Anemia management and outcomes from 12 countries in the dialysis outcomes and practice patterns study (DOPPS)

Dialysis Therapies

Anemia Management and Outcomes From 12 Countries in the Dialysis Outcomes and Practice Patterns Study (DOPPS) Ronald L. Pisoni, PhD, Jennifer L. Bragg-Gresham, MS, Eric W. Young, MD, Tadao Akizawa, MD, PhD, Yasushi Asano, MD, PhD, Francesco Locatelli, MD, Juergen Bommer, MD, Jose Miguel Cruz, MD, Peter G. Kerr, MD, David C. Mendelssohn, MD, Philip J. Held, PhD, and Friedrich K. Port, MD, MS ● Background: Anemia is common in hemodialysis (HD) patients. Methods: Data collected from nationally representative samples of HD patients (n ⴝ 11,041) in 2002 to 2003 were used to describe current anemia management for long-term HD patients at 309 dialysis units in 12 countries. Analyses of associations and outcomes were adjusted for demographics, 15 comorbid classes, laboratory values, country, and facility clustering. Results: For patients on dialysis therapy for longer than 180 days, 23% to 77% had a hemoglobin (Hgb) concentration less than 11 g/dL (<110 g/L), depending on country; 83% to 94% were administered erythropoietin (EPO). Mean Hgb levels were 12 g/dL (120 g/L) in Sweden; 11.6 to 11.7 g/dL (116 to 117 g/L) in the United States, Spain, Belgium, and Canada; 11.1 to 11.5 g/dL (111 to 115 g/L) in Australia/New Zealand, Germany, Italy, the United Kingdom, and France; and 10.1 g/dL (101 g/L) in Japan. Hgb levels were substantially lower for new patients with end-stage renal disease, and EPO use before ESRD ranged from 27% (United States) to 65% (Sweden). By patient, EPO use significantly declined with greater Hgb concentration (adjusted odds ratio, 0.61 per 1-g/dL [10-g/L] greater Hgb level; P < 0.0001), as did EPO dosage. Case-mix–adjusted mortality and hospitalization risk declined by 5% and 6% per 1-g/dL greater patient baseline Hgb level (P < 0.003 each), respectively. Furthermore, patient mortality and hospitalization risks were 10% to 12% lower for every 1-g/dL greater facility mean Hgb level. Patients were significantly more likely to have Hgb levels of 11 g/dL or greater (>110 g/L) if they were older; were men; had polycystic kidney disease; had greater albumin, transferrin saturation, or calcium levels; were not dialyzing with a catheter; or had lower ferritin levels. Facilities with greater intravenous iron use showed significantly greater facility mean Hgb concentrations. Mean EPO dose varied from 5,297 (Japan) to 17,360 U/wk (United States). Greater country mean EPO doses were significantly associated with greater country mean Hgb concentrations. Several patient characteristics were associated with greater EPO doses. Even in some countries with high intravenous iron use, 35% to 40% of patients had a transferrin saturation less than 20% (below guidelines). Conclusion: These findings indicate large international variations in anemia management, with significant improvements during the last 5 years, although many patients remain below current anemia guidelines, suggesting large and specific opportunities for improvement. Am J Kidney Dis 44:94-111. © 2004 by the National Kidney Foundation, Inc. INDEX WORDS: Anemia; hemodialysis (HD); hemoglobin (Hgb); erythropoietin (EPO); iron; Dialysis Outcomes and Practice Patterns Study (DOPPS).

A

NEMIA MANAGEMENT is an important aspect of care for hemodialysis (HD) patients because anemia is one of the common consequences of chronic kidney disease. A major advance in anemia management of patients with chronic kidney disease was achieved with the

introduction of recombinant human erythropoietin (EPO) in 1989. Using national data from the US Renal Data System, we showed that the requirement for blood transfusions decreased markedly, from 16% of HD patients administered at least 1 transfusion per quarter in early

From the University Renal Research and Education Association; Division of Nephrology, Veterans Affairs Medical Center and University of Michigan, Ann Arbor, MI; Center of Blood Purification Therapy, Wakayama Medical University, Wakayama; Department of Internal Medicine, Sashima Redcross Hospital, Ibaragi, Japan; Division of Nephrology and Dialysis, A. Manzoni Hospital, Lecco, Italy; Nephrology Section, University of Heidelberg, Germany; Nephrology Service, Hospital General Universitario La Fe, Valencia, Spain; Nephrology Department, Monash Medical Centre, Melbourne, Victoria, Australia; and Division of Nephrology, Humber River Hospital and University of Toronto, Weston, Ontario, Canada.

Received January 15, 2004; accepted in revised form March 11, 2004. The Dialysis Outcomes and Practice Patterns Study is supported by research grants from Amgen Inc and Kirin Brewery Ltd without restrictions on publications. Address reprint requests to Ronald L. Pisoni, PhD, URREA, 315 W Huron, Ste 260, Ann Arbor, MI 48103. E-mail: [email protected] © 2004 by the National Kidney Foundation, Inc. 0272-6386/04/4401-0003$30.00/0 doi:10.1053/j.ajkd.2004.03.023

94

American Journal of Kidney Diseases, Vol 44, No 1 (July), 2004: pp 94-111

ANEMIA MANAGEMENT AND OUTCOMES IN 12 COUNTRIES

1989 to 3.3% in 1992.1 During the same period, the introduction of EPO increased from 0% to 88% of HD patients, at which level it subsequently stabilized. Moreover, mean hemoglobin (Hgb) level has increased markedly for HD patients in the United States during the last 13 years, from a mean of approximately 9 g/dL (90 g/L; hematocrit, 27%) in 19891 to 11.7 g/dL (117 g/L) in 2002 to 2003, as indicated in the present study. In addition to the use of EPO, the development of safer intravenous (IV) iron preparations has allowed practitioners to maintain greater iron stores for HD patients for achieving gains in anemia control for this patient population. The management of anemia in HD patients requires attention to a variety of factors, including EPO dosing, inflammation, iron deficiency from thrice-weekly blood loss during HD or blood loss caused by other conditions, adequacy of dialysis, and shortened red blood cell life time, among others. Numerous benefits have been associated with good anemia control in HD patients, including lower mortality and hospitalization risks,2-6 reduced occurrence of left ventricular hypertrophy,7-9 and improved mental function.10 The EPO-induced increase in hematocrit also has been associated with substantial improvement in most indicators of health-related quality of life. This improvement is of a magnitude observed in the general population between individuals without versus with diabetes mellitus or without versus with chronic back pain.11-17 These benefits have been reviewed extensively in the literature. A large number of clinical trials and prospective observational studies have provided key insights into optimizing the management of anemia for HD patients. These studies have served as the basis for the development of clinical practice guidelines for anemia management by The National Kidney Foundation–Kidney and Dialysis Outcomes Quality Initiative (NKF-K/ DOQI) in 1997,18 the European Best Practice Guidelines (EBPG) on Anemia Management in 1999,19 and the Clinical Practice Guidelines of the Canadian Society of Nephrology (CSN).20 The recommended Hgb target level ranges from these 3 sets of clinical guidelines are greater than 11 g/dL (⬎110 g/L) for the EBPG and CSN and between 11 and 12 g/dL (110 and 120 g/L) for the K/DOQI. Recent studies have shown that the large variability in individual patient Hgb concen-

95

trations during relatively short intervals is important to consider with regard to achieving recommended Hgb target levels.21,22 Few studies have evaluated anemia management internationally.23 The Dialysis Outcomes and Practice Patterns Study (DOPPS) was designed to increase longevity of HD patients by evaluating HD practices and outcomes such as morbidity, mortality, and quality of life.24 The present study evaluated anemia management with EPO and IV iron preparations and the control of anemia over time based on data collected in 7 countries in the DOPPS during 1996 to 2000 and in 12 countries during 2002 to 2003. Additionally, we studied associations between anemia control and mortality or hospitalization while adjusting for country effects and differences in case mix, including comorbid conditions. METHODS

Data Sources Anemia management was assessed by using data from both DOPPS I (data collection, 1996 to 2001) and DOPPS II (data collection, 2002 to present). The DOPPS is a prospective observational study involving adult HD patients randomly selected from 308 representative dialysis facilities in 7 countries (France, Germany, Italy, Japan, Spain, the United Kingdom, and the United States) for DOPPS I and 309 representative dialysis facilities in 12 countries (Australia, Belgium, Canada, France, Germany, Italy, Japan, New Zealand, Spain, Sweden, the United Kingdom, and the United States) for DOPPS II. In DOPPS II, 59 facilities participated from Japan; 20 facilities each from Belgium, Canada, Germany, Italy, Spain, and Sweden; 19 facilities from the United Kingdom; 18 facilities each from Australia and France; 2 facilities from New Zealand; and 74 facilities from the United States. The DOPPS sampling plan and study methods have been described previously.24 Patient information was collected without patient identifiers, and patient consent was obtained as required from local or national ethics committees or institutional review boards. In DOPPS I, US facilities started the study in 1996; European facilities, in 1998; and Japanese facilities, in 1999. Data collection for DOPPS II began in 2002 for all 12 countries. Nationally representative samples were obtained by using a stratified random selection of dialysis units in each country to provide facility samples proportional to the major types of dialysis units and geographic regions within each country. Furthermore, within each dialysis unit, patients were randomly selected for study participation. Detailed patient data were collected at study entry (baseline) and 4-month intervals thereafter. Data were collected from 20 to 40 prevalent HD patients at each facility (dependent on facility size) and up to 15 new patients with end-stage renal disease (ESRD) from each facility when initiating long-term

96

HD therapy. The total number of DOPPS II patients for whom data were available was 11,041.

