Medicare Costs Associated With Arteriovenous Fistulas Among US Hemodialysis Patients

Original Investigation

Medicare Costs Associated With Arteriovenous Fistulas Among US Hemodialysis Patients Mae Thamer, Timmy C. Lee, Haimanot Wasse, Marc H. Glickman, Joyce Qian, Daniel Gottlieb, Scott Toner, and Timothy A. Pflederer Background: An arteriovenous fistula (AVF) is the recommended vascular access for hemodialysis (HD). Previous studies have not examined the resources and costs associated with creating and maintaining AVFs. Study Design: Retrospective observational study. Setting & Participants: Elderly US Medicare patients initiating hemodialysis therapy during 2010 to 2011. Predictor: AVF primary and secondary patency and nonuse in the first year following AVF creation. Outcomes: Annualized vascular access costs per patient per year. Results: Among patients with only a catheter at HD therapy initiation, only 54% of AVFs were successfully used for HD, 10% were used but experienced secondary patency loss within 1 year of creation, and 83% experienced primary patency loss within 1 year of creation. Mean vascular access costs per patient per year in the 2.5 years after AVF creation were $7,871 for AVFs that maintained primary patency in year 1, $13,282 for

S

AVFs that experienced primary patency loss in year 1, $17,808 for AVFs that experienced secondary patency loss in year 1, and $31,630 for AVFs that were not used. Similar patterns were seen among patients with a mature AVF at HD therapy initiation and patients with a catheter and maturing AVF at HD therapy initiation. Overall, in 2013, fee-forservice Medicare paid $2.8 billion for dialysis vascular access–related services, w12% of all end-stage renal disease payments.

Complete author and article information provided before references.

Limitations: Lack of granularity with certain billing codes.

© 2018 The Authors. Published by Elsevier Inc. on behalf of the National Kidney Foundation, Inc. This is an open access article under the CC BY-NC-ND license (http:// creativecommons.org/ licenses/by-nc-nd/4.0/).

Conclusions: AVF failure in the first year after creation is common and results in substantially higher health care costs. Compared with patients whose AVFs maintained primary patency, vascular access costs were 2 to 3 times higher for patients whose AVFs experienced primary or secondary patency loss and 4 times higher for patients who never used their AVFs. There is a need to improve AVF outcomes and reduce costs after AVF creation.

uccessful treatment of end-stage renal disease (ESRD) with maintenance hemodialysis (HD) is inextricably dependent on reliable access to the bloodstream, typically 3 times a week. The vascular access is the “lifeline” for the HD patient. There are 3 types of vascular access for HD: the arteriovenous fistula (AVF), arteriovenous graft (AVG), and central venous catheter (CVC). In numerous studies, including a meta-analysis of 62 cohort studies with 586,337 participants, patients dialyzing with an AVF have been observed to have less morbidity and mortality.1 Compared with patients with an AVF, those with a CVC have higher rates of cardiovascular events and all-cause and infectionrelated mortality.1 The Centers for Medicare & Medicaid Services (CMS), the principal payer of dialysis services in the United States, has actively promoted use of AVFs with the twin goals of improving health outcomes and lowering costs. Specifically, CMS created the Fistula First Initiative in 2002,2 initially with a target of AVF use in 40% of prevalent HD patients; subsequently, the goal was increased to 66%. Most recently, the CMS Quality Incentive Program provides a financial incentive for providers to increase AVF use by imposing reimbursement penalties to dialysis units based in part on AVF prevalence. Cumulatively, these vascular access initiatives have been successful, and the proportion of prevalent US HD patients using an AVF has increased from 24%3 in 1998 to 2000 to 64% in August 2016.4

AJKD Vol XX | Iss XX | Month 2018

Correspondence to T.A. Pflederer (tap@ renalcareassoc.com) Am J Kidney Dis. XX(XX): 1-9. Published online Month X, 2018. doi: 10.1053/ j.ajkd.2018.01.034

