International Comparisons to Assess Effects of Payment and Regulatory Changes in the United States on Anemia Practice in Patients on Hemodialysis: The Dialysis Outcomes and Practice Patterns Study

Abstract

For years, erythropoiesis-stimulating agent (ESA) use among patients on dialysis was much higher in the United States than in Europe or Japan. Sweeping changes to dialysis reimbursement and regulatory policies for ESA in the United States in 2011 were expected to reduce ESA use and hemoglobin levels. We used the Dialysis Outcomes and Practice Patterns Study (DOPPS) data from 7129 patients in 223 in–center hemodialysis facilities (average per month) to estimate and compare time trends in ESA dose and hemoglobin levels among patients on hemodialysis in the United States, Germany, Italy, Spain, the United Kingdom, and Japan. From 2010 to 2013, substantial declines in ESA use and hemoglobin levels occurred in the United States but not in other DOPPS countries. Between August of 2010 and April of 2013, mean weekly ESA dose in the United States decreased 40.4% for black patients and 38.0% for nonblack patients; mean hemoglobin decreased from 11.5 g/dl in black patients and 11.4 g/dl in nonblack patients to 10.6 g/dl in both groups. In 2010 and 2013, adjusted weekly ESA doses per kilogram were 41% and 11% lower, respectively, in patients in Europe and 60% and 18% lower, respectively, in patients in Japan than in nonblack patients in the United States. Adjusted hemoglobin levels in 2010 and 2013 were 0.07 g/dl lower and 0.56 g/dl higher, respectively, in patients in Europe and 0.93 and 0.01 g/dl lower, respectively, in patients in Japan than in nonblack patients in the United States. In conclusion, ESA dosing reductions in the United States likely reflect efforts in response to changes in reimbursement policy and regulatory guidance.

Deficient renal erythropoietin production is a major factor contributing to the anemic status that affects the vast majority of patients on dialysis.¹ Treatment with erythropoiesis-stimulating agents (ESAs) and intravenous (iv) iron has been the cornerstone of anemia management for over two decades. Since the late 1990s, the international Dialysis Outcomes and Practice Patterns Study (DOPPS) has shown differences in anemia management among patients on dialysis in the United States compared with 11 other countries, with hemoglobin (Hb) levels equal to or higher than, and ESA doses substantially exceeding those in other countries.^2,3 Although these differences might be attributable, in part, to greater ESA resistance in United States patients on dialysis than in patients elsewhere,³ for many years, ESAs in the United States were reimbursed on a per unit dose basis, thus providing a direct financial incentive for higher ESA doses. Historically higher ESA doses among for–profit United States dialysis centers also supported the possibility that financial incentives influenced ESA dose.⁴ Conversely, no direct financial or regulatory incentives were present in other countries.^3,5

Clinical trial evidence published in 2006 (e.g., the Correction of Hemoglobin and Outcomes in Renal Insufficiency [CHOIR] Trial ⁶ and the Cardiovascular Risk Reduction by Early Anemia Treatment with Epoetin-β [CREATE] Trial⁷; later, the Trial to Reduce Cardiovascular Events with Aranesp Therapy (TREAT)⁸ in 2009) raised serious concerns about ESA safety. During a 2007 Congressional subcommittee hearing⁹ on the matter, data from the US Renal Data System were shared that indicated that more than one half of United States patients on dialysis had Hb levels above the US Food and Drug Administration¹⁰ (FDA) target range of 10–12 g/dl, and the US Government Accounting Office¹¹ (GAO) presented a 2006 report recommending implementation of an expanded bundled payment for dialysis services including renal medications and laboratory services. Subsequent regulatory action culminated in three important changes that occurred in 2011. In January of 2011, the US Centers for Medicare and Medicaid Services (CMS) introduced a major modification to the ESRD Prospective Payment System (PPS) that was fully implemented nearly 90% of United States dialysis units, with the remainder opting to phase in over 4 years. The revised PPS added approximately $3 billion worth of separately billable services to the bundled payment, of which dose-based reimbursement for ESAs and iv iron accounted for approximately 67% and 8%, respectively.¹² In June of 2011, the FDA approved a revision to the ESA prescribing information that removed the Hb target range of 10–12 g/dl for patients with CKD, instead advising initiation of ESA therapy at Hb<10 g/dl and reduction or interruption of treatment in patients on dialysis when Hb nears or exceeds 11 g/dl.¹³ The updated label cited three studies assessing the effect of different Hb targets, which showed higher mortality rates in dialysis (Normal Hematocrit Trial; target 14 versus 10 g/dl) and CKD (CHOIR; target 13.5 versus 11.3 g/dl) and higher stroke rates in CKD (TREAT; target 13 versus ≥9 g/dl) among patients randomized to the higher targets.^6,8,14,15 Finally, in July of 2011, the CMS proposed a modification to the Quality Incentive Payment (QIP), which was approved in November of 2011, that removed the payment penalty for Hb<10 g/dl but retained the penalty for Hb>12 g/dl.^16,17