Statistical Methods Hgb levels and outcomes, DOPPS I. Using DOPPS I data, associations between Hgb levels and both mortality and hospitalization were investigated by using Cox proportional hazards models adjusted for age; sex; race; 15 summary comorbid conditions (coronary artery disease, congestive heart failure, other cardiac disease, hypertension, diabetes mellitus, cerebrovascular disease, peripheral vascular disease, cancer, human immunodeficiency virus/acquired immune deficiency syndrome, lung disease, neurological disorders, psychiatric disease, recurrent cellulitis/gangrene, dyspnea, and gastrointestinal bleeding); single-pool Kt/V; serum concentrations of phosphorus, calcium, and albumin; and country of residence. These models used a robust estimator25 to account for facility clustering. Patients on dialysis therapy for 180 days or less were excluded from several analyses to allow evaluation of a steady-state effect of anemia management after initiation of dialysis therapy.5,26 To evaluate the effect of facility anemia control practice, additional Cox models tested the correlation of dialysis facility mean Hgb concentrations with patient mortality and hospitalization risks. These models accounted for facility clustering effects and were adjusted for the same covariates used in the Cox models based on patient Hgb values. Each facility’s mean Hgb concentration was calculated as the mean baseline Hgb value for all patients in the dialysis unit who had been on dialysis therapy for more than 180 days at the time of study entry. Descriptive statistics by country, DOPPS II. Hgb values and EPO use were investigated by country by using descriptive statistics for patients in DOPPS II. These values were calculated for 2 samples of patients. The first sample examined patients who entered the study within 7 days of their first long-term HD treatment to characterize EPO use and Hgb values for new ESRD patients. The second sample was restricted to patients who at the time of entering the study had been on dialysis therapy for more than 180 days. IV iron use, serum iron and serum ferritin concentrations, and transferrin saturation (TSAT) also were described by country for a prevalent cross-section of patients in DOPPS II, again excluding patients on dialysis therapy for 180 days or less. Calculations of EPO use included all forms of recombinant human EPO currently used for anemia treatment of dialysis patients (EPO alfa, EPO beta, and darbepoetin). However, darbepoetin (ie, Aranesp; Amgen, Inc, Thousand Oaks, CA) was not included in any EPO dosing analyses because of uncertainty regarding the exact conversion factor to use for the equivalence of darbepoetin units with EPO alfa and EPO beta units. Predictors of Hgb level 11 g/dL or greater (ⱖ110 g/L) and of EPO use, DOPPS II. Logistic regression analysis was used to examine associations between patient characteristics and an Hgb value of 11 g/dL or greater (ⱖ110 g/L versus ⬍11 g/dL [⬍110 g/L]) for patients on dialysis therapy for longer than 180 days. These models were adjusted for age; sex; black race; years on dialysis therapy; 15 summary comorbid conditions previously listed; body weight; polycystic kidney disease; serum albumin, serum ferritin, and serum

PISONI ET AL

calcium levels; TSAT; catheter use; and country. Generalized estimating equations were used to account for clustering at the facility level, assuming a compound symmetry covariance structure.27 Logistic regression also was used to examine the odds of having an Hgb value of 11 g/dL or greater (ⱖ110 g/L versus ⬍11 g/dL [⬍110 g/L]) for 2 facility practice patterns: facility catheter use and percentage of patients in the facility with a TSAT less than 20%. A logistic regression model with the same adjustments as indicated above was used to determine the likelihood of patients being administered EPO as a function of the patient’s Hgb concentration at study entry. EPO dosing, DOPPS II. The association between weekly EPO dose and Hgb level was examined by country by using DOPPS II data. Average weekly EPO doses were calculated as prescribed EPO units per week averaged during a 4-week period. Route of administration also was investigated as percentage of patients administered IV versus subcutaneous (SC) EPO in 2002 and 2003. Unadjusted differences in EPO dosing across countries were investigated, looking at mean and median EPO units per week, as well as the distribution of patients across dose ranges. Differences in EPO dosing across countries also were examined after adjustment for case mix by using linear mixed statistical models to account for facility clustering. The model, which included only the baseline prevalent cross-section of patients who had been treated in the unit for longer than 28 days, was adjusted for age; sex; race; time on dialysis therapy; the 15 summary comorbid conditions listed; body weight; visual signs of malnourishment; ability to eat independently or walk without assistance; polycystic kidney disease as cause of ESRD; values for albumin, ferritin, TSAT, and serum calcium; catheter use; and route of EPO administration. All countries were compared with the United States as the reference. Patient characteristics associated with a greater administered EPO dose were examined by using linear mixed models and accounted for facility clustering by using the same adjustments listed previously.

RESULTS

Relationship of Patient Hgb Concentrations With Mortality and Hospitalization Risks In Cox models, relative risks (RRs) for mortality and all-cause hospitalization were 5% and 6% lower, respectively, for every 1-g/dL (10-g/L) greater Hgb concentration when using baseline Hgb level of each patient as a continuous variable and adjusting for a large number of patient demographic and comorbidity characteristics (RR, 0.95 and 0.94, respectively; P ⱕ 0.003 each). As shown in Fig 1, the relationship with both mortality and hospitalization risk varied across categories of Hgb concentrations, being steeper at lower levels of Hgb. For Hgb concentrations of 12 g/dL or greater (ⱖ120 g/L), a trend toward lower mortality risk was observed (RR, 0.92; P ⫽ 0.19), but there was no difference for

ANEMIA MANAGEMENT AND OUTCOMES IN 12 COUNTRIES

97

Fig 1. Greater patient Hgb values are associated with lower risk for mortality and hospitalization; DOPPS I data, 7 countries, 1996 to 2001. Separate Cox proportional hazards models were used to estimate RR for mortality and RR for hospitalization among patients on dialysis therapy for longer than 180 days based on Hgb level at study start. Both models were adjusted for age, sex, black race, years with ESRD, 15 comorbid conditions, single-pool Kt/V, serum phosphorus level, calcium level, albumin level, and country of residence and accounted for facility clustering. Overall RRs use Hgb level as a continuous variable. Mean patient follow-up time was 1.67 years for the mortality analysis and 0.98 years for the hospitalization analysis. To convert Hgb in g/dL to g/L, multiply by 10.

hospitalization compared with the group with Hgb levels of 11.0 to 11.9 g/dL (110 to 119 g/L). In additional analyses using the same adjustments, patient mortality risk was found to be significantly lower in dialysis units that had a greater mean Hgb concentration (RR for death, 0.90 for every 1-g/dL [10-g/L] greater facility mean Hgb concentration; P ⫽ 0.02). Similarly, patient hospitalization risk was lower in facilities that had greater mean Hgb concentrations (RR for hospitalization, 0.88 for every 1-g/dL [10-g/ L] greater facility mean Hgb concentration; P ⫽ 0.006). It is noteworthy that adding adjustments for EPO dose administered during the previous 4 weeks to the models did not diminish the relationship between patient baseline Hgb level and

mortality or hospitalization risk. Because of these insignificant effects, EPO dose was not included as an adjustment in the final mortality and hospitalization Cox models shown in Fig 1. Country Differences in Hgb Concentrations and EPO Use As listed in Table 1, there is a broad range in mean Hgb concentrations across the 12 countries in DOPPS II for 2002 to 2003. Mean Hgb concentrations ranged from 12.0 g/dL (120 g/L) in Sweden to 10.1 g/dL (101 g/L) in Japan. The United States and Spain showed the second highest mean Hgb value of 11.7 g/dL (117 g/L). However, despite relatively high mean Hgb concentrations in these countries, there were still substantial percentages of patients with an Hgb