Although AVFs are the preferred form of vascular access, they are limited by high rates of nonuse, patency loss, and abandonment. In an attempt to increase use, many AVFs undergo surgical or endovascular procedures, but these are associated with higher rates of subsequent maintenance procedures and shorter overall AVF survival.5,6 Permanent AVF failure within the first year after surgical creation is reported to occur in up to 40% of patients7-10 and can result in prolonged CVC dependence and subsequent morbidity, as well as the need for additional access surgery. Despite recognition of the complexities associated with AVF creation, few studies have evaluated the costs related to AVF management in a representative US HD population. In this study, we examine the costs to Medicare of vascular access creation, maintenance, and associated complications using national claims data. Our analyses stratify patients into cohorts based on timing of AVF creation relative to HD therapy initiation and evaluate costs over 2.5 years of follow-up based on AVF clinical outcomes. Methods Study Design, Data Source, and Patient Population The observational equivalent of intention-to-treat principle guided the design of this retrospective study to compare 1

Original Investigation costs associated with different clinical scenarios for AVFs among a cohort of HD patients. We used data from the 2010 to 2013 US Renal Data System (USRDS), which includes demographic and clinical data for patients with ESRD and their institutional providers of dialysis treatment. CMS Medical Evidence Form 2728, a validated and reliable research tool,11 was used to determine vascular access status at HD therapy initiation. Starting in July 2010, all dialysis units reported vascular access in use for all HD patients on a monthly basis according to the following modifiers: V5 (CVC), V6 (AVG), or V7 (AVF). In addition to the USRDS, 1-year pre-ESRD CMS Medicare claims data were used to identify vascular access procedures conducted in the year before dialysis therapy initiation. Figure 1 shows the patient cascade using USRDS and CMS Medicare pre-ESRD data to identify 3 study cohorts based on AVF status at the time of first dialysis: cohort 1 initiated HD therapy with a mature AVF, cohort 2 initiated HD therapy with a maturing AVF, and cohort 3 initiated HD therapy with a CVC only and underwent AVF creation within 9 months after dialysis therapy initiation. All incident HD patients who were 66 years and older and initiated dialysis therapy between July 1, 2010, and June 30, 2011, were included for cohorts 1 and 2. Those who initiated dialysis therapy between April 1, 2010, and September 30, 2010, were included for cohort 3 to allow for a 9-month window after dialysis therapy initiation for

Selecon criteria: 1. Age ≥ 66 years & iniate dialysis during:

7/1/2010 - 6/30/2011 (12 mo. period)

48,713 2. VA type used at dialysis iniaon:

3. Medicare primary payor (MPP) with both part A & B coverage 4. First HD claim within 90 days 5. AVF surgery within selected me period: 6. Remove transplant or PD use anyme during study follow-up or died on day of surgery AVF status at me of first dialysis:

AVF

AVF creation. The study population was limited to patients 66 years and older to enable inclusion of AVF creations that were performed in the prior year. We used the CMS 2728 form to identify vascular access used at dialysis therapy initiation. An index AVF creation for each patient was identified using Current Procedural Terminology, 4th Edition codes 36818, 36819, 36820, and 36821 and International Classification of Diseases, Ninth Revision (ICD-9) procedure code 39.27. For cohorts 1 and 2, the most recent AVF creation in the year before dialysis therapy initiation was identified; because codes 36818, 36819, and 36820 can be billed for the second surgery of a 2-stage AVF creation, we used a prior AVF creation as the index procedure if it occurred within the prior 3 months. For cohort 3, only patients who had an AVF creation within 9 months after dialysis therapy initiation were included. This index AVF creation is the start of follow-up for collecting cost data for each patient. Additional patient exclusion criteria related to insurance coverage and other issues are presented in Fig 1. The final study populations were 2,704 for cohort 1, 3,530 for cohort 2, and 3,901 for cohort 3. Table S1 provides the definition and sample size for the 3 AVF patient cohorts based on timing of AVF creation and by primary and secondary patency loss and AVF nonuse. Because we used encrypted patient information and reported aggregate data, we did not require research ethics committee approval. Informed consent was also waived due to deidentified information.

