Abstract

Background. Our objective was to determine whether patient factors, processes of care and measures of erythropoietin (EPO) responsiveness were associated with successful anemia management at the individual patient level.

Methods. We retrospectively reviewed laboratory and demographic data from 1499 patients receiving hemodialysis in 15 units operated by the same dialysis provider. We performed univariate and multivariate logistic regression analysis to determine predictors of an average 3-month hemoglobin level below or above the target interval of 11.0–12.5 g/dL. To explain the effect of facility on anemia performance, we calculated correlations between measures of EPO responsiveness and the probability of achieving the target interval by facility.

Results. Patients above the target hemoglobin range demonstrated an association with parathyroid hormone (PTH) (OR = 0.96 per 100 pg/mL increase), female gender (OR = 0.68), EPO protocol use (OR = 0.94 per 10% increase in use) and facility (range of OR = 0.26–2.59 for 15 participating sites). Patients below the target hemoglobin range demonstrated an association with CRP (OR = 1.10 per mg/L increase), PTH (OR = 1.07 per 100 pg/mL increase), iron deficiency (OR = 1.07 per 10% increase), EPO protocol use (OR = 0.89 per 10% increase in use), iron protocol use (OR = 0.93 per 10% increase in use) and facility (range of OR = 0.58–3.41 over 15 units). EPO index ( r = 0.71), EPO dose ( r = 0.73), hemoglobin ( r = −0.60) and EPO per unit weight ( r = 0.76) were significantly correlated with the probability of achieving the target hemoglobin by facility.

Conclusions. The facility significantly influences the outcome of anemia management in patients with ESRD. In part, this is due to the patients’ EPO responsiveness, which may be influenced by facility care patterns.

Introduction

Nearly all of the 340 000 patients on chronic renal replacement therapy in the United States suffer from anemia [ 1 ] and can be successfully treated with appropriate doses of erythropoisis-stimulating agents and iron supplementation. Recent evidence has emphasized the importance of maintaining hemoglobin levels in an optimal range [ 2,3 ]. Nevertheless, reported anemia management is suboptimal among patients on chronic hemodialysis. Only 39% of 31 267 ESRD patients in the United States achieved the target hemoglobin level of 11–12 g/dL [ 4 ] recommended by The Kidney Disease Outcome Quality Initiative (K/DOQI) [ 5 ]. Despite the fact that it remains uncertain what hemoglobin level is truly best for dialysis patients, these findings are concerning given the possible consequences of inappropriately treated anemia. Patients with inadequate treatment have been shown to have worse cardiovascular function [ 6 ], poorer quality of life [ 7 ], increased number of hospitalizations [ 8 ] and decreased patient survival [ 9,10 ], while targeted treatment of anemia to return hemoglobin levels to those found in the general population has been associated with higher morbidity and mortality in many studies, and by meta analysis of the literature [ 2 , 11 , 12 ]. For example, in one study aiming for hemoglobin levels of 13.5 versus 11.3 g/dL [ 11 ], there was no improved quality of life and significantly more adverse events, while another comparing a range of 13–15 g/dL versus 10–11.0 g/dL showed improved quality of life but increased dialysis need and no cardiovascular benefit [ 12 ].

There are many potential explanations for the suboptimal clinical effectiveness of anemia management in the ESRD population. These include multiple patient factors, processes of care and, as recently described, facility factors may also play an important role [ 13 ]. Furthermore, these explanations may differentially affect patients who are normally responsive to erythropoietin (EPO) compared to patients who are EPO resistant. The goal of the present study was to determine the factors associated with successful, on target, anemia management.

Subjects and methods

Database

The study was performed at Satellite Dialysis, Inc., a nonprofit dialysis provider. Patient data for the analysis from 15 participating hemodialysis units were abstracted from a centralized electronic database that stores all demographic, clinical, procedural and hemodialysis events associated with each patient encounter. Satellite Dialysis, through its Medical Policy Committee, utilizes protocol-driven therapy for adequacy, anemia, bone disease, reuse and immunizations using similar equipment and protocols for patient care. Physicians can request their patients be opted out of any protocol.

Subjects

All 1954 patients receiving chronic hemodialysis for >31 days within the study period of August 2005–October 2005 were initially considered for the analysis. Patients were eligible for the study if they received chronic hemodialysis for the entire study period, had three consecutive months of anemia monitoring (at least one hemoglobin laboratory reading per month) and at least one reading for each predictor value over the 3-month study period.