98

PISONI ET AL

Table 1. Mean Hgb Levels and Percentage of Patients With Hgb Levels Less Than 11 g/dL for Patients on Dialysis Therapy Longer Than 180 Days and at the Time of Starting Dialysis, and Percentage of EPO Use for HD Patients on Dialysis Therapy for Longer Than 180 Days and During the Pre-ESRD Period by Country: DOPPS II Among Patients on Dialysis ⬎ 180 Days

Country

n1

Sweden 466 United States 1,690 Spain 513 Belgium 442 Canada 479 Australia/New Zealand 423 Germany 459 Italy 447 United Kingdom 436 France 341 Japan 1,210

EPO Use Mean Hgb Hgb ⬍ 11 g/dL (% of patients) (g/dL) (% of patients)

94 91 93 94 91 86 86 87 94 83 84

12.0 11.7 11.7 11.6 11.6 11.5 11.4 11.3 11.2 11.1 10.1

23 27 31 29 29 36 35 38 40 45 77

Among Patients New to ESRD, at Start of Dialysis*

n2

168 458 170 213 150 108 142 167 93 86 131

EPO Use Before Mean Hgb Hgb ⬍ 11 g/dL ESRD (% of patients) (g/dL) (% of patients)

65 27 56 33 43 50 46 59 44 43 62

10.7 10.4 10.6 10.3 10.1 10.1 10.5 10.2 10.2 10.1 8.3

55 65 61 66 70 70 61 68 67 65 95

NOTE. To convert Hgb in g/dL to g/L, multiply by 10. *Includes patients who were new to ESRD and entered DOPPS within 7 days of first-ever long-term dialysis treatment. Those administered EPO before ESRD had a 0.35-g/dL (3.5-g/L) higher Hgb level at the time of starting dialysis therapy compared with patients not administered EPO during the pre-ESRD period (P ⬍ 0.001).

value less than the K/DOQI and EBPG guideline target level of 11 g/dL (110 g/L): 23% to 29% of HD patients were below the Hgb guideline target in Sweden, the United States, Belgium, and Canada, and 31% to 45% of HD patients were below the target in Spain, Germany, Australia/New Zealand, Italy, the United Kingdom, and France. Furthermore, 77% of Japanese HD patients had an Hgb level less than 11 g/dL (⬍110 g/L), although current guidelines in Japan do not promote an Hgb target level of 11 g/dL (110 g/L). Although a mean Hgb concentration of 10.1 g/dL (101 g/L) was observed in Japan, 10% of Japanese dialysis units had a facility mean Hgb level of 10.9 to 11.5 g/dL (109 to 115 g/L). It should be noted that in the United Kingdom, 78% of patients had an Hgb concentration greater than 10 g/dL (⬎100 g/L), which closely approaches the recommendation set in the UK minimum standards document for greater than 85% of UK HD patients to have an Hgb concentration greater than 10 g/dL (⬎100 g/L).6 EPO use was calculated as percentage of patients on dialysis therapy for longer than 180 days who were administered EPO during a 4-week study period. As listed in Table 1, EPO use varied from 83% of patients in France to 94% in Belgium, Sweden, and the United Kingdom. Nearly identical percentages of EPO use by

country were seen in cross-sectional pointprevalent samples of HD patients that included patients on dialysis therapy for 180 days or less. The likelihood of HD patients being administered EPO was strongly related to a patient’s Hgb concentration (adjusted odds ratio [AOR] for being administered EPO, 0.61 per 1-g/dL [10-g/ L] greater Hgb level; P ⬍ 0.0001). Patients with an Hgb concentration less than 10 g/dL (⬍100 g/L) had 1.9-fold greater odds for being administered EPO compared with patients with an Hgb concentration of 11 to 11.99 g/dL (110 to 119 g/L). Moreover, patients with an Hgb concentration of 12 to 13 g/dL (120 to 130 g/L) had an AOR of 0.43 (P ⬍ 0.001) for being administered EPO compared with patients with an Hgb concentration of 11 to 11.99 g/dL (110 to 119 g/L). Country Differences in Anemia Management at Onset of ESRD Patients starting ESRD therapy with HD had a mean Hgb concentration of 10.2 g/dL (102 g/L), which was 1.1 g/dL (11 g/L) lower (country range, 0.8 to 1.8 g/dL [8 to 18 g/L] lower) than for patients on dialysis therapy for longer than 180 days (Table 1). A large country variation was seen in the percentage of HD patients administered EPO during the pre-ESRD period, ranging from as low as 27% of patients in the United

ANEMIA MANAGEMENT AND OUTCOMES IN 12 COUNTRIES Table 2.

99

Patient- and Facility-Level Characteristics Associated With Adjusted Odds of Patient Hgb Level of 11 g/dL or Greater, 2002 to 2003

Covariate

Patient-level characteristics* (n ⫽ 6,611) Polycystic kidney disease Serum albumin (per 0.3-g/dL higher) TSAT (per 10% higher) Male (v female) Serum calcium (per 1-mg/dL higher) Age (per 10 y older) Ferritin (per 100-ng/mL higher) Catheter use for vascular access Gastrointestinal bleeding in previous year Facility-level characteristics† Facility catheter use (per 10% higher use) (n ⫽ 6,389) Percent of facility patients with TSAT ⬍ 20% (per 10% more patients) (n ⫽ 4,755)‡

AOR (Hgb ⱖ11 v ⬍11 g/dL)

P

1.62 1.29 1.22 1.21 1.15 1.09 0.96 0.73 0.63

0.0002 ⬍0.0001 ⬍0.0001 0.001 ⬍0.0001 0.0002 ⬍0.0001 0.0001 ⬍0.0001

0.92

0.04

0.82

0.07

NOTE. To convert Hgb and albumin in g/dL to g/L, multiply by 10; calcium in mg/dL to mmol/L, multiply by 0.02495. *Adjusted for listed covariates and nonsignificant factors, including 14 comorbid classes, ESRD cause, body weight, time on ESRD, facility clustering, and country of residence. Among patients on dialysis therapy for longer than 180 days. †Adjusted for demographics, 15 comorbid classes, ESRD cause, body weight, albumin level, ferritin level, calcium level, time on ESRD, facility clustering, and country. Among patients on dialysis therapy for longer than 180 days. ‡Excludes facilities with TSAT measurements for less than 80% of patients.

States and 33% in Belgium to 43% to 65% in the other countries participating in the DOPPS. Patients who were administered EPO before starting dialysis therapy had a 0.35-g/dL (3.5-g/L) greater mean Hgb level at the time of initiating dialysis therapy (P ⬍ 0.001), compared with new HD patients who did not receive EPO during the pre-ESRD period. Furthermore, mean EPO dose at the time of starting dialysis therapy was 8,520 U/wk for patients who had been administered EPO during the pre-ESRD period compared with 10,229 U/wk for patients not administered EPO during the pre-ESRD period. Patient and Facility Characteristics Associated With Patient Hgb Concentrations of 11 g/dL or Greater (ⱖ110 g/L) An analysis that adjusted for numerous patient characteristics, country, and facility clustering effects was performed to determine the relationship between specific patient characteristics and the likelihood of patients having an Hgb concentration of 11 g/dL or greater (ⱖ110 g/L) versus less than 11 g/dL (⬍110 g/L). As listed in Table 2, the adjusted odds of having an Hgb level of 11 g/dL or greater (ⱖ110 g/L) was significantly greater for HD patients who were men, were

older, had polycystic kidney disease, or had greater levels of serum albumin, serum calcium, or TSAT. However, patients who were dialyzing with a catheter, had gastrointestinal bleeding in the previous 12 months, or had a greater ferritin concentration showed a significantly lower likelihood of having an Hgb level of 11 g/dL or greater (ⱖ110 g/L). Categorical analysis of the relationship between Hgb level and different levels of serum albumin, TSAT, and ferritin indicated that the relationships were approximately linear across all levels of each of these measures. Patients older than 75 years showed the highest adjusted odds of having an Hgb level of 11 g/dL or greater (ⱖ110 g/L; AOR, 1.5; P ⬍ 0.0001 compared with patients 18 to 49 years old). Additional investigation focused on facility practices and patient Hgb concentrations. As listed in Table 2, for every 10% greater use of catheters for vascular access within a dialysis facility, patients in those facilities had 8% lower odds of reaching the target Hgb level of 11 g/dL or greater (ⱖ110 g/L; AOR, 0.92; P ⫽ 0.04). Facility catheter use varied from 0% to more than 50% across the 12 DOPPS countries (median, 11%). In one quarter of DOPPS dialysis units, more than 27% of patients used a catheter.