The GAO, the American Society of Nephrology, and others anticipated that financial incentives resulting from the PPS would favor lower ESA dosing and greater use of less expensive iv iron.^18–20 Combined with the bundled reimbursement for ESA, the removal of specific floor Hb levels in the updated ESA label and QIP proposed rule increased the likelihood that Hb levels would also decline.²¹ Indeed, the Dialysis Outcomes and Practice Patterns Study Practice Monitor (DPM) reported substantial declines in average Hb levels and ESA doses in the United States between 2010 and 2012 along with a temporary increase in iv iron use and a sustained rise in clinical measures of iron stores.²²

Because no significant anemia–related policy changes were implemented in other DOPPS countries during this same time period, in this study, we compare recent trends in anemia practices and outcomes in the United States, Europe, and Japan. We also evaluate whether and to what extent international differences in ESA doses over time are explained by differences in patient characteristics known to influence ESA dose and other aspects of anemia management.

Results

Study Sample

Mean monthly patient (facility) counts were 2023 (70), 1664 (57), and 3442 (96) for Europe, Japan, and the United States, respectively (Supplemental Table 1). During the study period, modest declines were observed for mean vintage (Europe and Japan), ESA delivery by iv route (Japan), and catheter use (United States black patients compared with arteriovenous fistula or graft) (Table 1). The European sample also tended over time toward older age and proportionally more men, whereas United States black patients tended toward higher serum albumin levels and predialysis weight.

Table 1.

Means or percentages of selected patient characteristics by region and calendar time

Month/Yeara	Europeb		Japan		United States Black		United States Nonblack
	N	Mean (%)	N	Mean (%)	N	Mean (%)	N	Mean (%)
Age, yr
08/2010	2186	65.8	1708	65.0	1053	58.6	2451	65.6
12/2011	2103	66.3	1672	65.5	1261	59.0	2794	64.8
04/2013	2011	66.6	1681	65.6	1099	58.5	2195	64.6
P value for trend		0.07		0.14		0.95		0.18
Vintage, yr
08/2010	2160	5.2	1702	8.4	1046	4.6	2433	3.5
12/2011	2084	5.4	1664	8.8	1252	4.8	2777	3.8
04/2013	1995	5.2	1672	8.0	1082	4.7	2167	3.6
P value for trend		0.001		<0.001		0.10		0.18
Men, %
08/2010	2186	59.5	1707	64.1	1053	55.5	2451	56.1
12/2011	2103	59.6	1671	64.0	1261	56.1	2794	55.4
04/2013	2003	61.8	1678	64.5	1098	54.7	2196	56.3
P value for trend		0.05		0.56		0.87		0.88
Predialysis weight, kg
08/2010	2040	73.9	1606	56.7	1024	86.6	2304	81.4
12/2011	1316	74.3	1595	57.7	1179	87.0	2508	81.8
04/2013	1577	75.4	1569	58.2	1050	88.2	2051	81.0
P value for trend		0.13		0.04		0.02		0.96
Diabetes, %
08/2010	2154	34.7	1698	38.9	1042	61.5	2385	66.3
12/2011	1370	35.7	1659	38.9	1213	62.1	2610	67.4
04/2013	1657	35.0	1596	41.8	1081	60.0	2135	63.0
P value for trend		0.83		0.11		0.76		0.10
Gastrointestinal bleeding in prior 12 mo
08/2010	2142	3.8	1681	4.3	991	3.4	2298	2.9
12/2011	1360	4.6	1642	4.0	1123	3.3	2466	2.8
04/2013	1627	4.1	1573	3.9	931	2.9	1888	2.8
P value for trend		0.67		0.62		0.33		0.48
Serum albumin, g/dl
08/2010	1662	3.81	1470	3.67	1034	3.86	2369	3.82
12/2011	1221	3.74	1533	3.74	1177	3.94	2623	3.87
04/2013	1301	3.72	1518	3.72	970	3.90	1869	3.87
P value for trend		0.24		0.65		0.06		0.31
Serum creatinine, mg/dl
08/2010	1511	8.28	1536	10.93	1009	9.76	2253	7.54
12/2011	1194	8.02	1577	10.65	1072	9.52	2438	7.64
04/2013	1366	8.16	1500	10.56	915	9.64	1707	7.74
P value for trend		0.89		0.16		0.70		0.20
Catheter, %
08/2010	2126	19.1	1564	0.2	982	16.8	2270	16.8
12/2011	1640	16.9	1550	0.2	1157	18.0	2546	16.5
04/2013	1841	19.8	1584	0.2	1069	11.4	2149	12.3
P value for trend		0.46		0.40		0.001		0.30
ESA route, % iv
08/2010	1792	90.1	1271	99.8	849	99.4	1939	98.8
12/2011	1624	87.5	1276	99.6	945	99.0	2000	94.9
04/2013	1445	90.6	1348	96.1	898	99.9	1660	97.4
P value for trend		0.53		0.004		0.53		0.72

P values for trend obtained from models adjusted for monthly seasonality.