Anemia management

Unless specifically excluded by the patient's nephrologists, all study patients were evaluated once per month by a standardized nurse-administered anemia management protocol to titrate the dosages of recombinant EPO (EPOGEN, Amgen Inc., Thousand Oaks, CA, USA) and IV iron (Venofer, American Reagents, Shirley, NY, USA) according to current laboratory values. To briefly summarize, the implemented protocol targeted a hemoglobin level between 11 and 12.5 g/dL, an iron saturation level between 25% and 50%, and a ferritin level between 200 and 800 ng/mL. These targets were consistent with KDOQI recommendations and Medicare guidance at the time. Patients with iron levels below target would receive 500–1000 mg of IV iron over five sessions and those within target would receive 100–200 mg of maintenance iron every other week. All iron products were held if the ferritin exceeded 800 ng/mL, or there was clinical evidence of infection. EPO doses could be increased or decreased up to 25%. Titration was based on a patient's hemoglobin trend over the past 30–90 days. The EPO dose would be immediately dropped to 500 units if the hemoglobin level exceeded 14 g/dL and restarted with a 25% dose reduction once the hemoglobin level fell below 12.5 g/dL. The protocol was developed by Satellite Dialysis, was utilized at all facilities and was a written policy administered by the nurses.

Definitions

Optimal anemia management in the ESRD population for the purpose of this study was defined as a 3-month average hemoglobin level between 11 and 12.5 g/dL [ 14,15 ]. The use of an averaged hemoglobin level minimizes the inherent and unavoidable longitudinal variability in hemoglobin readings [ 16 ] and was the metric used to monitor anemia management in the ESRD population by the Centers for Medicare and Medicaid Services (CMS) [ 17 ] at the time of the study. C-reactive protein (CRP) levels (by latex immuno-turbidimetry) were used as a measure of systemic inflammation [ 18 ]. Iron deficiency was defined as one of the following: (1) ferritin <200 ng/mL or (2) ferritin <800 ng/mL and saturation <20% [ 19 ]. EPO responsiveness was measured using the EPO index and EPO per unit weight [ 20 ]. The EPO index was calculated by a simple division of the total weekly EPO dose by hematocrit [ 21 ]. The intra-patient standard deviation of hemoglobin was used as a surrogate marker for hemoglobin cycling [ 22 ].

Laboratory data

Monthly pre- and post-dialysis blood samples were obtained immediately before the first session of the week for standard chemistries, liver enzymes, complete blood counts, bone markers and iron studies. All patients also received routine CRP monitoring semi-annually (drawn on September 2005, coinciding with the data period). Formal kinetic modeling and nPNA were used to quantitate dialysis adequacy and nutritional status, respectively. All laboratory measurements were performed by Satellite Laboratories (Redwood City, CA, USA).

Statistical methods

Demographic and laboratory values were summarized as mean ± standard deviation. Univariate analysis was performed to compare characteristics among patients with a 3-month hemoglobin level within and outside the target interval of 11.0–12.5 g/dL. ANOVA and Chi-square analyses were used to detect significant differences among continuous and categorical variables, respectively. A post hoc pairwise Scheffé analysis was performed on all continuous predictor variables with a statistically significant ANOVA outcome ( P  < 0.05). Distributions for all continuous predictor variables were plotted and examined to confirm the assumption of normality. Further analysis included a univariate analysis to compare characteristics of patients above, within and below the target interval of 11.0–12.5 g/dL.

Two parallel multivariate logistic regression analyses using backward elimination techniques (α = 0.05) were performed to identify correlates of suboptimal anemia management outcomes. The first analysis identified significant factors for a 3-month hemoglobin level that were below versus within the interval of 11.0–12.5 g/dL, excluding those above range. A second analysis identified significant factors for a 3-month hemoglobin level above versus within the interval of 11.0–12.5 g/dL, excluding patients below range. Results were reported as odds ratios with 95% confidence intervals.

To determine whether assignment to an EPO or iron protocol confounded a facility's effect on achieving the target hemoglobin level, the above logistic regression analysis was repeated with only facility and significant patient characteristics [iron deficiency, parathyroid hormone (PTH), CRP and gender] in the model. We followed this by another logistic regression analysis to re-determine the significance of facility on achieving the target hemoglobin level, this time adjusted for patient characteristics and assignment to EPO/iron protocol. To determine whether confounding was occurring, we then compared the significance of facility both before and after the addition of protocol assignment to the logistic regression model.

We used linear regression to determine the effect of facility on hemoglobin, intra-patient standard deviation, EPO index, weekly EPO dose and weekly EPO dose per unit weight with adjustment for covariates found to be statistically significant in the previous backward elimination logistic regression analysis. A Pearson's correlation coefficient was used to determine the linearity between the above measures and the facility performance, defined as the expected probability of achieving the target hemoglobin level at the specific facility, after adjustment for covariates. All statistical analysis was done using the SAS statistical software package (version 9.1). Institutional Review Boards at both the Stanford University School of Medicine and the Satellite Healthcare where the analysis was conducted approved the study.

Results

Anemia management

From August 2005 to September 2005, the 1499 patients who qualified for this study had a 3-month average hemoglobin of 12 ± 1.1 g/dL. Only 58.4% of the population had a 3-month hemoglobin level within the recommended target of 11–12.5 g/dL; 13.4% of the population had an average hemoglobin level of <11 g/dL (below target) and 28.2% had an average hemoglobin level >12.5 g/dL (above target).

The majority of EPO (85.9%) and IV iron doses (86.9%) during the study period were administered to patients who were assigned to a nurse-administered anemia management protocol. Patients who were assigned to the anemia management protocol were more often within the target range than those who had a portion of their EPO or IV iron titration managed off protocol (see Figure 1 ).