100

PISONI ET AL

Fig 2. Variability and mean net change in Hgb levels during 4 months; DOPPS II data, HD patients on dialysis therapy for longer than 180 days. Mean Hgb concentrations at baseline (start of the 4-month interval) are shown by 4 different categories of baseline Hgb level. Mean 4-month changes in Hgb concentrations are shown by the length and direction of the arrows, and SD is noted. To convert Hgb in g/dL to g/L, multiply by 10.

Catheter use among patients showed a large variation across regions: 1% in Japan; 5% to 12% in Australia, New Zealand, France, Germany, Italy, and Spain; 25% to 27% in Sweden, the United Kingdom, and the United States; 33% in Canada; and 38% in Belgium (data not shown). Patients also had a lower probability of having an Hgb level greater than 11 g/dL (⬎110 g/L) if they were treated in a dialysis unit that had a larger fraction of patients with a TSAT less than 20% (AOR, 0.82 for every 10% more patients with a TSAT ⬍ 20%; P ⫽ 0.07). Mean Changes in Patient Hgb Concentrations During a 4-Month Interval An analysis was performed to determine the degree to which patient Hgb values changed during a 4-month interval, depending on the patient’s Hgb concentration at the start of the interval (baseline Hgb concentration). The Hgb measurement 4 months later showed large intrapatient fluctuations in Hgb concentrations, with an SD of 1.15 to 1.3 g/dL (12 to 13 g/L) for patients who had a baseline Hgb level less

than 10 g/dL (⬍100 g/L) or 12 g/dL or greater (ⱖ120 g/L) and a lower SD of 0.82 to 0.95 g/dL (8.2 to 9.5 g/L) for patients who had a baseline Hgb level of 10 to 11.99 g/dL (100 to 119.9 g/L; Fig 2). Furthermore, mean change in patient Hgb concentrations during a 4-month period was strongly dependent on the patient’s baseline Hgb value. As shown in Fig 2, this analysis indicated that a patient’s Hgb concentration increased by an average of 1.08 g/dL (10.8 g/L) for patients who had a baseline Hgb concentration of less than 10 g/dL (⬍100 g/L), with an average increase of 0.48 g/dL (4.8 g/L) in patient Hgb levels for patients with a baseline Hgb concentration of 10 to 10.99 g/dL (100 to 109 g/L). Conversely, patient Hgb concentrations decreased by an average of 0.91 g/dL (9.1 g/L) for patients with a baseline Hgb value of 12 g/dL or greater (ⱖ120 g/L). This latter finding likely reflects the observed reduction in EPO use when patient Hgb values exceed 12 g/dL (120 g/L).

ANEMIA MANAGEMENT AND OUTCOMES IN 12 COUNTRIES Table 3.

101

EPO Dose for 2002 to 2003, Cross-Section of HD Patients Administered EPO by DOPPS Country EPO Dose (U/wk)

Patients With Indicated Level of EPO U/wk (%)

Country

Mean

Median

1K-18K

18K-24K

24K-36K

36K-75K

⬎75K

Japan Germany France Spain United Kingdom Italy Australia/New Zealand Canada Sweden Belgium United States restricted* United States all

5,297 6,846 7,401 7,607 8,010 8,118 8,725 10,808 12,202 12,312 12,947 17,360

4,875 6,000 6,000 6,000 6,000 6,000 8,000 8,000 10,000 10,000 12,000 12,000

98 99 96 96 96 95 91 86 78 85 75 69

— 1 2 3 2 3 5 7 15 7 11 10

2 — 2 1 2 2 3 6 5 6 14 13

— — — — — — — — 1 3 — 7

— — — — — — — — — — — 2

NOTE. Mean values shown were weighted for the fraction of patients sampled in each facility. Excludes darbepoetin (Aranesp). *In United States, when restricted to the 91% of patients administered EPO 36,000 U/wk or less.

Country Variations in EPO Dosing and Relationship to Patient Hgb Level Mean and median weekly EPO doses were calculated for 2002, excluding patients administered darbepoetin (4% to 17% use in Europe and Australia/New Zealand and essentially no darbepoetin use in Japan

or the United States). Mean weekly EPO dose varied from 5,297 U/wk in Japan to 17,360 U/wk in the United States (Table 3). However, as shown in Fig 3, a clear trend was seen in the relationship between greater country mean EPO dose and a greater country mean Hgb concentration.

Fig 3. Unadjusted weekly EPO dose and mean Hgb level by country; DOPPS II data, 12 countries, 2002 to 2003, among a prevalent cross-section of patients. **US data for patients administered EPO, 1 to 36,000 U/wk. Abbreviations: SW, Sweden; US, United States; SP, Spain; BE, Belgium; CA, Canada; ANZ, Australia/New Zealand; GE, Germany; IT, Italy; UK, United Kingdom; FR, France; JA, Japan. To convert Hgb in g/dL to g/L, multiply by 10.

102

PISONI ET AL

Table 4. Mean EPO Dose by Country With and Without Case-Mix Adjustment (for Patients Receiving EPO Equal to or Less Than 36,000 U/wk) and Percentage and Odds of Hgb Level of 11 g/dL or Greater: DOPPS II, 2002 to 2003

No Adjustment for Case-Mix

Adjusted for Case-Mix

Percentage of Patients With Hgb ⱖ 11 g/dL (unadjusted)

⫺1,272 Reference* ⫺1,693† ⫺2,627† ⫺4,575† ⫺4,660† ⫺4,775† ⫺5,455† ⫺5,713† ⫺7,413† ⫺6,395†

⫹365 (NS) Reference* ⫺737 (NS) ⫺1,158 (NS) ⫺2,922† ⫺2,925† ⫺2,980† ⫺3,436† ⫺4,027† ⫺5,858† ⫺6,038†

74 76 68 66 63 56 58 67 51 19 61

Mean EPO U/wk (difference from US)

Country

Sweden United States Belgium Canada Australia/New Zealand Italy United Kingdom Spain France Japan Germany

AOR of Hgb ⱖ11 v ⬍11 g/dL (relative to United States)

1.18 (NS) 1.00 (reference) 0.87 (NS) 0.86 (NS) 0.54† 0.48† 0.53† 0.62† 0.40† 0.08† 0.57†

NOTE. Among prevalent cross-section of patients, mean during previous 28 days; adjusted for age, sex, black race, time on ESRD, 15 comorbidity classes, body weight, malnourishment, ability to eat independently or walk without assistance, polycystic kidney disease as ESRD cause, serum albumin level, ferritin level, TSAT, serum calcium level, catheter use, route of EPO administration, facility clustering, and country (not adjusted for patient Hgb level). Appropriate standard error adjustments were made to account for clustering of patients and their treatments within facilities. (n ⫽ 5,364.) Excludes darbepoetin (Aranesp). To convert Hgb in g/dL to g/L, multiply by 10. Abbreviation: NS, not significant at P ⬎ 0.05. *US reference ⫽ 12,947 U/wk. †Significantly different from the United States at P ⬍ 0.05.