^aFacility sample transitioned from the DOPPS 4 to the DOPPS 5 during January to April of 2012.

^bEurope includes the DOPPS facility samples in Germany, Italy, Spain, and the United Kingdom.

Long-Term Trends: 2002–2013

Mean prescribed ESA dose among ESA-treated patients in the United States rose steadily from 2002 to 2005, reached a plateau between 2006 and 2010, and then, sharply declined from 2010 to 2013 (Figure 1A). By contrast, mean prescribed ESA dose in Europe rose from 2002 to 2005 but was stable between 2006 and 2012. In Japan, mean prescribed ESA dose declined between 2002 and 2006 but then, increased slightly through 2013.

Figure 1.

Temporal trends (2002–2013) in prescribed ESA dose and Hb level among patients managed with ESAs by region. (A) Prescribed ESA dose and (B) Hb level among patients managed with ESAs by region. Values were reported at seven indicated time points using data from the DOPPS 2 (2002), the DOPPS 3 (2005 and 2006), the DOPPS 4 (2010 and 2011), and the DOPPS 5 (2012 and 2013).

Mean Hb levels among ESA-treated patients in the United States and Europe reached their peaks in 2005 and 2006, respectively (Figure 1B). Although mean Hb levels in the United States were substantially higher than those in Europe from 2002 to 2006, Hb levels declined in both regions and reached parity in 2010–2011. Hb levels in Japan increased steadily from 2002 to 2010, and by 2013, they nearly reached the mean Hb level in the United States.

ESA Prescription and Dose Trends: August of 2010 to April of 2013

ESA prescription in a 3-month period increased from 87.4% to 90.8% in Japan but fell from 89.6% to 88.0% in Europe, from 94.5% to 92.3% in United States black patients, and from 95.5% to 89.9% in United States nonblack patients (Supplemental Figure 1A). Seasonality–adjusted absolute differences during this period were 3.7% (95% confidence interval [95% CI], 1.2 to 6.2) in Japan, −1.7% (95% CI, −4.3 to 1.0) in Europe, −2.2% (95% CI, −4.9 to 0.4) in United States black patients, and −3.5% (95% CI, −6.1 to −1.0) in United States nonblack patients.

The mean (median) prescribed ESA dose in a 3-month period increased from 5171 (3933) to 5848 (5000) units/wk in Japan but fell from 9272 (7500) to 8216 (6249) units/wk in Europe, from 19,865 (14,250) to 13,834 (8655) units/wk in United States black patients, and from 18,392 (12,467) to 11,046 (7311) units/wk in United States nonblack patients (Figure 2A, Table 2, Supplemental Table 2). Seasonality–adjusted relative differences during this period were 19.4% (95% CI, 5.9 to 34.7) in Japan, −6.4% (95% CI, −17.7 to 6.5) in Europe, −40.4% (95% CI, −50.6 to −28.0) in United States black patients, and −38.0% (95% CI, −44.9 to −30.2) in United States nonblack patients. Results were similar in analyses restricted to iv epoetin alfa doses, analyses expressing ESA dose per unit of Hb (Supplemental Figure 1B) or per kilogram predialysis body weight (Figure 2B), and analyses using alternate conversion ratios for darbepoetin (200:1) and subcutaneous epoetin alfa (1.3:1).

Figure 2.

Temporal trends (by region and month [August 2010 to April 2013]) in weekly prescribed ESA dose, weekly prescribed ESA dose per kilogram, Hb level among ESA-treated patients, and proportion of patients prescribed iv iron. (A) Weekly prescribed ESA dose, (B) weekly prescribed ESA dose per kilogram, (C) Hb level among ESA-treated patients, and (D) proportion of patients prescribed iv iron. Markers indicate monthly observed means. Regional trend lines were estimated using a restricted cubic spline model with seven knots. Vertical lines indicate significant changes to United States dialysis reimbursement policy or regulatory updates. PPS indicates the January of 2011 New Medicare ESRD PPS. QIP indicates the Medicare ESRD QIP program proposed rule removing Hb<10 g/dl as a measure. ESA dose conversions: subcutaneous epoetin, ×1.15; darbepoetin (iv or subcutaneous), ×250 units/μg; and pegylated epoetin-β (iv or subcutaneous), ×208 units/μg. ¹Introduction of the revised PPS in January of 2011. ²ESA label revision in June of 2011. ³ESRD QIP proposed rule in July of 2011.

Table 2.