7/1/2010 - 6/30/2011 (12 mo. period)

4/1/2010 - 9/30/2010 (6 mo. period)

48,713

24,017

CVC w/ maturing AVF

CVC only no maturing AVF/AVG

7,719

7,685

15,099

4,709

4,746

9,658

4,583

4,640

9,142

1 year prior to dialysis

1 year prior to dialysis

9 months aer dialysis iniaon

2,845

3,664

4,040

2,704

3,530

3,901

Cohort 1—

Cohort 2—

Cohort 3—

Mature AVF

Maturing AVF

AVF not yet created

Figure 1. Patient cascade using US Renal Data System and Medicare claims pre–end-stage renal disease data to identify 3 cohorts based on arteriovenous fistula (AVF) status at time of first dialysis. Abbreviations: AVG, arteriovenous graft; CVC, central venous catheter; HD, hemodialysis; PD, peritoneal dialysis; VA, vascular access.

2

AJKD Vol XX | Iss XX | Month 2018

Original Investigation Identification of Vascular Access Claims An expert panel of vascular access surgeons and nephrologists (all co-authors on this article) identified a complete list of all dialysis vascular access–related claims and codes. Vascular access procedures were organized into 5 categories: noninvasive diagnostic imaging, open surgical procedures, invasive imaging and endovascular interventions, inpatient admissions, and anesthesiology associated with vascular access procedures (Item S1). To ensure that inpatient admissions were related to vascular access, we included only Medical Severity Diagnostic Related Groups 314 to 316 and 264 or Septicemia-related Medical Severity Diagnostic Related Groups 870 to 872 billed with specific ICD-9 diagnosis codes in Item S1. Similarly, to ensure that anesthesia for access to the central venous system was used for vascular access and not for other indications, it had to be accompanied with a diagnosis for ESRD (585.*). Ascertainment of Outcomes AVF use was defined as at least 1 month in which an AVF was recorded on ESRD billing claims as the type of vascular access in use using monthly V codes as follows. For cohort 1, AVF use was examined in months 1 to 3 after dialysis therapy initiation; for cohort 2, AVF use was examined in months 1 to 5 after dialysis therapy initiation; and for cohort 3, AVF use was examined within 6 months after AVF surgery. AVF nonuse was defined as no V code indicating AVF use in the periods described previously. Maintain primary patency was defined as no surgical or endovascular procedure in the year following AVF surgical creation. Procedures included angioplasty, thrombectomy, revision, ligation, banding, and embolization of accessory veins (panel a of Table S2). Primary patency loss was defined as at least 1 surgical or endovascular procedure performed on the AVF within 1 year of surgical creation. Secondary patency loss was defined as 3 months of continuous CVC use after first AVF use or creation of a subsequent permanent vascular access (panel b of Table S2) within 1 year of the index AVF creation date. Clinical Scenarios and Study Follow-up In this study, we compare vascular access costs for 3 cohorts based on timing of AVF creation and 4 clinical scenarios in the first year after surgical creation: (1) maintain primary patency, (2) primary patency loss, (3) secondary patency loss after AVF use, and (4) AVF nonuse. Vascular access costs, including the cost of the index AVF creation, are calculated for years 1 and 2 after the index AVF creation date and total 2.5 years after initial AVF creation. Analytic Approach Baseline characteristics were compared using analysis of variance and Kruskal-Wallis test for continuous variables and χ 2 test for categorical variables. The frequency of AVF use for dialysis, primary patency loss, and secondary AJKD Vol XX | Iss XX | Month 2018