Median weekly EPO dose was lower than its mean value in all countries (Table 3), with an especially large difference seen in the United States, where the median weekly EPO dose (12,000 U/wk) was dramatically lower than the mean EPO dose of 17,360 U/wk. The difference between mean and median EPO dose values results from a skewed distribution, with particularly high EPO doses prescribed for small fractions of patients. Histograms of EPO dosing by country indicated that a weekly dose greater than 36,000 U/wk was administered to 9% of patients in the United States, 3% in Belgium, 1% in Sweden; and essentially no patients in the DOPPS samples of any other countries. When patients administered more than 36,000 U/wk of EPO were excluded from the analysis, mean EPO dose in the United States was substantially lower (12,947 U/wk). Additional investigation and characterization of US patients administered the greater EPO doses of greater than 36,000 U/wk is described by Held et al (manuscript submitted for publication). Additional analysis was performed to describe patient characteristics associated with EPO dosing and how adjustment for patient characteris-

tics affects differences in EPO dosing across countries. These analyses were restricted to patients administered 36,000 U/wk or less of EPO to provide comparisons based on the usual practice that pertains to 91% of patients in the United States and 97% to 100% of HD patients in the other 11 DOPPS countries. The cutoff value of 36,000 U/wk was chosen from a histogram analysis as the point at which the long tail in greater EPO doses begins in the United States. Results listed in Table 4 indicate that case-mix adjustment for the characteristics of US patients reduces the differences in EPO dosing between the United States and other countries. The EPO dose after case-mix adjustment was found to be not significantly different between Belgium, Canada, Sweden, and the United States; these 4 countries also showed the greatest adjusted odds of their patients having an Hgb level of 11 g/dL or greater (ⱖ110 g/L). Other countries showed a significantly and substantially lower EPO dose than these 4 countries, and their lower EPO dose also was associated with significantly lower odds of patients in each of those countries having an Hgb level of 11 g/dL or greater (ⱖ110 g/L). Patient characteristics that were significantly

ANEMIA MANAGEMENT AND OUTCOMES IN 12 COUNTRIES Table 5. Patient Characteristics Associated With Administration of Greater Weekly EPO Dose: DOPPS II, 2002 to 2003

Covariate

TSAT (per 5% lower) Serum albumin (per 0.3 g/dL lower) Body weight (per 10 kg higher) Age (per 10 y younger) Coronary artery disease (yes v no) Congestive heart failure (yes v no) Other cardiovascular disease (yes v no) Female (v male) Dyspnea (yes v no) Psychiatric disorders (yes v no) Catheter use (yes v no) Not having polycystic kidney disease Hgb (per 1 g/dL lower) Gastrointestinal bleed in previous year (yes v no) IV EPO administration (v SC)

Difference in EPO (U/wk)

P

93

0.006

207 298 363

0.03 ⬍0.0001 ⬍0.0001

389

0.05

521

0.01

592 615 641 710 735

0.003 ⬍0.0001 0.04 0.002 0.004

763 962

0.05 ⬍0.0001

1,354 1,420

⬍0.0001 ⬍0.0001

NOTE. Difference in EPO dose adjusted for listed covariates, facility clustering, country of residence, and nonsignificant factors of black race, time on ESRD, 9 comorbidity classes, weight, malnourishment, ability to eat or walk independently, serum ferritin level, and calcium level (n ⫽ 5,364). All values are positive, indicating higher EPO dose. Restricted to patients administered EPO, 36,000 U/wk or less per week. Overall mean EPO dose from model ⫽ 9,164 U/wk. Excludes darbepoetin (Aranesp). Characteristics associated with EPO dose greater than 36,000 U/wk have been described by Held et al [manuscript submitted]. To convert albumin and Hgb in g/dL to g/L, multiply by 10.

related to patients being administered greater weekly EPO doses are listed in Table 5, along with the incremental effect of each characteristic on mean weekly EPO dose across the 12 countries in DOPPS. In this adjusted analysis, significantly greater weekly EPO doses were observed for patients who were younger or female; had greater body weight; did not have polycystic kidney disease as cause of ESRD; had a lower value for TSAT, Hgb, or serum albumin; had coronary artery disease, congestive heart failure, other cardiovascular disease, dyspnea, a psychiatric disorder, or gastrointestinal bleeding in the previous year; were using a catheter for vascular access; or were administered IV (versus SC)

103

EPO. This EPO dose analysis, adjusted for patient mix and Hgb level, indicated a 14% lower EPO dose for patients administered EPO by SC (versus IV) route (P ⬍ 0.0001). However, in the United States, SC EPO doses were, on average, only 3% lower (P ⫽ 0.73) than IV EPO doses after adjustment for case mix and at the same Hgb concentrations. Trends in Route of EPO Administration and Associated EPO Dosing In 2002 and 2003, health policies or recommendations in some countries discouraged and prohibited the use of certain EPO preparations to be administered by the SC route to avoid the potential risk for red cell aplasia that has been associated in rare occurrences with SC administration of certain EPO preparations. An analysis of DOPPS data showed a large shift from SC toward IV EPO administration since 2002 (Fig 4). By November 2003, IV EPO administration served as the major route in 11 of 12 countries in the DOPPS, whereas in early 2002, IV EPO administration was the predominant route in only 4 of the countries (Belgium, Germany, Japan, and the United States). Indicators of Iron Deficiency and IV Iron Therapy by Country The percentage of HD patients on dialysis therapy for longer than 180 days who were administered IV iron during a 4-month study period varied greatly, from 89% IV iron use in Belgium to 38% in Japan (Fig 5). However, in some countries with a high percentage of patients administered IV iron (eg, Sweden, Belgium, and Germany), a large fraction of patients (31% to 38%) had indications for iron deficiency by TSAT values less than 20% (below EBPG and K/DOQI guidelines). Analyses that adjusted for numerous patient characteristics did not show a significant relationship between patient TSAT values and the percentage of patients administered IV iron at a dialysis unit or the number of IV iron doses in a 4-month period. These results indicate that substantial deficits in iron stores exist in HD patients despite a large fraction of patients administered IV iron. The percentage of HD patients within DOPPS dialysis units administered IV iron during a 4-month interval was found to be significantly

104

PISONI ET AL

Fig 4. Percentage of EPO-treated patients administered EPO through the IV route in 2002 versus 2003. Calculations for the year 2002 consider the first reported EPO use between January 1 and August 31, 2002. Calculations for the year 2003 consider the most recently reported EPO use between January 1 and August 31, 2003. The remainder of patients were administered EPO SC. Darbepoetin (Aranesp) therapy is excluded. Abbreviations: SW, Sweden; US, United States; SP, Spain; BE, Belgium; CA, Canada; ANZ, Australia/New Zealand; GE, Germany; IT, Italy; UK, United Kingdom; FR, France; JA, Japan.

related (P ⬍ 0.0001) to the dialysis unit’s mean Hgb concentration when adjusted for country and facility median EPO dose. An especially large variation of 0% to 100% was seen in IV iron use among Japanese dialysis units. In Japan, greater facility IV iron use also was strongly associated with greater facility mean Hgb concentrations (every 20% greater facility IV iron use was associated with 0.18-g/dL [1.8-g/L] greater facility mean Hgb concentration; P ⫽ 0.002 after adjustment for facility median weekly EPO dose). Moderate country variation was seen in the percentage of patients with a serum ferritin concentration less than 100 ng/mL, ranging from 4% in the United Kingdom to 13% in Italy; all other countries had 7% to 12% of patients with such a low serum ferritin concentration. Across the DOPPS II countries with at least 60% reporting of ferritin and TSAT values, 4% to 7% of HD patients had a TSAT less than 20% and a serum ferritin concentration less than 100 ng/mL. A median serum iron concentration of 59 to 62 ␮g/dL (10.6 to 11.1 ␮mol/L) was observed for HD patients in 9 of 12 DOPPS II countries; 56

␮g/dL (10.0 ␮mol/L) in Canada, and 53 ␮g/dL (9.5 ␮mol/L) in the United States and Belgium. Despite this similarity in country median serum iron concentrations, large variations in mean serum iron concentrations were seen across facilities, ranging from 49 ␮g/dL (8.8 ␮mol/L) for the lower 10th percentile to 78 ␮g/dL (14.0 ␮mol/L) for the upper 90th percentile of the distribution. One notable trend in the 2 cross-sections of prevalent HD patients between 1999 and 2002 to 2003 was a substantial increase in median serum ferritin concentration in each of the 7 countries that participated in DOPPS I and II (data not shown). Median serum ferritin concentration increased by 24%-36% in France, Germany, and the United States; by 53%-63% in Spain, Italy, and the United Kingdom; and by 114% in Japan. Conversely, median TSAT values did not show corresponding increases during this period. Hospitalization and Patient Hgb Concentrations An analysis of EPO-treated patients in DOPPS I before and after hospitalization indicates that in

ANEMIA MANAGEMENT AND OUTCOMES IN 12 COUNTRIES

105

Fig 5. Percentages of prevalent HD patients administered IV iron and patients with a low TSAT (<20%) by country; DOPPS II data, 2002 to 2003, restricted to patients on dialysis therapy for longer than 180 days. TSAT data are based on the most recent value, and receipt of IV iron is from the previous 4-month period. Data reporting for TSAT varied. Abbreviations: SW, Sweden; US, United States; SP, Spain; BE, Belgium; CA, Canada; ANZ, Australia/ New Zealand; GE, Germany; IT, Italy; UK, United Kingdom; FR, France; JA, Japan; NA, results not shown because of less than 50% data reporting for TSAT (potential selection); 52% to 64% TSAT reporting in France, Germany, Italy, and Sweden (interrupted lines); 75% to 92% TSAT reporting in all other countries.