Prescribed ESA dose quantiles by region

Month/Yeara	Scaled by Predialysis Weight						Not Scaled
	N	10th	25th	50th	75th	90th	N	10th	25th	50th	75th	90th
Europeb
08/2010	1528	33	56	101	175	274	1628	2500	4000	7500	11,997	19,663
12/2011	881	31	52	101	173	303	1419	2236	4000	6944	11,854	20,000
04/2013	1158	30	54	90	144	232	1338	2297	3900	6249	10,000	15,000
Japan
08/2010	832	26	41	76	124	193	872	1500	2500	3933	6525	10,000
12/2011	973	24	43	76	131	190	1020	1404	2500	4315	7500	10,000
04/2013	1092	30	51	88	136	192	1111	1774	2924	5000	7500	10,000
United States black
08/2010	825	49	88	167	317	554	833	4230	7229	14,250	25,230	46,087
12/2011	872	37	64	131	227	396	916	2933	4769	9523	17,415	30,782
04/2013	813	35	56	107	222	401	839	2918	4604	8655	18,262	31,751
United States nonblack
08/2010	1835	42	88	159	292	549	1883	3299	6522	12,467	23,077	43,786
12/2011	1744	36	61	115	214	361	1912	2933	4597	8416	16,025	27,526
04/2013	1481	24	46	94	176	328	1540	1983	3640	7311	13,735	24,334

Values expressed as iv epoetin–equivalent units per week were calculated using the following conversions: subcutaneous epoetin (×1.15), darbepoetin (×250 units/μg), and pegylated epoetin-β (×208 units/μg).

^aFacility sample transitioned from the DOPPS 4 to the DOPPS 5 during January to April of 2012.

^bEurope includes the DOPPS facility samples in Germany, Italy, Spain, and the United Kingdom.

Between August of 2010 and April of 2013, adjusted differences in prescribed ESA dose scaled by predialysis body weight compared with United States nonblack patients (Table 3, scaled by predialysis body weight) declined from −41% (95% CI, −51% to −30%) to −11% (95% CI, −25% to 6%) in Europe, declined from −60% (95% CI, −66% to −51%) to −18% (95% CI, −33% to 0%) in Japan, and increased from 2% (95% CI, −11% to 16%) to 17% (95% CI, 4% to 32%) in United States black patients. These differences were slightly larger in unscaled analyses (Table 3, not scaled) and analyses expressing ESA dose per unit of Hb (not shown).

Table 3.

Percentage differences (with 95% CIs) in mean prescribed ESA dose in Europe, Japan, and United States black patients compared with United States nonblack patients by calendar time and model used to adjust for covariates

Month/Yeara	Scaled by Predialysis Body Weight			Not Scaled
	Unadjusted	Model 1b	Model 2c	Unadjusted	Model 1b	Model 2c	Model 3d
Europee
08/2010	−37% (−46% to −28%)	−41% (−49% to −32%)	−41% (−51% to −30%)	−43% (−50% to −35%)	−44% (−51% to −36%)	−46% (−54% to −37%)	−45% (−53% to −36%)
12/2011	−17% (−29% to −2%)	−28% (−39% to −15%)	−21% (−35% to −6%)	−22% (−31% to −11%)	−27% (−38% to −14%)	−25% (−37% to −10%)	−22% (−35% to −8%)
04/2013	−4% (−17% to 10%)	−13% (−24% to 0%)	−11% (−25% to 6%)	−11% (−22% to 2%)	−16% (−26% to −4%)	−17% (−29% to −4%)	−15% (−27% to −2%)
Japan
08/2010	−54% (−61% to −47%)	−58% (−64% to −51%)	−60% (−66% to −51%)	−68% (−72% to −63%)	−69% (−73% to −65%)	−71% (−75% to −66%)	−70% (−75% to −65%)
12/2011	−36% (−45% to −26%)	−41% (−49% to −31%)	−39% (−50% to −26%)	−53% (−60% to −46%)	−55% (−62% to −48%)	−56% (−64% to −47%)	−56% (−63% to −46%)
04/2013	−9% (−20% to 4%)	−13% (−25% to 1%)	−18% (−33% to −0%)	−34% (−42% to −26%)	−36% (−44% to −26%)	−43% (−52% to −31%)	−41% (−51% to −29%)
United States black
08/2010	4% (−9% to 20%)	2% (−11% to 17%)	2% (−11% to 16%)	13% (−1% to 28%)	7% (−6% to 22%)	7% (−5% to 21%)	6% (−6% to 20%)
12/2011	8% (−4% to 20%)	5% (−8% to 19%)	3% (−10% to 18%)	10% (−2% to 23%)	4% (−8% to 19%)	3% (−10% to 18%)	2% (−10% to 17%)
04/2013	21% (7% to 37%)	19% (6% to 34%)	17% (4% to 32%)	30% (14% to 48%)	25% (11% to 42%)	23% (9% to 40%)	23% (8% to 39%)

Doses expressed as iv epoetin–equivalent units per week were calculated using the following conversions: subcutaneous epoetin (×1.15), darbepoetin (×250 units/μg), and pegylated epoetin-β (×208 units/μg).

^aFacility sample transitioned from the DOPPS 4 to the DOPPS 5 during January to April of 2012.

^bModel 1 includes age, years on dialysis, black race (United States only), sex, catheter use, diabetes, history of gastrointestinal bleeding, creatinine, and most recent albumin level in the prior 3 months.

^cModel 2 includes model 1 adjustments plus iv iron use in the prior 3 months (yes or no), most recent transferrin saturation and ferritin levels in the prior 3 months, and facility–reported upper Hb target.