Fig. 1

Hemoglobin outcomes by method of anemia management. Note: group assignment was determined by an individual's 3-month average hemoglobin being above, below or within the target of 11–12.5 g/dL. P < 0.0001 for chi-square analysis between patients who were assigned to the anemia management protocol versus those patients with some portion of their anemia managed off protocol.

Fig. 1

Hemoglobin outcomes by method of anemia management. Note: group assignment was determined by an individual's 3-month average hemoglobin being above, below or within the target of 11–12.5 g/dL. P < 0.0001 for chi-square analysis between patients who were assigned to the anemia management protocol versus those patients with some portion of their anemia managed off protocol.

Univariate analysis

Several relevant demographic and laboratory factors differed according to the average hemoglobin group ( Table 1 ). Dialysis adequacy and other demographic and lab values that were not significant in univariate models are not shown. Compared to patients within the target range, patients outside the target range demonstrated a trend toward higher CRP levels, but significantly had lower use of the anemia protocol, were more likely to be male and there were highly significant facility differences (see Table 2 ). The outside the target range patients were divided into either above target (>12.5 g/dL) or below target (<11.0 g/dL) groups for further analysis. Patients below target were characterized as relatively iron deficient, inflamed (low albumin or elevated CRP levels) and had an elevated PTH when compared with patients within the target hemoglobin interval. Lower doses of recombinant EPO were administered to patients with above range hemoglobin levels. Patients with a 3-month average hemoglobin level within the target range had a significantly higher proportion of their EPO managed with a standardized protocol, whereas patients in the below target group were significantly less likely to receive their iron on protocol.

Table 1

Mean (standard error) of patients by average hemoglobin level

Factor Below group Within target group Above group P -value  
N 201 876 422 N/A 
Hemoglobin (g/dL) 10.4 (0.05) 11.9 (0.01) 13.2 (0.03) <0.0001 
Assignment to EPO protocol (%) 78.5 (2.14) 91.6 (0.68) 85.5 (1.29) 0.0018 
Assignment to iron protocol (%) 81.3 (2.37) 90.5 (0.80) 91.6 (1.06) <0.0001 
EPO per HD session (units) 9785 (646) 5040 (167) 3646 (182) <0.0001 
Venofer over 3 months (mg) 741 (40) 794 (17) 949 (28) <0.0001 
Albumin (g/dL) 3.7 (0.03) 3.8 (0.01) 3.8 (0.01) <0.0001 
C-reactive protein (mg/L) 15.3 (2.1) 8.5 (0.7) 6.1 (0.7) <0.0001 
Total cholesterol (mg/dL) 131 (2.5) 141 (1.2) 143 (1.7) <0.0001 
Ferritin (ng/mL) 668 (23) 631 (7.9) 601 (11.5) <0.0001 
Percent iron deficiency 39 (3) 31 (1) 27 (1) <0.0001 
Transferrin saturation (%) 23 (0.8) 25 (0.3) 28 (0.4) <0.0001 
Transferrin (ug/dL) 171 (2.3) 175 (0.9) 179 (1.3) 0.0004 
Parathyroid hormone (pg/mL) 461 (38) 443 (12) 364 (13) <0.0001 
Gender (percent male) 53.3 53.1 60.7 0.0238 
Age (years) 63(1.1) 64 (0.5) 65 (0.7) 0.1747 
Days on hemodialysis 1245 (77) 1233 (42) 1066 (62) 0.6861 
Calcium (mg/dL) 8.8 (0.05) 8.8 (0.02) 8.8 (0.03) 0.2126 
Phosphorus (mg/dL) 5.4 (0.10) 5.3 (0.04) 5.4 (0.06) 0.8685 
Kt/V with Kru 1.7 (0.03) 1.7 (0.02) 1.7 (0.02) 0.1063 
Mean corpuscular volume (fL) 93 (0.6) 94 (0.2) 93 (0.3) 0.0388 
Post-dialysis weight (kg) 72.8 (1.4) 74.1 (0.7) 73.0 (0.9) 0.4222 
nPNA (g/kg/day) 0.93 (0.02) 0.97 (0.01) 0.97 (0.01) 0.2949 
Factor Below group Within target group Above group P -value  
N 201 876 422 N/A 
Hemoglobin (g/dL) 10.4 (0.05) 11.9 (0.01) 13.2 (0.03) <0.0001 
Assignment to EPO protocol (%) 78.5 (2.14) 91.6 (0.68) 85.5 (1.29) 0.0018 
Assignment to iron protocol (%) 81.3 (2.37) 90.5 (0.80) 91.6 (1.06) <0.0001 
EPO per HD session (units) 9785 (646) 5040 (167) 3646 (182) <0.0001 
Venofer over 3 months (mg) 741 (40) 794 (17) 949 (28) <0.0001 
Albumin (g/dL) 3.7 (0.03) 3.8 (0.01) 3.8 (0.01) <0.0001 
C-reactive protein (mg/L) 15.3 (2.1) 8.5 (0.7) 6.1 (0.7) <0.0001 
Total cholesterol (mg/dL) 131 (2.5) 141 (1.2) 143 (1.7) <0.0001 
Ferritin (ng/mL) 668 (23) 631 (7.9) 601 (11.5) <0.0001 
Percent iron deficiency 39 (3) 31 (1) 27 (1) <0.0001 
Transferrin saturation (%) 23 (0.8) 25 (0.3) 28 (0.4) <0.0001 
Transferrin (ug/dL) 171 (2.3) 175 (0.9) 179 (1.3) 0.0004 
Parathyroid hormone (pg/mL) 461 (38) 443 (12) 364 (13) <0.0001 
Gender (percent male) 53.3 53.1 60.7 0.0238 
Age (years) 63(1.1) 64 (0.5) 65 (0.7) 0.1747 
Days on hemodialysis 1245 (77) 1233 (42) 1066 (62) 0.6861 
Calcium (mg/dL) 8.8 (0.05) 8.8 (0.02) 8.8 (0.03) 0.2126 
Phosphorus (mg/dL) 5.4 (0.10) 5.3 (0.04) 5.4 (0.06) 0.8685 
Kt/V with Kru 1.7 (0.03) 1.7 (0.02) 1.7 (0.02) 0.1063 
Mean corpuscular volume (fL) 93 (0.6) 94 (0.2) 93 (0.3) 0.0388 
Post-dialysis weight (kg) 72.8 (1.4) 74.1 (0.7) 73.0 (0.9) 0.4222 
nPNA (g/kg/day) 0.93 (0.02) 0.97 (0.01) 0.97 (0.01) 0.2949 