the United States, Hgb concentrations were, on average, 0.53 g/dL (5.3 g/L) lower within the 30 days after a hospitalization compared with patient Hgb concentrations measured approximately 3 months before hospitalization (P ⬍ 0.001). Across all 7 DOPPS I countries, the decline in Hgb concentrations with hospitalization was 0.42 g/dL (4.2 g/L; P ⬍ 0.001). These results further substantiate the expectation of worse anemia control for patients who are hospitalized or have comorbidities that induce EPO hyporesponsiveness (eg, infections). Large variation was observed in hospitalization rates across the 7 countries in DOPPS I, ranging from 1.7 hospitalizations/patient-year in the United States to 0.65 to 0.8 hospitalizations/patient-year in Japan, Italy, and Spain (Fig 6). DISCUSSION

These current results from the international DOPPS indicate large improvements in anemia management during the last several years, with HD populations in more than half the DOPPS countries showing a mean Hgb level of 11.5 g/dL or greater (ⱖ115 g/L). Recently, Locatelli et al5

showed significant increases in HD patients’ Hgb levels in 5 European countries within 1 year after publication of the EBPG on anemia management.19 These guidelines, along with the NKF-K/ DOQI guidelines first published in the United States in 1997,18 the CSN guidelines published in Canada in 1999,20 and the Renal Association’s Standards Document in the United Kingdom,6 have provided benchmarks for improving anemia control in this patient population. Different ethnic populations may have slight differences in their mean Hgb levels. However, at Hgb concentrations observed in this study among long-term HD patients, greater Hgb levels were associated with better outcomes independent of country and adjusted for race. Our analyses show mortality and hospitalization risks to be 5% and 6% lower for every 1-g/dL (10-g/L) greater Hgb concentration, respectively. Although not significant with the current sample size, our results suggest a lower mortality risk for patients who have an Hgb concentration of 12 g/dL or greater (ⱖ120 g/L; RR, 0.92; P ⫽ 0.19) compared with an Hgb concentration of 11 to 11.9 g/dL (110 to 119 g/L). This observation

106

PISONI ET AL

Fig 6. Hospitalization rates by country; DOPPS I data, prevalent cross-section of patients, 7 countries. Unadjusted hospitalization rates were calculated as the number of hospitalizations per patient-year at risk for the period between May 1, 1999, and April 30, 2000. Hospitalization rates by European DOPPS country also are reported by Rayner et al.53

agrees with recent significant results of Ofsthun et al,28 which showed an RR for mortality of 0.84 (P ⬍ 0.007) for patients who had an Hgb level of 12 to 12.9 g/dL (120 to 129 g/L) compared with 11 to 11.9 g/dL (110 to 119 g/L), suggesting a possible mortality benefit for patients beyond the current Hgb target level recommended by the Centers for Medicare and Medicaid Services and K/DOQI guidelines. The analysis of Ofsthun et al28 was based on 44,550 US patients treated at Fresenius dialysis units between July 1, 1998, and June 30, 2000, and was adjusted for age, race, sex, diabetes, body mass index, albumin level, missed treatments, and urea reduction ratio. The current DOPPS results do not show as large a decline in mortality risk per unit higher Hgb concentration as that shown by Ofsthun et al28; this difference may be attributed to the use of more comorbid factors and patient laboratory measures as adjustments in the present study, as well as the longer mean follow-up time between baseline Hgb measurements and death or censoring in the DOPPS. Finally, these DOPPS results further indicate that mortality and hospitalization

risks for patients were significantly lower when patients were treated in a dialysis unit that had a greater mean Hgb concentration. In addition to the relationship of improved anemia control with lower mortality risk and hospitalization risk, numerous other studies have indicated significant increases in patient quality of life and physical functioning.2,8,11,15,29-32 Although anemia management has improved considerably during the last decade, representative cross-sections of HD patients from the current DOPPS results indicate that in 2002 to 2003, 23% to 77% of HD patients by country were still not reaching the guideline Hgb level of 11 g/dL or greater (ⱖ110 g/L). Thus, DOPPS findings suggest that additional efforts to achieve the recommended guideline levels for Hgb will likely benefit patients. Lacson et al21 described large variability in an individual patient’s Hgb concentration, with an SD of 1.3 g/dL (13 g/L) observed during a 3-month interval in 65,000 Fresenius dialysis patients in the United States during the first quarter of 2000. Similar results have been re-

ANEMIA MANAGEMENT AND OUTCOMES IN 12 COUNTRIES

ported by Berns et al.22 The present DOPPS analyses further indicate large intrapatient variability across the 12 DOPPS countries during a 4-month period, with a greater SD for Hgb concentrations less than 10 g/dL (⬍100 g/L) or 12 g/dL or greater (ⱖ120 g/L; SD ⫽ 1.15 to 1.3 g/dL) than for Hgb concentrations of 10 to 11.9 g/dL (100 to 119 g/L; SD ⫽ 0.82 to 0.95 g/dL). Mean 4-month change in Hgb levels was ⫹1.08 g/dL (⫹10.8 g/L) for patients who had a baseline Hgb less than 10 g/dL (⬍100 g/L). In sharp contrast, patients who had a baseline Hgb level of 12 g/dL or greater (ⱖ120 g/L) showed an average decline in mean Hgb levels of similar magnitude (0.91 g/dL [9.1 g/L]) during the same interval. Although patients with an Hgb level of 11 to 11.9 g/dL (110 to 119 g/L) showed very little change in mean Hgb levels, their SD in 4-month intrapatient Hgb level variability remained high. These results, along with those of Lacson et al21 and Berns et al,22 highlight the difficulties of maintaining patient Hgb values within a narrow range, even during a relatively short period, with this variability determined by baseline Hgb level and patterns of EPO dosing. Although 83% to 94% of prevalent HD patients were administered EPO across the 12 DOPPS countries in 2002 to 2003, this study indicated large variation in EPO use before the initiation of dialysis therapy, from 27% in the United States to 65% in Sweden. On average, new patients with ESRD initiate HD therapy with a substantially lower mean Hgb concentration (1.1 g/dL [11 g/L] lower) than that of prevalent HD patients. Several other recent reports have indicated low mean Hgb concentrations and low EPO use during the pre-ESRD period.33,34 Our results also indicate that patients administered EPO during the pre-ESRD period had a greater mean Hgb level (by 0.35 g/dL [3.5 g/L]) when initiating HD than patients not administered EPO before the onset of ESRD. It is conceivable that patients administered EPO more frequently and for longer duration during stage 4 of chronic kidney disease would show an even greater benefit. Substantial benefits have been described for patients with chronic kidney disease as a result of improved anemia management through greater EPO use during the pre-ESRD period, including lower hospitalization during the first month after the start of dialysis therapy,

107

lower mortality risk during the first year of dialysis therapy, improved quality of life, and reduced occurrence of left ventricular hypertrophy.8,29,32,33,35-38 Thus, DOPPS confirms from an international perspective that opportunities to improve anemia control are large for prevalent patients and even larger for patients before requiring long-term dialysis therapy. The present study indicates significantly better anemia control for patients who are older; are men; have greater TSAT, serum albumin, and serum calcium levels; have lower serum ferritin levels; and are not using a catheter for vascular access. The inverse relationship between serum ferritin level and patient Hgb level also was reported recently by Reddan et al.39 Furthermore, patients dialyzing in facilities with high catheter use had significantly lower odds of having an Hgb level of 11 g/dL or greater (ⱖ110 g/L). In a previous work from the DOPPS, Combe et al40 showed that vascular access infection rates are 5- to 8-fold greater for patients dialyzing with a catheter compared with those using an arteriovenous fistula. Additionally, Pisoni et al41 found that risk for infection-related hospitalization is 60% greater for patients dialyzing in facilities with catheter use greater than 28%; all-cause mortality and hospitalization risk also were greater for patients in these facilities. Catheter use among prevalent patients is common in some of the countries studied, ranging from 38% catheter use in Belgium, 33% in Canada, and 25% to 27% in Sweden, the United Kingdom, and the United States. Considering the numerous negative aspects associated with catheter use, efforts to reduce catheter use are likely to result in decreased hospitalization and mortality risk, improved anemia control, and, consequently, decreased EPO use. Furthermore, the recent study of Nassar et al42 provides an excellent example of successfully showing the underlying source of accessrelated infection in patients who were EPO hyporesponsive but without other clinical signs or symptoms of localized or systemic infection. They found that 75% of EPO-hyporesponsive patients with an asymptomatic nonfunctioning arteriovenous graft showed fully positive scans for indium-labeled white blood cells at the site of the nonfunctioning graft,

