowrie EG: Dose of
erythropoietin after intravenous and subcutaneous injections in hemodialysis and survival: Differences by race and sex. JAMA
healthy volunteers. Blood 76: 1718–1722, 1990 280: 1764–1768, 1998
11. Besarab A: Physiological and pharmacodynamic considerations for 32. de Jager DJ, Grootendorst DC, Jager KJ, van Dijk PC, Tomas LM,
route of EPO administration. Semin Nephrol 20: 364–374, 2000 Ansell D, Collart F, Finne P, Heaf JG, De Meester J, Wetzels JF,
12. USRDS: 2013 USRDS Annual Data Report: Atlas of chronic Rosendaal FR, Dekker FW: Cardiovascular and non-
kidney disease and end-stage renal disease in the United States. cardiovascular mortality among patients starting dialysis. JAMA
Am J Kidney Dis 63[Suppl]: e1–e478, 2014 302: 1782–1789, 2009
13. National Kidney Foundation: NFK-DOQI clinical practice 33. Cheung AK, Sarnak MJ, Yan G, Berkoben M, Heyka R, Kaufman
guidelines for the treatment of anemia of chronic renal failure. A, Lewis J, Rocco M, Toto R, Windus D, Ornt D, Levey AS; HEMO
National Kidney Foundation-Dialysis Outcomes Quality Initia- Study Group: Cardiac diseases in maintenance hemodialysis
tive. Am J Kidney Dis 30[4 Suppl 3]: S192–S240, 1997 patients: Results of the HEMO Study. Kidney Int 65: 2380–2389,
14. National Kidney Foundation: IV. NKF-K/DOQI clinical practice 2004
guidelines for anemia of chronic kidney disease: Update 2000. 34. Casserly LF, Reddy SM, Dember LM: Venous thromboembolism
Am J Kidney Dis 37[Suppl 1]: S182–S238, 2001 in end-stage renal disease. Am J Kidney Dis 36: 405–411, 2000
15. Hynes DM, Stroupe KT, Greer JW, Reda DJ, Frankenfield DL, 35. Lindner A, Charra B, Sherrard DJ, Scribner BH: Accelerated
Kaufman JS, Henderson WG, Owen WF, Rocco MV, Wish JB, Kang atherosclerosis in prolonged maintenance hemodialysis. N Engl J
J, Feussner JR: Potential cost savings of erythropoietin administra- Med 290: 697–701, 1974
tion in end-stage renal disease. Am J Med 112: 169–175, 2002 36. Zhang Y, Thamer M, Cotter D, Kaufman J, Hernán MA: Estimated
16. Department of Veterans Affairs: FY97 Veterans Affairs Dialysis effect of epoetin dosage on survival among elderly hemodialysis
Case Mix Study 1998, Milwaukee, WI, National Center for Cost patients in the United States. Clin J Am Soc Nephrol 4: 638–644, 2009
Containment, 1998 37. Arcasoy MO: The non-haematopoietic biological effects of
17. Hynes DM, Stroupe KT, Kaufman JS, Reda DJ, Peterman A, erythropoietin. Br J Haematol 141: 14–31, 2008
Browning MM, Huo Z, Sorbara D: Adherence to guidelines for 38. Watanabe D, Suzuma K, Matsui S, Kurimoto M, Kiryu J, Kita M,
ESRD anemia management. Am J Kidney Dis 47: 455–461, 2006 Suzuma I, Ohashi H, Ojima T, Murakami T, Kobayashi T, Masuda
18. Bennett CL, Luminari S, Nissenson AR, Tallman MS, Klinge SA, S, Nagao M, Yoshimura N, Takagi H: Erythropoietin as a retinal
McWilliams N, McKoy JM, Kim B, Lyons EA, Trifilio SM, Raisch angiogenic factor in proliferative diabetic retinopathy. N Engl J
DW, Evens AM, Kuzel TM, Schumock GT, Belknap SM, Locatelli Med 353: 782–792, 2005
F, Rossert J, Casadevall N: Pure red-cell aplasia and epoetin 39. Anagnostou A, Liu Z, Steiner M, Chin K, Lee ES, Kessimian N,
therapy. N Engl J Med 351: 1403–1408, 2004 Noguchi CT: Erythropoietin receptor mRNA expression in human
19. McKoy JM, Stonecash RE, Cournoyer D, Rossert J, Nissenson AR, endothelial cells. Proc Natl Acad Sci U S A 91: 3974–3978, 1994
Raisch DW, Casadevall N, Bennett CL: Epoetin-associated pure 40. Bahlmann FH, De Groot K, Spandau JM, Landry AL, Hertel B,
red cell aplasia: Past, present, and future considerations. Trans- Duckert T, Boehm SM, Menne J, Haller H, Fliser D: Erythropoietin
fusion 48: 1754–1762, 2008 regulates endothelial progenitor cells. Blood 103: 921–926, 2004
20. Gershon SK, Luksenburg H, Coté TR, Braun MM: Pure red-cell 41. Beleslin-Cokic BB, Cokic VP, Yu X, Weksler BB, Schechter AN,
aplasia and recombinant erythropoietin. N Engl J Med 346: Noguchi CT: Erythropoietin and hypoxia stimulate erythropoietin
1584–1586, 2002 receptor and nitric oxide production by endothelial cells. Blood
21. KDOQI; National Kidney Foundation: II. Clinical practice 104: 2073–2080, 2004
guidelines and clinical practice recommendations for anemia in 42. George J, Goldstein E, Abashidze A, Wexler D, Hamed S,
chronic kidney disease in adults. Am J Kidney Dis 47[5 Suppl 3]: Shmilovich H, Deutsch V, Miller H, Keren G, Roth A: Erythro-
S16–S85, 2006 poietin promotes endothelial progenitor cell proliferative and
22. Thamer M, Zhang Y, Kaufman J, Stefanik K, Cotter DJ: Factors influ- adhesive properties in a PI 3-kinase-dependent manner.