^dModel 3 includes model 2 adjustments plus predialysis body weight.

^eEurope includes the DOPPS facility samples in Germany, Italy, Spain, and the United Kingdom.

Hb Trends: August of 2010 to April of 2013

The mean Hb level among ESA-treated patients changed little from 10.4 to 10.6 g/dl in Japan and from 11.3 to 11.2 g/dl in Europe but fell from 11.5 to 10.6 g/dl in United States black patients and from 11.4 to 10.6 g/dl in United States nonblack patients (Figure 2C). Seasonality–adjusted absolute differences during this period were 0.10 g/dl (95% CI, −0.06 to 0.25) in Japan, −0.19 g/dl (95% CI, −0.30 to −0.09) in Europe, −0.91 g/dl (95% CI, −1.02 to −0.79) in United States black patients, and −0.84 g/dl (95% CI, −0.94 to −0.74) in United States nonblack patients.

Between August of 2010 and April of 2013, the unadjusted mean Hb differences compared with United States nonblack patients increased from −0.10 g/dl (95% CI, −0.23 to 0.03) to 0.53 g/dl (95% CI, 0.41 to 0.65) in Europe; it decreased from −1.01 g/dl (95% CI, −1.20 to −0.82) to −0.05 g/dl (95% CI, −0.19 to 0.09) in Japan and from 0.11 g/dl (95% CI, 0.01 to 0.21) to −0.03 g/dl (95% CI, −0.13 to 0.06) in United States black patients (Table 4).

Table 4.

Differences in mean Hb in Europe, Japan, and United States black patients compared with United States nonblack patients by calendar time and model used to adjust for covariates

Month/Yeara	Unadjusted	95% CI	Model 1b	95% CI	Model 2c	95% CI	Model 3d	95% CI
Europee
08/2010	−0.10	−0.23 to 0.03	−0.08	−0.21 to 0.05	−0.08	−0.23 to 0.08	−0.07	−0.22 to 0.09
12/2011	0.55	0.39 to 0.70	0.60	0.45 to 0.75	0.57	0.39 to 0.75	0.58	0.40 to 0.76
04/2013	0.53	0.41 to 0.65	0.58	0.44 to 0.72	0.56	0.38 to 0.74	0.56	0.38 to 0.74
Japan
08/2010	−1.01	−1.20 to −0.82	−0.94	−1.14 to −0.74	−0.97	−1.19 to −0.75	−0.93	−1.15 to −0.72
12/2011	−0.20	−0.35 to −0.05	−0.15	−0.32 to 0.02	−0.17	−0.38 to 0.03	−0.14	−0.34 to 0.06
04/2013	−0.05	−0.19 to 0.09	−0.01	−0.17 to 0.14	−0.04	−0.25 to 0.16	−0.01	−0.21 to 0.19
United States black
08/2010	0.11	0.01 to 0.21	0.14	0.04 to 0.24	0.13	0.04 to 0.23	0.13	0.04 to 0.22
12/2011	−0.06	−0.23 to 0.10	−0.05	−0.21 to 0.12	−0.03	−0.19 to 0.13	−0.03	−0.19 to 0.12
04/2013	−0.03	−0.13 to 0.06	−0.01	−0.09 to 0.08	0.00	−0.09 to 0.08	−0.01	−0.09 to 0.07

^aFacility sample transitioned from the DOPPS 4 to the DOPPS 5 during January to April of 2012.

^bModel 1 includes age, years on dialysis, black race (United States only), sex, catheter use, diabetes, history of gastrointestinal bleeding, creatinine, and most recent albumin level in the prior 3 months.

^cModel 2 includes model 1 adjustments plus iv iron use in the prior 3 months (yes or no), most recent transferrin saturation and ferritin levels in the prior 3 months, and facility–reported upper Hb target.

^dModel 3 includes model 2 adjustments plus predialysis body weight.

^eEurope includes the DOPPS facility samples in Germany, Italy, Spain, and the United Kingdom.

Iron Prescription Trends: August of 2010 to April of 2013

The proportion of patients prescribed iv iron in a 3-month period increased from 75.4% to 75.8% in Europe, from 36.3% to 40.7% in Japan, from 73.5% to 82.9% in United States black patients, and from 72.0% to 82.1% in United States nonblack patients (Figure 2D). Seasonality–adjusted absolute differences during this period were −1.1% (95% CI, −4.5 to 2.2) in Europe, 7.9% (95% CI, −2.2 to 18.0) in Japan, 7.4% (95% CI, 1.4 to 13.3) in United States black patients, and 10.7% (95% CI, −1.6 to 22.9) in United States nonblack patients. Median monthly iv iron doses (among treated patients) in the United States remained fairly stable during this period (approximately 50 mg/wk), and oral iron use was <5% in all regions (not shown).