Results displayed as mean (SE). Group assignment was determined by an individual's 3-month average hemoglobin being above, below or within the target of 11–12.5 g/dL. P -value calculations are based on a one-way analysis of variance. Values in bold represent statistically significant findings via post hoc pairwise Scheffe comparison to the group who achieved a target hemoglobin level ( P < 0.05). EPO, erythropoietin; HD, hemodialysis.

Table 2

Odds ratios relating selected characteristics to likelihood of out of target range hemoglobin values

Variable Odd ratio for out of target versus in target hemoglobin value 
CRP (mg/L increase) 1.01 (1.00, 1.01) 
PTH (100 pg/ml increase) 1.01 (0.98, 1.03) 
Iron deficiency (10% increase) 1.01 (0.98, 1.04) 
Assignment to iron protocol (10% increase) 0.98 (0.94, 1.01) 
Assignment to EPO protocol (10% increase) 0.93 (0.89, 0.97) 
Gender (female versus male) 0.79 (0.64, 0.97) 
Facility (15 sites)  0.49–1.63 ( P = 0.005)  
Variable Odd ratio for out of target versus in target hemoglobin value 
CRP (mg/L increase) 1.01 (1.00, 1.01) 
PTH (100 pg/ml increase) 1.01 (0.98, 1.03) 
Iron deficiency (10% increase) 1.01 (0.98, 1.04) 
Assignment to iron protocol (10% increase) 0.98 (0.94, 1.01) 
Assignment to EPO protocol (10% increase) 0.93 (0.89, 0.97) 
Gender (female versus male) 0.79 (0.64, 0.97) 
Facility (15 sites)  0.49–1.63 ( P = 0.005)  

Variables included: facility name, gender, percent on iron protocol, percent on EPO protocol, age, CRP, PTH, MCV, spKTV, wKrU, nPNA, days onHD and percent iron deficient.

The backward elimination removed variables with P > 0.05.

Multivariate analysis

In a multivariate logistic regression analysis ( Table 3 ), increased CRP levels, increasing PTH levels, iron deficiency and decreasing assignment to protocol-driven iron dosing were significantly associated with below target hemoglobin levels. In contrast, only male gender and decreasing PTH levels were clinical factors associated with above target hemoglobin levels. Facility and the use of a standardized protocol to titrate EPO were also significant predictors of achieving the target hemoglobin level.

Table 3

Odds ratios relating selected characteristics to likelihood of low and high hemoglobin levels

Variable Odds ratio for below group Odds ratio for above group 
CRP (mg/L increase) 1.01 (1.01, 1.02) – 
PTH (100 pg/mL increase) 1.07 (1.03, 1.11) 0.96 (0.92, 0.99) 
Iron deficiency (10% increase) 1.07 (1.03, 1.12) – 
Assignment to iron protocol (10% increase) 0.93 (0.88, 0.97) – 
Assignment to EPO protocol (10% increase) 0.89 (0.84, 0.95) 0.94 (0.90, 0.98) 
Gender (female versus male) – 0.68 (0.53, 0.86) 
Facility (15 sites)  0.26–2.59 ( P = 0.0035)   0.58–3.41 ( P = 0.0002)  
Variable Odds ratio for below group Odds ratio for above group 
CRP (mg/L increase) 1.01 (1.01, 1.02) – 
PTH (100 pg/mL increase) 1.07 (1.03, 1.11) 0.96 (0.92, 0.99) 
Iron deficiency (10% increase) 1.07 (1.03, 1.12) – 
Assignment to iron protocol (10% increase) 0.93 (0.88, 0.97) – 
Assignment to EPO protocol (10% increase) 0.89 (0.84, 0.95) 0.94 (0.90, 0.98) 
Gender (female versus male) – 0.68 (0.53, 0.86) 
Facility (15 sites)  0.26–2.59 ( P = 0.0035)   0.58–3.41 ( P = 0.0002)  