encing route of administration for epoetin treatment among hemo- Cardiovasc Res 68: 299–306, 2005
dialysis patients in the United States. Am J Kidney Dis 48: 77–87, 2006 43. Miskulin DC, Zhou J, Tangri N, Bandeen-Roche K, Cook C,
23. Singh AK, Szczech L, Tang KL, Barnhart H, Sapp S, Wolfson M, Ephraim PL, Crews DC, Scialla JJ, Sozio SM, Shafi T, Jaar BG,
Reddan D; CHOIR Investigators: Correction of anemia with epoetin Boulware LE; DEcIDE Network Patient Outcomes in End Stage
alfa in chronic kidney disease. N Engl J Med 355: 2085–2098, 2006 Renal Disease Study Investigators: Trends in anemia manage-
24. Szczech LA, Barnhart HX, Inrig JK, Reddan DN, Sapp S, Califf ment in US hemodialysis patients 2004-2010. BMC Nephrol 14:
RM, Patel UD, Singh AK: Secondary analysis of the CHOIR trial 264, 2013
epoetin-a dose and achieved hemoglobin outcomes. Kidney Int 44. Brunelli SM, Monda KL, Burkart JM, Gitlin M, Neumann PJ, Park
74: 791–798, 2008 GS, Symonian-Silver M, Yue S, Bradbury BD, Rubin RJ: Early
25. Palmer SC, Navaneethan SD, Craig JC, Johnson DW, Tonelli M, trends from the Study to Evaluate the Prospective Payment System
Garg AX, Pellegrini F, Ravani P, Jardine M, Perkovic V, Graziano G, Impact on Small Dialysis Organizations (STEPPS). Am J Kidney
McGee R, Nicolucci A, Tognoni G, Strippoli GF: Meta-analysis: Dis 61: 947–956, 2013
Erythropoiesis-stimulating agents in patients with chronic kidney 45. Brenner R: Amgen letter to Health Care Professionals Available
disease. Ann Intern Med 153: 23–33, 2010 at: http://www.fda.gov/downloads/Safety/MedWatch/
26. Pfeffer MA, Burdmann EA, Chen CY, Cooper ME, de Zeeuw D, SafetyInformation/SafetyAlertsforHumanMedicalProducts/
Eckardt KU, Feyzi JM, Ivanovich P, Kewalramani R, Levey AS, UCM164481.pdf. Accessed
Lewis EF, McGill JB, McMurray JJ, Parfrey P, Parving HH, Remuzzi 46. Drüeke TB: Anemia treatment in patients with chronic kidney
G, Singh AK, Solomon SD, Toto R; TREAT Investigators: A trial disease. N Engl J Med 368: 387–389, 2013
of darbepoetin alfa in type 2 diabetes and chronic kidney disease. 47. Altman DG, Andersen PK: Calculating the number needed to
N Engl J Med 361: 2019–2032, 2009 treat for trials where the outcome is time to an event. BMJ 319:
27. Frankenfield DL, Rocco MV, Roman SH, McClellan WM: Sur- 1492–1495, 1999
vival advantage for adult Hispanic hemodialysis patients? Find- Received: February 10, 2015 Accepted: June 30, 2015
ings from the end-stage renal disease clinical performance
measures project. J Am Soc Nephrol 14: 180–186, 2003
28. Collins AJ, Li S, Adamson JW, Gilbertson DT: Assessment of pure Published online ahead of print. Publication date available at www.
red cell aplasia in US dialysis patients: The limits of the Medicare cjasn.org.
data. Am J Kidney Dis 43: 464–470, 2004
29. Bloembergen WE, Port FK, Mauger EA, Wolfe RA: Causes of This article contains supplemental material online at http://cjasn.
death in dialysis patients: Racial and gender differences. J Am Soc asnjournals.org/lookup/suppl/doi:10.2215/CJN.01590215/-/
Nephrol 5: 1231–1242, 1994 DCSupplemental.