Anemia Treatment Targets: 2010–2013

Among United States study sites, 94% reported an upper Hb target of 12 g/dl in 2010, but they were evenly split between 11 and 12 g/dl (46% each) in 2011. In 2013, targets of 11, 11.1–11.9, and 12 g/dl were reported by 62%, 22%, and 16% of facilities, respectively (Figure 3A). During the same period, Hb upper targets of 12 g/dl declined slightly from 92% to 87% in Europe and increased gradually from 66% to 82% in Japan. Hb lower targets ≤10 g/dl increased from 70% to 96% in the United States and from 40% to 79% in Europe but remained relatively stable at 91%–94% in Japan (Figure 3B).

Figure 3.

Temporal trends in upper and lower Hb target levels reported by the DOPPS site medical directors by region and year (2010–2013). (A) Upper and (B) lower Hb target levels reported by the DOPPS site medical directors by region and year (2010–2013).

Discussion

Since the introduction of significant United States dialysis payment and regulatory changes in 2011, there have been substantial decreases in mean ESA dose and Hb levels and a modest increase in the proportion of patients receiving iv iron. Observed ESA and Hb trends in the United States from 2010 to 2013 exceed prior trends from 2002 to 2010 or trends in other DOPPS countries from 2002 to 2013. Preliminary data for 2014 and 2015 suggest that Hb levels have achieved a relative stability, whereas ESA doses have continued to decline slowly (data not shown). Notably, the most rapid decline occurred soon after the ESA label revision (in June of 2011) rather than after the PPS bundling (in January of 2011). We infer from this that the United States dialysis community did not primarily change anemia practice for cost reasons alone but changed in the context of supportive clinical guidance from the FDA–approved ESA label change. In contrast, we observed rising ESA dose, iv iron prescription, and Hb levels in Japan and mostly flat trends in four European countries.

Before the US PPS revision, the maximum Medicare–reimbursable ESA dose was 400,000 units/mo, and doses of that magnitude were not uncommon in highly ESA–resistant patients. With the inclusion of ESAs into the bundled payment, the DPM indicates that many dialysis facilities in the United States may be voluntarily limiting prescribed ESA doses at much lower maximum levels, with 90% percentile doses now around 30,000 units/wk, suggesting a dramatic change in treatment strategy for these patients.²² However, ESA doses at or approaching even this reduced ceiling dose remain uncommon in Europe, where many countries allow separate billing for ESA therapy, and are absent in Japan, where the maximum reimbursable dose (determined from prescribing information on the ESA label) is 9000 units/wk.

Similar changes to reimbursement and payment–linked quality initiatives in other countries seem to have had a more modest effect on anemia management than in the United States. In Germany, an expanded, weekly flat–rate dialysis payment was introduced in 2002, and a quality assurance program—proposed in 2006 and implemented in 2009—included administrative penalties for facilities failing to achieve targets for Hb levels and three other nonanemia measures.^23,24 Kleophas et al.²⁴ reported that the proportion with Hb<10 g/dl remained steady at approximately 10% from 2006 to 2011 along with an approximately 17% decline in ESA dose and an approximately 20% increase in iv iron dose. However, these results should be viewed in the context of declining ESA use trends in Europe because of a changing balance of clinical evidence after publication of results from the CHOIR and CREATE clinical trials in 2006.^6,7

In 2006, Japan introduced a bundled payment for ESA within the overall reimbursement for dialysis services. Hasegawa et al.²⁵ showed that this shift in reimbursement policy was associated with an increase of 9.6% of patients receiving iv iron and an 11.8% reduction in mean EPO dose between January of 2006 and January of 2007. The percentage reduction in mean ESA dose in Japan was much less than in the United States after ESA bundling in 2011; a possible explanation is that prebundle ESA doses in Japan were already quite low and could be reduced little to still maintain an erythropoietic effect.

ESA doses, which had been twice as high in the United States than elsewhere to support a given Hb level, are now only modestly higher than in some European countries, indicating that, in contrast to what may have been previously inferred, clinical ESA requirements for patients differ little across these regions. Turenne et al.²⁶ reported comparable relative reductions in mean ESA doses adjusted for Hb among black and nonblack patients in the United States subsequent to the 2011 policy changes, indicating that ESA doses to support a given Hb level could be reduced in both groups. This study extends this result, showing that adjustment for an array of common patient case–mix factors (including iv iron use and ferritin levels) did not explain much of the observed differences in prescribed ESA dose among the DOPPS regions at several time points (Table 3). Together, these data support the speculation by McFarlane et al.³ that between-country differences in anemia management have been related primarily to clinical guidelines and reimbursement policies rather than specific physiologic requirement.