Results displayed as odds ratio (95% confidence interval). Hosmer and Lemeshow goodness-of-fit P = 0.4987 and P = 0.5061 for below and above group analysis, respectively. CRP, C-reactive protein; PTH, parathyroid hormone; EPO, erythropoietin.

Facility

The dialysis facility remained significantly and strongly associated with achieving the target hemoglobin level, even after statistical adjustment for patient characteristics and for assignment to the nurse-administered EPO and iron protocol (see Table 4 ). This suggests that facility is independent of patient characteristics and protocol assignment in predicting achievement of the targeted hemoglobin level. Figure 2 demonstrates the persistent large inter-facility variation in performance despite statistical adjustment. The 15 units studied were all in Northern California. They varied in geography from city to suburban and rural settings. The number of patients per facility ranged from 30 to 276 patients, with 1 nurse per 12 patients and 1 tech per 4 patients as well as multiple physicians responsible for patient care at each site (range 4–50 physicians/facility). While there were many significant differences between facilities in patient and treatment characteristics, they had homogeneous standards for processes of care through the implementation of the same algorithms to manage anemia, bone disease, nutrition, reuse and dialysis adequacy. The only significant correlation with facility anemia performance was in CRP (ranging from an average of 4.4–13.5 mg/dL, with r = 0.61, P < 0.01) and in percent of patients on EPO protocol (ranging from 72% to 93%, with r = 0.50, P < 0.01). See supplemental Table B for further description of the patients at each of the facilities.

Fig. 2

Anemia management outcomes adjusted by patient characteristics and protocol assignment by facility. Note: results adjusted for PTH, CRP, gender, iron deficiency and full assignment to EPO/iron protocol.

Fig. 2

Anemia management outcomes adjusted by patient characteristics and protocol assignment by facility. Note: results adjusted for PTH, CRP, gender, iron deficiency and full assignment to EPO/iron protocol.

Table 4

Significance of facility on anemia outcomes with and without adjustment for assignment to protocol

 Covariates in model  Facility ( P -value)  
Risk for low hemoglobin Patient characteristics 0.0001 
 Patient characteristics, assignment to EPO and iron protocol 0.0002 
Risk for high hemoglobin Patient characteristics <0.0001 
 Patient characteristics, assignment to EPO and iron protocol <0.0001 
 Covariates in model  Facility ( P -value)  
Risk for low hemoglobin Patient characteristics 0.0001 
 Patient characteristics, assignment to EPO and iron protocol 0.0002 
Risk for high hemoglobin Patient characteristics <0.0001 
 Patient characteristics, assignment to EPO and iron protocol <0.0001 

Multivariate logistic regression analysis. Outcome was the probability of a 3-month hemoglobin level below versus within the target interval of 11–12.5 g/dL and above versus within the target interval. Results displayed as Wald statistic for the predictor variable facility, adjusted for patient characteristics and assignment to protocol.

EPO doses and facility

Facilities that had fewer patients within target tended to have a greater proportion of their patients above target, with higher than average hemoglobin levels than other facilities (Figures 2 and 3 ). Despite this, the EPO doses were significantly lower in these units (Figure 4 ). EPO responsiveness, as measured by the EPO index and EPO per unit weight, was strongly correlated with facility assignment (Figure 5 ). Thus, the facilities with a lower proportion of patients within the target hemoglobin interval maintained higher hemoglobin levels, required less EPO and were more EPO responsive. No correlation could be seen between patient hemoglobin variability and facility (Figure 3 ).

Fig. 3

Hemoglobin and intra-patient standard deviation by facility rank. r = −0.60 ( P = 0.02) for hemoglobin. r = −0.09 ( P = 0.75) for intra-patient standard deviation of hemoglobin.

Fig. 3

Hemoglobin and intra-patient standard deviation by facility rank. r = −0.60 ( P = 0.02) for hemoglobin. r = −0.09 ( P = 0.75) for intra-patient standard deviation of hemoglobin.

Fig. 4

Erythropoietin dose by facility performance. r = 0.73 ( P = 0.002).

Fig. 4

Erythropoietin dose by facility performance. r = 0.73 ( P = 0.002).

Fig. 5

EPO index and EPO by unit weight by facility rank. r = 0.71 ( P = 0.003) for EPO index. r = 0.76 ( P = 0.001) for EPO per unit weight.

Fig. 5

EPO index and EPO by unit weight by facility rank. r = 0.71 ( P = 0.003) for EPO index. r = 0.76 ( P = 0.001) for EPO per unit weight.