patency loss in the first year were compared using Pearson χ 2 test. Costs shown in this study are amounts paid by Medicare to providers of service recorded in the USRDS. Costs include all Medicare payments (primary and secondary) for Parts A and B (institutional and physician supplier claims) associated with vascular access creations, interventions, maintenance, and ancillary costs, including related hospitalizations. Annualized per-patient per-year (PPPY) costs for vascular access management were calculated from and including the index AVF creation date with costs censored at death in each period. Annualized mean and standard deviation costs are presented as the primary analysis (median values as sensitivity analysis) and compared using Kruskal-Wallis test. In a secondary analysis, we calculated total aggregate Medicare payments for vascular access management among the ESRD population in 2011, 2012, and 2013. In this analysis, costs were calculated for all vascular access claims and by vascular access category. We used the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines12 to improve the reporting of our observational research study. Results This study identified 3 cohorts based on AVF status at time of first dialysis: cohort 1 initiated HD therapy with a mature AVF (n=2,704), cohort 2 initiated HD therapy with a CVC and a maturing AVF (n=3,530), and cohort 3 initiated HD therapy with a CVC only and an AVF was created within 9 months after dialysis therapy initiation (n=3,901; Table 1). Patients in cohort 1 had the AVF placed approximately 5 months (a median of 144 days) before dialysis therapy initiation. Patients in cohort 1 were less likely to be institutionalized, were more likely to ambulate, had less cardiac and pulmonary disease, and had higher albumin concentrations than the other cohorts (P < 0.001 for all comparisons). Cohort 1 also had significantly more women and a higher proportion of patients with ESRD caused by hypertension versus diabetes compared with the other cohorts (P < 0.001). Conversely, patients in cohort 3 were more likely to live in areas with the lowest median household income and were significantly less likely to have predialysis nephrology care (49%) compared with the other cohorts (P < 0.001 for both comparisons). Among patients in cohorts 2 and 3 initiating HD therapy using a CVC with or without a maturing AVF, nearly 46% of AVFs were not used for HD (Table 2). Of AVFs that were used for HD, 40% to 63% required 1 or more interventional or surgical procedure before the first AVF use (Table S3) and 70% to 86% required 1 or more interventional or surgical procedure within 1 year of surgical creation (primary patency loss in year 1; Table 2). Patients who initiated HD therapy with a mature AVF were more likely to maintain AVF primary patency in the first year (P < 0.001 for all comparisons). If the AVF was 3

Original Investigation Table 1. Sociodemographics, Comorbid Conditions, ESRD Information, and Laboratory Data for 3 Cohorts Based on AVF Status at Time of First Dialysis Cohort 1 Initiate w/ Mature AVF 2,704 No. of patients in study cohort No. of days of follow-up for 2.5-y study 782.2 ± 233.6a Deaths (calculated from index AVF surgery) Year 1 9.3%a Year 2 13.3% Year 3 (1st 6 mo) 6.2% Age at dialysis initiation 66-<75 y 44.0% 75-<85 y 44.8% ≥85 y 11.2% Sex Male 37.9% Female 62.1% Race White 77.0% Black 18.7% Other/unknown 4.3% Inability to ambulate/institutionalized 12.2% Cause of ESRD Hypertension 41.4% Diabetes 43.3% Glomerulonephritis 4.5% Other 10.8% Predialysis nephrology care None 8.0% 6-12 mo 43.0% >12 mo 49.0% Coexisting diseases Diabetes 52.3% Cancer 12.0% Congestive heart failure 33.7% Atherosclerotic heart disease 30.3% Other cardiac disease 20.8% Chronic obstructive pulmonary disease 10.5% Peripheral vascular disease 16.7% Cerebrovascular disease 11.2% Laboratory measuresb Body mass index, kg/m2 28.2 ± 6.7 Hemoglobin, g/dL 10.6 ± 18.8 Serum albumin, g/dL 3.6 ± 6.7 Median household income quartilec Quartile 1 (<$38,000/y) 22.0% Quartile 2 ($38,000-<$48,000/y) 24.6% Quartile 3 ($48,000-<$63,000/y) 25.6% Quartile 4 (≥$63,000/y) 25.6%