Indeed, inter-regional differences in treatment targets for Hb level have been shrinking. Before 2004, anemia treatment targets in Japan were based informally on insurance pricing tables that suggested maintaining hematocrit levels at 30% (approximately 10.0 g/dl). However, the Japanese Society for Dialysis Therapy (JSDT) published anemia management guidelines in 2004 that recommended an Hb target level of 10–11 g/dl for most patients and 11–12 g/dl for younger, more active patients.²⁷ A 2008 update to the JSDT guidelines added 12 g/dl as a criterion for reducing or interrupting ESA treatment.²⁸ Although the distribution of lower Hb targets reported by Japanese DOPPS facilities has not changed much since 2010 (Figure 3B), in recent years, we have observed a higher proportion of Japanese DOPPS facilities reporting upper Hb targets of 12 g/dl (Figure 3A). In contrast, nonguideline position statements regarding anemia treatment in European patients with CKD released by the European Renal Best Practice (ERBP) group through 2010 generally suggested maintaining Hb levels between 11 and 12 g/dl.^29–31 In 2012, responding to new guidelines from the Kidney Disease Improving Global Outcomes group,³² the ERBP group cited a lack of evidence of benefit or harm for maintaining Hb values in the range of 11.5 and 13.5 g/dl and suggested that Hb levels in European populations should be individualized between 10 and 12 g/dl.³³ Between 2010 and 2013, we have shown a steady increase in the proportion of European DOPPS facilities reporting a lower Hb target of 10 g/dl compared with 10.5 or 11 g/dl (Figure 3B), with relatively little change in the proportion of facilities reporting an upper Hb target of 12 g/dl (Figure 3A).

In addition to clinical guidelines, reimbursement policies may account for differences between United States and European Hb levels in the most recent years. Reimbursement to United States dialysis providers is reduced if there is an excess of patients with Hb>12 g/dl under the CMS QIP in place during the analysis time period¹⁷ (although this measure is now to be phased out for QIP Program Year 2017), and thus, this may explain, in part, the rapid reduction of Hb upper targets to 11.0 or 11.5 g/dl to avoid penalties. By contrast, reimbursement generally is not conditioned on achievement of specific Hb levels in many European countries.³⁴ Thus, European providers may be less concerned, for reimbursement reasons, with occasional excursions of Hb levels into the ERBP gray area of 11.5–13.5 g/dl.

United States ESA dosing guidance before 2011 recommended treating patients with CKD to an Hb level between 10 and 12 g/dl (well above common thresholds for using red blood cell transfusions to correct severe anemia)^35,36 and may have led to administration of higher doses than necessary in ESA-resistant patients. The revised ESA label approved by the FDA in June of 2011 did not specify a particular Hb target, instead recommending using the lowest dose required to avoid transfusion. Dialysis provider efforts to modify treatment protocols under this new guidance and the resulting reductions in ESA doses may partly explain a short-term increase during the study period in the proportion of Medicare ESRD beneficiaries receiving a transfusion (from 2.5%–3.0% per month in 2010 to 3.2%–3.6% per month in 2012) reported elsewhere.³⁷ Concerns regarding ESA safety (e.g., cancer progression and increased stroke risk) may have influenced the use of transfusions. Transfusions incur a risk of allosensitization, which may reduce the likelihood of a suitable renal transplant match. Although the proportion of patients on dialysis undergoing kidney transplant each year is low, careful monitoring of transfusion rates, particularly among transplant candidates, remains an important priority.

This study has several strengths. The DOPPS includes a wide range of data collected in standardized fashion internationally since 2002 (or earlier), enabling the regional comparisons presented in this paper. Much of our contemporary data collection is performed by direct electronic health record download, which allows routine collection of anemia management measures and improved data fidelity. The decline in United States Hb and ESA doses after implementation of the revised PPS was first described using the DOPPS data and has since been confirmed with national data sources.^37,38

We acknowledge several study limitations. The DPM extends backward only to August of 2010, precluding a broader analysis of anticipatory practice changes in response to the PPS proposed rule during 2010. Our trend models used a discontinuity parameter to account for the low retention of United States facilities from the DOPPS 4 to the DOPPS 5; this may result in a slight underestimation of actual trends. ESA dose comparisons by region are additionally dependent on global conversion factors for subcutaneous ESA or nonepoetin alfa preparations, which may not be appropriate for all patients. However, our results were robust to a variety of plausible scenarios related to these conversions. Other dialysis practices, which are not consistently captured in the DPM data, may partially explain residual differences in ESA dose or Hb between study regions.

In summary, we report dramatic trends in United States anemia management practices and laboratory outcomes between August of 2010 and April of 2013 after updates to United States regulatory guidance and reimbursement policy in 2011. Compared with the observed trends in the United States, changes in anemia management in other DOPPS countries during this period were minimal. United States ESA dosing reductions likely reflect efforts to limit high ESA doses and accept lower Hb levels in response to changes in reimbursement policy and regulatory guidance. Despite adjusted ESA dosing levels that have recently converged, Hb levels in the United States are now lower than in Europe for the first time in more than a decade; residual differences may reflect higher prevalence or severity of ESA resistance in United States patients. Ongoing observation and research are needed to estimate any lasting effects on the patient experience and clinical outcomes.

Concise Methods

Data Sources

The DOPPS is an international, multistage prospective cohort study of patients on in-center hemodialysis and practices ongoing since 1996. Hemodialysis facilities are randomly selected in each country according to factors, such as geographic location and facility setting or dialysis organization. Basic demographic information (age, sex, race, and years on dialysis) is collected for all facility patients, and a random sample of patients on hemodialysis is selected for DOPPS enrollment with expanded longitudinal data collection, including medical history, monthly laboratory values, vascular access history, and medication lists with dosing information (details of the study design are in the works by Young et al.³⁹ and Pisoni et al.⁴⁰).