Discussion

Anemia management is a key component in the care of patients with ESRD. Based on their exhaustive review of the present literature, maintaining the hemoglobin between 11 and 12 g/dL has been advocated by consensus groups in order to improve quality of life, limit hospitalizations and, perhaps, reduce cardiovascular mortality [ 23,24 ]. Given the inherent challenges of achieving a narrow anemia management target , providers had targeted a 3-month average hemoglobin level of 11.0–12.5 g/dL in practice, which had been endorsed by CMS at the time of the study data collection. We confirm the findings of prior studies that many ESRD patients do not achieve the fourth goal. For the first time, we analyzed factors within a single dialysis chain to understand the factors that influence the ability to achieve a predetermined target range. These findings should be of relevance regardless of where target hemoglobin levels are set in the future.

Nearly half of the patients in this study were outside of the target hemoglobin range, despite the utilization of a system-wide EPO and iron management policy. Indeed, the effort to correct the anemia in an appropriate fashion was evident by the significantly higher EPO dosing in the below target group compared with the target group as well as the significantly lower doses given to patients in the above target group. Univariate analysis suggested interplay between multiple patient characteristics and laboratory values and the ability to achieve target hemoglobin levels. However, multivariate analysis limited these relationships to show that only inflammation, hyperparathyroidism and iron deficiency were significant, yet relatively minor, clinical predictors of hemoglobin values below the target range. Patients with levels above the target range were characterized only by relatively low PTH levels and male gender. These findings are consistent with other studies [ 25 ], which have related inflammation to adverse patient outcomes including complicated anemia [ 26 ], hospitalization [ 27 ] and death [ 28 ]. However, while the above relationships were highly significant, the overall impact on the odds ratio was small, suggesting that these laboratory predictors explained relatively little of the variation in hemoglobin level.

In contrast, the manner in which patients were managed was a more powerful predictor of achieving hemoglobin levels within the target range. Off-protocol adjustments by physicians were associated with a lower chance of achieving target values. Furthermore, the dialysis unit in which the patient was treated was strongly associated with the achievement of target hemoglobin values independent of the individual facility's usage of the anemia management protocol. Unlike a prior study examining the influence of facility on hemoglobin [ 29 ], this study was completed in a chain of units controlling for profit versus nonprofit status and also utilizing the same chain-wide protocol for management of dialysis dose, comorbidities and anemia care. Why such large inter-facility differences occur is uncertain, as this relationship remains after adjustment for all significant demographic variables, laboratory variables and for the assignment to the anemia protocol.

To further explore this finding, we examined the patterns of EPO responsiveness by evaluating EPO dosing, EPO index and EPO by weight at the facility level. These measures of EPO responsiveness have been suggested to be surrogates of patient health, EPO responsiveness [ 21 ], inflammation [ 21 ] and prognosis [ 8 , 22 , 30 ]. All facilities appeared to have relatively few patients below target; however, units that did not perform as well in achieving the target hemoglobin level were characterized as having more patients above the target level. These facilities used less EPO and had patients with lower EPO indices, implying that higher hemoglobin values were not a result of excessive EPO and iron dosing, but rather the EPO responsiveness of the unit's patient population. While standard anemia protocols appear to appropriately decrease EPO doses in response to high hemoglobin values, our findings suggest that current protocols are likely better suited to patient populations who are relatively EPO resistant, perhaps with comorbidities that are not adequately accounted for by available markers. This may also be the reason that better performing facilities (in terms of achieving target) had more patients off EPO protocol and a higher average CRP level. Their patients received more EPO resulting in lower mean hemoglobin values but avoided below target levels. The hyporesponsiveness of this population resulted in fewer patients above target with the net result that these facilities had more patients in the target range . The lack of association between intra-patient standard deviation and facility rank suggested no relationship between the proportion of patients achieving target and hemoglobin cycling.

Taken altogether, these findings suggest that there remains a significant need to improve anemia management in ESRD patients. It is possible that reductions in inflammation and hyperparathyroid bone disease may increase the number of patients who achieved the target. Alternatively, measures of CRP and PTH could be adopted into EPO dose titration. However, the present study suggests that the most important determinant in achieving target hemoglobin may be at the dialysis facility level. Attention to the maintenance of EPO and iron protocols, including refinement of these protocols, is a likely necessary first step to improve hemoglobin levels in dialysis patients [ 31 ]. While we could not account for differences in EPO responsiveness by typical patient characteristics, it is possible that facility-specific geographic differences in EPO responsiveness partially underlie the facility effect. A geographic effect on hemoglobin levels has been previously suggested across widely separated regions of the USA [ 32 ], but this is the first study to suggest that this effect may exist even among facilities only miles apart. Additional facility-specific issues appear to contribute to the risk of not achieving target hemoglobin levels, but unlike other studies, this is the first to demonstrate differences among facilities that share nonprofit status and common patient care protocols [ 29 , 33 ]. Further specific issues contributing to the facility effect could not be identified in this study but could include differences in patient populations that could not be balanced by multivariate adjustment, unspecified covariates that effect anemia management outcomes, undetected differences in compliance with the EPO or iron protocols, varying hospitalization rates, use of transfusions, socio-economic status as well as differences in the technical care of the patients, such as access care and infiltrations.