108

with bacteriological infection confirmed after surgical resection of these grafts. After resection, patient Hgb and serum albumin concentrations increased, whereas EPO dose and serum ferritin and C-reactive protein levels declined. The significant correlation of high ferritin level with low Hgb level in the present study suggests that inflammation or infection has a major role in the anemia of dialysis patients. Large differences were seen in the DOPPS regarding EPO dosing across countries. An evaluation of the relationship between country mean EPO doses and mean patient Hgb concentrations indicated overall a significantly greater country mean Hgb concentration in countries with a greater mean EPO dose. In the United States, approximately 9% of patients were administered EPO doses greater than 36,000 U/wk. Because this is a greater fraction than in other countries, this topic is studied in greater detail in a report by Held et al (manuscript submitted). For typical patients (91% in the United States) administered EPO doses of 36,000 units or less, we found that EPO doses after case-mix adjustment were not significantly different between Belgium, Canada, Sweden, and the United States. All other DOPPS countries had a significantly lower mean weekly EPO dose, and each of these countries also had significantly lower odds of their patients having an Hgb level of 11 g/dL (110 g/L) or greater. In Japan, EPO doses and Hgb levels were both particularly low. The country’s maximum allowed EPO dose is 3,000 units 3 times weekly,43 unless written requests for individual patients are submitted. Even in Japan, a significant inverse association between Hgb level and mortality risk could be documented in the DOPPS. The relationship of greater EPO dose associated with greater Hgb levels was observed on a country basis. However, in a patient-based analysis adjusted for case mix, the opposite is observed: Patients with low Hgb levels are administered significantly greater EPO doses, and patients with high Hgb levels are administered lower EPO doses. This analysis shows that patients receive appropriate adjustments in their therapy. Several patient characteristics and treatment factors were associated with the administration of greater EPO doses, including younger age,

PISONI ET AL

female sex, greater body weight, and several comorbid conditions. Of particular interest is the association with iron deficiency (low TSAT), potential markers of inflammation/nutrition (such as low albumin level), and catheter use for vascular access. Greenwood et al23 recently showed a similar relationship of these demographic factors and measures of nutrition, inflammation, and iron stores on the EPO dose/Hgb ratio in a study of 64 HD units. Because EPO dosing differs depending on patient comorbidity, adjustments for numerous patient case-mix characteristics, shown in the present study, are important in comparing EPO dosing across countries. Furthermore, other factors in addition to patient mix must be considered when evaluating EPO dose needs on a patient level and mean Hgb values as a practice pattern of facilities. Some dialysis facilities have much greater hospitalization rates than other units in the same country, and hospitalization is associated with declines in patient Hgb levels. Facility mean hospitalization rates varied more than 2-fold across the 7 countries of DOPPS I. A large number of studies have examined the difference in mean weekly EPO dose for SC versus IV administration, with highly variable results obtained, ranging from no difference to 60% lower dose for EPO administered SC.44-49 In the present study, we found mean weekly EPO doses, adjusted for both case mix and Hgb concentrations, to be 14% lower for SC (versus IV) EPO throughout all DOPPS countries. It is unclear why this difference was smaller in the United States (3% lower). Our results are consistent with those recently reported by Moist et al,50 in which a 7.8% greater mean weekly EPO dose was found for IV administration after all patients within a dialysis unit were switched from SC to IV to maintain preswitch Hgb levels. The DOPPS documented a major change from SC to IV EPO use between 2002 and 2003 across all 12 DOPPS countries. This sudden change likely is a response to the recent rare occurrences of pure red cell aplasia that have been associated with SC, but not IV, administration of certain EPO preparations. Large differences also were seen by country in percentages of patients prescribed IV iron during a 4-month period, ranging from 38% in Japan to 89% in Belgium. However, even in

ANEMIA MANAGEMENT AND OUTCOMES IN 12 COUNTRIES

countries with high IV iron use, large proportions of patients continued to have a TSAT less than 20%, and we failed to see a significant relationship between facility IV iron use and percentage of patients with a TSAT less than 20%. This observation needs to be explored further, with consideration of cumulative iron doses and other factors. Because a greater TSAT (as a measure of iron stores) was positively associated with better anemia control, it is important to consider additional treatment practices, such as minimizing blood loss, reducing inflammation, and more vigorous iron therapy when both TSAT and ferritin levels are low. In summary, large variations are seen in anemia management practices across the 12 countries participating in the DOPPS. However, across these countries, substantial differences also exist regarding patient comorbidity,51 hospitalization rates, vascular access use,52 and other HD treatment practices that bear on anemia management practices. DOPPS analyses have the advantage of permitting adjustments for case mix, laboratory values, and practice patterns. Overall, changes in anemia management practices that have been guided by national and regional practice guidelines have resulted in large improvements in anemia control for HD patients during the last 5 years across nearly all DOPPS countries. Because better anemia control is associated with lower morbidity and mortality, continued focus on anemia management practices, coupled with improvements in other areas of HD practice, such as catheter use for vascular access, should lead to additional gains in anemia control for HD patients. ACKNOWLEDGMENT The authors gratefully acknowledge helpful suggestions from Dr Naoki Kimata; the assistance of Trinh Pifer, Justin Albert, Miles P. Finley, Caroline Shevrin, and Stacey Elder with preparation of this manuscript; the work of David Dickinson, Jon Bodfish, and Rameswari Metla in maintaining the DOPPS database; the multitude of activities of Michael Losey, Theresa Helm, and Patrick Carlson in working with the sites participating in the DOPPS; and the great dedication of staff members and medical directors from more than 300 dialysis units in the 12 countries in the DOPPS for all their efforts in participating in this study.

109

REFERENCES 1. US Renal Data System: USRDS Annual Data Report, treatment modalities. Am J Kidney Dis 24:S57-S75, 1994 (suppl 2) 2. Locatelli F, Conte F, Marcelli D: The impact of haematocrit levels and erythropoietin treatment on overall and cardiovascular mortality and morbidity—The experience of the Lombardy Dialysis Registry. Nephrol Dial Transplant 13:1642-1644, 1998 3. Ma JZ, Ebben J, Xia H, Collins AJ: Hematocrit level and associated mortality in hemodialysis patients. J Am Soc Nephrol 10:610-619, 1999 4. Xia H, Ebben J, Ma JZ, Collins AJ: Hematocrit levels and hospitalization risks in hemodialysis patients. J Am Soc Nephrol 10:1309-1316, 1999 5. Locatelli F, Pisoni RL, Combe C, et al: Anaemia in haemodialysis patients in five European countries: Association with morbidity and mortality in the Dialysis Outcomes and Practice Patterns Study (DOPPS). Nephrol Dial Transplant 19:121-132, 2004 6. Renal Association: Treatment of Adults and Children With Renal Failure: Standards and Audit Measures (ed 3). London, UK, Royal College of Physicians of London and the Renal Association, 2002 7. Foley RN, Parfrey PS, Harnett JD, Kent GM, Murray DC, Barre PE: The impact of anemia on cardiomyopathy, morbidity, and mortality in end stage renal disease. Am J Kidney Dis 31:53-61, 1998 8. Levin A, Thompson CR, Ethier J, et al: Left ventricular mass index increase in early renal disease: Impact of decline in hemoglobin. Am J Kidney Dis 34:125-134, 1999 9. O’Riordan E, Foley RN: Effects of anaemia on cardiovascular status. Nephrol Dial Transplant 15:S19-S22, 2000 (suppl 3) 10. Nissenson AR: Epoetin and cognitive function. Am J Kidney Dis 20:S21-S24, 1992 (suppl 1) 11. Beusterien KM, Nissenson AR, Port FK, Kelly M, Steinwald B, Ware JE Jr: The effects of recombinant human erythropoietin on functional health and well-being in chronic dialysis patients. J Am Soc Nephrol 7:763-773, 1996 12. Canadian Erythropoietin Study Group: Association between recombinant human erythropoietin and quality of life and exercise capacity of patients receiving hemodialysis. BMJ 3:573-578, 1990 13. Evans RW, Rader B, Manninen DL: The quality of life of hemodialysis recipients treated with recombinant human erythropoietin. Cooperative Multicenter EPO Clinical Trial Group. JAMA 263:825-830, 1990 14. Lundin AP, Akerman MJ, Chesler RM, et al: Exercise in hemodialysis patients after treatment with recombinant human erythropoietin. Nephron 58:315-319, 1991 15. McMahon LP, McKenna MJ, Sangkabutra T, et al: Physical performance and associated electrolyte changes after haemoglobin normalization: A comparative study in haemodialysis patients. Nephrol Dial Transplant 14:11821187, 1999 16. Moreno F, Sanz-Guajardo D, Lopez-Gomez JM, Jofre R, Valderrabano F: Increasing the hematocrit has a beneficial effect on quality of life and is safe in selected hemodialysis patients. Spanish Cooperative Renal Patients