On the basis of the DOPPS 4 (2009–2011) and the DOPPS 5 (2012–2015) samples, the DPM is designed to inform the community of contemporary trends in United States dialysis treatment practices before and after implementation of the revised PPS in January of 2011. Sampling design and analytic methods have been published,⁴¹ and the DPM results have been consistent with published national data.^37,38,42 This study extends the DPM framework to other DOPPS countries with complete follow-up during the analysis period (Germany, Italy, Spain, the United Kingdom, and Japan) to describe and contrast regional trends in anemia practices and outcomes between August of 2010 and April of 2013. Cross-sections from the DOPPS 2 (2002) and the DOPPS 3 (2005 and 2006) are also provided for historical context.

Treatment practices included prescription of ESA or iv iron during the prior 3 months and prescribed ESA dose. ESAs include epoetin (epoetin alfa in the United States and epoetin alfa and other forms elsewhere), darbepoetin (available in the United States but primarily used in other countries), and pegylated epoetin-β (not available in the United States). ESA prescribed doses were obtained at the end of each study month and scaled to a 7-day weekly dose. Administered ESA doses were not available in all countries during the study period and were not analyzed. On average, the ESA dose required to maintain a given Hb level tends to be higher in black patients than in patients of other races (nonblack).^16,43,44 To limit confounding by the large proportion of black patients in the United States, results are presented separately for black and nonblack patients in the United States.

Facility–reported Hb targets were obtained from annual surveys of the DOPPS site medical directors in September to November of 2010, 2011, 2012, and 2013.

Statistical Analyses

The DOPPS facility sample transitioned from the DOPPS 4 to the DOPPS 5 (January of 2012 to April of 2012) during the analysis period; we used the DOPPS 5 sample frame as the common reference population for weighting purposes. The proportions of facilities retained from the DOPPS 4 to the DOPPS 5 were 55% in the United States, 90% in Europe, and 88% in Japan.

Subcutaneous epoetin (×1.15)³ and darbepoetin (×250 units/μg)⁴⁵ doses were converted to iv epoetin–equivalent values, and pegylated epoetin-β (×208 units/μg) doses were converted by calculating a darbepoetin-equivalent dose (dividing by 1.2)⁴⁶ and applying the darbepoetin conversion. Last observation carry–forward imputation (up to 3 months) was used to populate missing values for serum albumin, which is not always measured monthly, particularly in Europe.

Observed monthly means and proportions are overlaid with smoothed trends estimated by a restricted cubic spline with knots at August of 2010, January of 2011, June of 2011, November of 2011, April of 2012, October of 2012, and March of 2013.⁴⁷ Seasonality-adjusted differences between August of 2010 and April of 2013 with 95% CIs were estimated separately using a continuous variable for calendar year, indicator variables for each nominal calendar month (e.g., January and December), and a discontinuity allowed at January of 2012 to account for potential differences resulting from the facility sample transition from the DOPPS 4 to the DOPPS 5.

Linear mixed models to investigate regional differences in prescribed ESA dose and mean Hb at three time points (August of 2010, December of 2011, and April of 2013) were estimated using progressive levels of adjustment and a random facility intercept to account for within-facility clustering. Because body weight may influence ESA dosing practices differently across regions, separate models are shown using ESA dose scaled per kilogram predialysis body weight and covariate adjustment for predialysis body weight. A natural log transform for ESA dose was used in trend analyses where it was the outcome. Sensitivity analyses for ESA dose included restriction to iv epoetin doses only, ESA dose expressed per unit Hb, and use of alternate conversion ratios for subcutaneous epoetin (×1.3)⁴⁸ and darbepoetin (×200 units/μg). Dataset development and analyses were performed with SAS 9.3 (SAS Institute Inc., Cary, NC).

Disclosures

The Dialysis Outcomes and Practice Patterns Study Program is supported by Amgen, Inc., Kyowa Hakko Kirin, AbbVie Inc., Sanofi Renal, Baxter Healthcare, and Vifor Fresenius Medical Care Renal Pharma Ltd. Additional support for specific projects and countries is also provided by Amgen, Inc., BHC Medical, Janssen Biotech, Takeda Chemical Industries, and Kidney Foundation of Canada (for logistics support) in Canada; Hexal, DGfN, Shire, and WiNe Institute in Germany; and the Japanese Society for Peritoneal Dialysis for Peritoneal Dialysis Outcomes and Practice Patterns Study in Japan. All support is provided without restrictions on publications. R.L.P. has received speaker’s fees from Amgen, Inc., Kyowa Hakko Kirin, and Vifor Fresenius Medical Care Renal Pharma Ltd. and has served on an advisory panel for Merck GmbH. B.M.R. has received a speaker’s fee from Kyowa Hakko Kirin. The other authors declare that they have no other relevant financial interests.