This study was limited by the relatively small number of patients who were studied for a limited amount of time. The population was selected based on the access to data and may not be representative of the US population. Further, the study was observational and is unable to distinguish cause from association. For example, the fact that patients treated off protocol were more likely to be outside of the target range does not necessarily imply that the protocol is more effective. It is possible that these patients are managed outside the protocol because of difficulties achieving the target. Finally, the database does not provide a mean to further evaluate some of the specific issues mentioned above to determine why additional differences remain among facilities after adjustment for patient demographics, measured care processes and EPO index.

Nonetheless, it would appear to be prudent to address differences in anemia care among facilities providing dialysis, and standardize and prioritize methods to achieve the desired hemoglobin levels in more patients.

The authors would like to graciously thank Dr Allen R. Nissenson, Dr Kristin L. Cobb and Isabella Taylor for their contributions toward this project. Dr Nissenson's ongoing advice was a critical component to the study's success. The statistical advice provided by Dr Cobb was invaluable in the analysis of the dataset. Finally, the study could not have succeeded without Ms Taylor's timely administrative support. The authors would also like to acknowledge Satellite Healthcare and the Kidney Foundation of Canada for their support.

Conflict of interest statement. Dr Chan's salary was partially funded by a research fellowship from Amgen. The fellowship did not have any restrictions on his research. The results presented in this paper have not been published previously in whole or part. Some of the findings in this manuscript were presented as an abstract at ASN Renal Week, San Diego, CA, USA, in November 2006.

Supplementary data

Supplementary data is available at NDT Journal online.

References

1
Adamson
JW
Anemia of end-stage renal disease
Kidney Int
 , 
1985
, vol. 
28
 (pg. 
1
-
5
)
2
Singh
AK
Szczech
L
Tang
KL
, et al.  . 
Correction of anemia with epoetin alfa in chronic kidney disease
N Engl J Med
 , 
2006
, vol. 
355
 (pg. 
2085
-
2098
)
3
Drüeke
TB
Locatelli
F
Clyne
N
, et al.  . 
Normalization of hemoglobin level in patients with chronic kidney disease and anemia
N Engl J Med
 , 
2006
, vol. 
355
 (pg. 
2071
-
2084
)
4
Collins
AJ
Brenner
RM
Ofman
JJ
, et al.  . 
Epoetin alfa use in patients with ESRD. An analysis of recent US prescribing patterns and hemoglobin outcomes
Am J Kidney Dis
 , 
2005
, vol. 
46
 (pg. 
481
-
488
)
5
National Kidney Foundation NKF-K/DOQI
Clinical practice guidelines for anemia of chronic kidney disease: update 2000
Am J Kidney Dis
 , 
2001
, vol. 
37
 
Suppl 1
(pg. 
S182
-
S238
)
6
Macdougall
IC
Coles
GA
Williams
JD
, et al.  . 
Long-term cardiorespiratory effects of amelioration of renal anemia by erythropoietin
Lancet
 , 
1990
, vol. 
335
 (pg. 
489
-
493
)
7
Evans
RW
Rader
B
Manninen
DL
Cooperative Multicenter EPO Clinical Trial Group. The quality of life of hemodialysis recipients treated with recombinant human erythropoietin
JAMA
 , 
1990
, vol. 
63
 (pg. 
825
-
820
)
8
Ofsthun
N
Labrecque
J
Lacson
E
The effects of higher hemoglobin levels on mortality and hospitalization in hemodialysis patients
Kidney Int
 , 
2003
, vol. 
63
 (pg. 
1908
-
1914
)
9
Lowrie
EG
Huang
WH
Lew
NL
, et al.  . 
Friedman
EA
The relative contribution of measured variables to death risk among hemodialysis patients?
Death on Hemodialysis. Preventable or Inevitable
 , 
1994
Boston, MA
Kluwer
(pg. 
121
-
141
)
10
Regidor
DL
Kopple
JD
Kovesdy
CP
, et al.  . 
Associations between changes in hemoglobin and administered erythropoiesis-stimulating agent and survival in hemodialysis patients
J Am Soc Nephrol
 , 
2006
, vol. 
17
 (pg. 
1181
-
1191
)
11
Phrommintikul
A
Haas
SJ
Maros
E
, et al.  . 
Mortality and target hemoglobin concentration in anemic patients with chronic kidney disease treated with erythropoietin: a meta analysis
Lancet
 , 
2007
, vol. 
369
 (pg. 
381
-
388
)
12
Besarab
A
Bolton
WK
Browne
JK
, et al.  . 
The effects of normal as compared with low hematocrit values in patients with cardiac disease who are receiving hemodialysis and epoetin
N Engl J Med
 , 
1998
, vol. 
339
 (pg. 
584
-
590
)
13
Fink
JC
Hsu
VD
Zhan
M
, et al.  . 
Center effects in anemia management of dialysis patients
J Am Soc Nephrol
 , 
2007
, vol. 
18
 (pg. 
646
-
653
)
14
Bems
J
Elzein
H
Lynn
R
, et al.  . 
Hemoglobin variability in epoetin-treated hemodialysis patients
Kidney Int
 , 
2003
, vol. 
64
 (pg. 
1514
-
1521
)
15
Improving the quality of care for beneficiaries with end-stage renal disease
The Medicare Payment Advisory Commission
 , 
1999
 