110

Quality of Life Study Group of the Spanish Society of Nephrology. J Am Soc Nephrol 11:335-342, 2001 17. Painter P, Moore G, Carlson L, et al: Effects of exercise training plus normalization of hematocrit on exercise capacity and health-related quality of life. Am J Kidney Dis 39:257-265, 2002 18. National Kidney Foundation: K/DOQI Clinical Practice Guidelines: 2000 Update. Am J Kidney Dis 37:S1-S238, 2001 (suppl 1) 19. European Best Practice Guidelines for the management of anaemia in patients with chronic renal failure. Nephrol Dial Transplant 14:S1-S50, 1999 (suppl 5) 20. Toffelmire EB, Barrett BJ, Fenton SS, et al: Clinical practice guidelines for the management of anemia coexistent with chronic renal failure. J Am Soc Nephrol 10:S292-S296, 1999 (suppl 13) 21. Lacson E Jr, Ofsthun N, Lazarus JM: Effect of variability in anemia management on hemoglobin outcomes in ESRD. Am J Kidney Dis 41:111-124, 2003 22. Berns JS, Elzein H, Lynn RI, Fishbane S, Meisels IS, Deoreo PB: Hemoglobin variability in epoetin-treated hemodialysis patients. Kidney Int 64:1514-1521, 2003 23. Greenwood RN, Ronco C, Gastaldon F, et al: Erythropoietin dose variation in different facilities in different countries and its relationship to drug resistance. Kidney Int Suppl 87:S78-S86, 2003 24. Young EW, Goodkin DA, Mapes DL, et al: The Dialysis Outcomes and Practice Patterns Study: An international hemodialysis study. Kidney Int Suppl 74:S74-S81, 2000 25. Klein J, Moeschberger M: Survival Analysis Techniques for Censored and Truncated Data. New York, NY, Springer, 1997, pp 416-418 26. Snyder JJ, Foley RN, Gilbertson DT, Vonesh EF, Collins AJ: Hemoglobin levels and erythropoietin doses in hemodialysis and peritoneal dialysis patients in the United States. J Am Soc Nephrol 15:174-179, 2004 27. SAS Institute Inc: SAS/STAT User’s Guide, Version 8, vol 2, Cary, NC, SAS Institute Inc, 2000, p 1452 28. Ofsthun N, Labrecque J, Lacson E, Keen M, Lazarus JM: The effects of higher hemoglobin levels on mortality and hospitalization in hemodialysis patients. Kidney Int 63:1908-1914, 2003 29. Clyne N, Jogestrand T: Effect of erythropoietin treatment on physical exercise capacity and on renal function in predialytic uremic patients. Nephron 60:390-396, 1992 30. Foley RN, Parfrey PS, Morgan J, et al: Effect of hemoglobin levels in hemodialysis patients with asymptomatic cardiomyopathy. Kidney Int 58:1325-1335, 2000 31. Hampl H, Sternberg C, Berweck S, et al: Regression of left ventricular hypertrophy in hemodialysis patients is possible. Clin Nephrol 58:S73-S96, 2002 (suppl 1) 32. Silverberg DS, Wexler D, Blum M, et al: The effect of correction of anaemia in diabetics and non-diabetics with severe resistant congestive heart failure and chronic renal failure by subcutaneous erythropoietin and intravenous iron. Nephrol Dial Transplant 18:141-146, 2003 33. Xue JL, St Peter WL, Ebben JP, Everson SE, Collins AJ: Anemia treatment in the pre-ESRD period and associated mortality in elderly patients. Am J Kidney Dis 40:11531161, 2002

PISONI ET AL

34. Valderrabano F, Horl WH, Macdougall IC, Rossert J, Rutkowski B, Wauters JP: Pre-dialysis survey on anaemia management. Nephrol Dial Transplant 18:89-100, 2003 35. Collins AJ: Anaemia management prior to dialysis: Cardiovascular and cost-benefit observations. Nephrol Dial Transplant 18:S2-S6, 2003 (suppl 2) 36. Furuland H, Linde T, Ahlmen J, Christensson A, Strombom U, Danielson BG: A randomized controlled trial of haemoglobin normalization with epoetin alfa in predialysis and dialysis patients. Nephrol Dial Transplant 18: 353-361, 2003 37. Walters BA, Hays RD, Spritzer KL, Fridman M, Carter WB: Health-related quality of life, depressive symptoms, anemia, and malnutrition at hemodialysis initiation. Am J Kidney Dis 40:1185-1194, 2002 38. Owen WF Jr: Patterns of care for patients with chronic kidney disease in the United States: Dying for improvement. J Am Soc Nephrol 14:S76-S80, 2003 (suppl 2) 39. Reddan DN, Frankenfield DL, Klassen PS, et al: Regional variability in anaemia management and haemoglobin in the US. Nephrol Dial Transplant 18:147-152, 2003 40. Combe Ch, Pisoni RL, Port FK, et al: Dialysis Outcomes and Practice Patterns Study: Donne´ es sur l’utilisation des cathe´ ters veineux centraux en he´ modialyse chronique. Nephrologie 22:379-384, 2001 41. Pisoni RL, Young EW, Combe C, et al: Higher catheter use within facilities is associated with increased mortality and hospitalization: Results from DOPPS. J Am Soc Nephrol 12:299A, 2001 (abstr) 42. Nassar GM, Fishbane S, Ayus JC: Occult infection of old nonfunctioning arteriovenous grafts: A novel cause of erythropoietin resistance and chronic inflammation in hemodialysis patients. Kidney Int Suppl 80:S49-S54, 2002 43. Ministry of Health and Welfare: Revised Point Fee System for Social Insurance (in Japanese), Tokyo, Japan, Shakai-hoken Kenkyujo, 2002 44. Canaud B, Bennhold I, Delons S, et al: What is the optimum frequency of administration of r-HuEPO for correcting anemia in hemodialysis patients? Dial Transplant 24:306329, 1995 45. Derose SF, Diesto JT, Crooks PW: Sustainability of subcutaneous erythropoietin dosing. J Am Soc Nephrol 12:325A, 2001 (abstr) 46. Jensen JD, Madsen JK, Jensen LW: Comparison of dose requirement, serum erythropoietin and blood pressure following intravenous and subcutaneous erythropoietin treatment of dialysis patients. Eur J Clin Pharmacol 50:171-177, 1996 47. Kaufman J, Reda DJ, Fye CL, et al: Subcutaneous compared with intravenous epoetin in patients receiving hemodialysis. N Engl J Med 339:578-583, 1998 48. Parker KP, Mitch WE, Stivelman JC, et al: Safety and efficacy of low-dose subcutaneous dosing of epoetin alfa in hemodialysis patients. J Am Soc Nephrol 8:288-293, 1997 49. Stockenhuber F, Liobl U, Jahn C, et al: Intravenous versus subcutaneous application of erythropoietin in patients on RDT and CAPD. Meeting of the American Society of Nephrology. Kidney Int 37:379A, 1990 (abstr)

ANEMIA MANAGEMENT AND OUTCOMES IN 12 COUNTRIES

50. Moist L, House A, Wazny L, Muirhead N: Switching from SC to IV EPO: No increase in dose required. J Am Soc Nephrol 14:265A, 2003 (abstr) 51. Goodkin DA, Bragg-Gresham JL, Koenig KG, et al: Association of comorbid conditions and mortality in hemodialysis patients in Europe, Japan, and the United States: The Dialysis Outcomes and Practice Patterns Study (DOPPS). J Am Soc Nephrol 14:3270-3277, 2003

111

52. Pisoni RL, Young EW, Dykstra DM, et al: Vascular access use in Europe and the United States: Results from the DOPPS. Kidney Int 61:305-316, 2002 53. Rayner HC, Pisoni RL, Bommer J, et al: Mortality and hospitalization in haemodialysis patients in five European countries: Results from the Dialysis Outcomes and Practice Patterns Study (DOPPS). Nephrol Dial Transplant 19:108-120, 2004