16
Lacson
E
Ofsthun
N
Lazarus
JM
Effect of variability in anemia management on hemoglobin outcomes in ESRD
Am J Kidney Dis
 , 
2003
, vol. 
41
 (pg. 
111
-
124
)
17
Messana
T
Erythropoietin claims monitoring policy. Implications for clinical practice
Nephrol Nurs J
 , 
2006
, vol. 
33
 (pg. 
209
-
222
)
18
Kalantar-Zadeh
K
McAllister
CJ
Lehn
RS
, et al.  . 
Effect of malnutrition-inflammation complex on EPO hyporesponsiveness in maintenance hemodialysis patients
Am J Kidney Dis
 , 
2003
, vol. 
42
 (pg. 
761
-
773
)
19
NKF-K/DOQI
Clinical practice guidelines for anemia of chronic disease
 
20
NKF-DOQI
Clinical practice guidelines for anemia of chronic renal failure: IV. Administration of epoetin
Am J Kidney Dis
 , 
2001
, vol. 
37
 
Suppl 1
pg. 
S207
 
21
Gunnell
J
Yeun
JY
Depner
TA
, et al.  . 
Acute-phase response predicts erythropoietin resistance in hemodialysis and peritoneal dialysis patients
J Am Soc Nephrol
 , 
1999
, vol. 
33
 (pg. 
63
-
71
)
22
Berns
J
Elzein
H
Lynn
R
, et al.  . 
Hemoglobin variability in epoetin-treated hemodialysis patients
Kidney Int
 , 
2003
, vol. 
64
 (pg. 
1514
-
1521
)
23
National Kidney Foundation. KDOQI
Clinical practice guideline and clinical practice recommendations for anemia in chronic kidney disease: 2007 update of hemoglobin target
Am J Kidney Dis
 , 
2007
, vol. 
50
 (pg. 
471
-
530
)
24
National Kidney Foundation. K/DOQI
Clinical practice guidelines for chronic kidney disease. Evaluation, classification, and stratification
Am J Kidney Dis
 , 
2002
, vol. 
39
 
Suppl 1
(pg. 
S1
-
S266
)
25
Drueker
T
Hyporesponsiveness to recombinant human erythropoietin
Nephrol Dial Transplant
 , 
2001
, vol. 
16
 
Suppl 7
(pg. 
25
-
28
)
26
Gunnell
J
Yeun
JY
Depner
TA
, et al.  . 
Acute-phase response predicts erythropoietin resistance in hemodialysis and peritoneal dialysis patients
Am J Kidney Dis
 , 
1999
, vol. 
33
 (pg. 
63
-
71
)
27
Kalantar-Zadeh
K
Kopple
JD
Block
G
, et al.  . 
A malnutrition-inflammation score is correlated with morbidity and mortality in maintenance hemodialysis patients
Am J Kidney Dis
 , 
2001
, vol. 
38
 (pg. 
1251
-
1263
)
28
Yeun
JY
Kaysen
GA
C-reactive protein predicts all-cause and cardiovascular mortality in hemodialysis patients
Am J Kidney Dis
 , 
2000
, vol. 
35
 (pg. 
469
-
476
)
29
Fink
JC
Hsu
VD
Zhan
M
, et al.  . 
Center effects in anemia management of dialysis patients
J Am Soc Nephrol
 , 
2007
, vol. 
18
 (pg. 
646
-
653
)
30
Kaysen
GA
Muller
HG
Ding
J
, et al.  . 
Challenging the validity of EPO index
Am J Kidney Dis
 , 
2006
, vol. 
47
 pg. 
166
 
31
Brimble
KS
Rabbat
CG
McKenna
P
, et al.  . 
Protocolized anemia management with erythropoietin in hemodialysis patients. A randomized controlled trial
J Am Soc Nephrol
 , 
2003
, vol. 
14
 (pg. 
2654
-
2661
)
32
Coladonato
JA
Frankenfield
DL
Reddan
DN
, et al.  . 
Trends in anemia management among US hemodialysis patients
J Am Soc Nephrol
 , 
2002
, vol. 
13
 (pg. 
1288
-
1295
)
33
Reddan
DN
Frankenfield
DL
Klassen
PS
, et al.  . 
Regional variability in anaemia management and haemoglobin in the US
Nephrol Dial Transplant
 , 
2003
, vol. 
18
 (pg. 
147
-
152
)

Author notes

*
Kevin E. Chan and Richard A. Lafayette served as co-first authors on this manuscript.

Comments

0 Comments