Cohort 2 Initiate w/ CVC and Maturing AVF 3,530 660.7 ± 311.7

Cohort 3 Initiate w/ CVC Only (AVF Not Yet Created) 3,901 651.1 ± 319.3

24.4% 15.5% 6.1%

25.9% 14.3% 7.1%

44.7% 43.8% 11.5%

44.0% 43.3% 12.7%

45.2% 54.8%

44.5% 55.5%

73.5% 21.0% 5.6% 23.4%

76.8% 18.4% 4.8% 23.1%

37.7% 48.1% 3.9% 10.3%

37.5% 40.6% 3.5% 18.4%

26.3% 38.0% 35.8%

49.2% 29.1% 21.7%

58.8% 10.9% 44.9% 34.1% 26.5% 13.5% 20.4% 13.5%

53.6% 10.6% 41.9% 28.1% 25.0% 14.3% 17.3% 11.2%

<0.001 0.2 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001

28.5 ± 7.2 9.9 ± 3.2 3.3 ± 4.6

28.3 ± 7.2 10.2 ± 15.6 3.1 ± 0.7

0.2 0.1 <0.001 0.01

23.3% 26.0% 24.7% 26.0%

25.1% 26.9% 24.4% 23.7%

P <0.001 <0.001

0.2

<0.001

0.003

<0.001 <0.001

<0.001

Note: Unless otherwise indicated, continuous data presented as mean ± standard deviation; categorical variables given as percentage. Abbreviations: AVF, arteriovenous fistula; CVC, central venous catheter; ESRD, end-stage renal disease. a By definition, cohort 1 initiates dialysis therapy with an AVF and has a lower death rate in year 1 because of left truncation; ie, to be included, these individuals must have AVF surgery before dialysis that has matured for use and they have survived to initiate dialysis therapy. b Based on availability of laboratory data. c Based on census data using 5-digit zip code for patient residence with 1% missing income.

4

AJKD Vol XX | Iss XX | Month 2018

Original Investigation Table 2. AVF Outcome Measures for 3 Study Cohorts by Demographics

AVF used for dialysis Age 66-<75 y 75-<85 y ≥85 y Sex Male Female Race (other not included) White Black Primary patency loss in y 1 Age 66-<75 y 75-<85 y ≥85 y Sex Male Female Race (other not included) White Black Secondary patency loss in y 1; AVF used and abandoned Age 66-<75 y 75-<85 y ≥85 y Sex Male Female Race (other not included) White Black AVF nonuse; AVF created but never used Age 66-<75 y 75-<85 y ≥85 y Sex Male Female Race (other not included) White Black

Cohort 1 Initiate w/ Mature AVF (n = 2,704) 92.2%a

Cohort 2 Initiate w/ CVC and Maturing AVF (n = 3,530) 54.3%

Cohort 3 Initiate w/ CVC Only (AVF Not Yet Created) (n = 3,901) 54.3%

1,102 (44.2%) 1,115 (44.7%) 276 (11.7%)

876 (45.7%) 836 (43.6%) 205 (10.7%)

960 (45.4%) 908 (42.9%) 249 (11.8%)

1,574 (63.1%) 919 (36.9%)

1,117 (58.3%) 800 (41.7%)

1,290 (60.9%) 827 (39.1%)

1,937 (77.7%) 448 (18.0%) 70.0%

1,416 (73.9%) 379 (19.8%) 85.7%

1,660 (78.4%) 356 (16.8%) 83.2%

752 (43.1%) 788 (45.2%) 204 (11.7%)

759 (46.20%) 708 (43.1%) 176 (10.7%)

814 (46.2%) 740 (42.0%) 208 (11.8%)

1,066 (61.1%) 678 (38.9%)

950 (57.8%) 693 (42.2%)

1,053 (59.8%) 709 (40.2%)

1,345 (77.1%) 331 (19.0%) 9.6%

1,201 (73.1%) 339 (20.6%) 10.6%

1,363 (77.4%) 320 (18.2%) 9.5%

114 (44.0%) 120 (46.3%) 25 (9.7%)

185 (49.6%) 153 (41.0%) 35 (9.4%)

169 (45.4%) 153 (41.1%) 50 (13.4%)

154 (59.5%) 105 (40.5%)

179 (48.0%) 194 (52.0%)

209 (56.2%) 163 (43.8%)

181 (69.9%) 63 (24.3%) 7.8%a

261 (70.0%) 86 (23.1%) 45.7%

284 (76.3%) 71 (19.1%) 45.7%

88 (41.7%) 96 (45.5%) 27 (12.8%)

702 (43.5%) 710 (44.0%) 201 (12.5%)

757 (42.4%) 779 (43.7%) 248 (13.9%)

104 (49.3%) 107 (50.7%)

816 (50.6%) 797 (49.4%)

875 (49.1%) 909 (51.0%)

145 (68.7%) 57 (27.0%)

1,177 (73.0%) 362 (22.4%)

1,336 (74.9%) 363 (20.4%)

P <0.001 0.6

0.005

0.002

<0.001 0.002

0.1

0.003

<0.001 0.2

0.01

0.2

<0.001 0.8

0.6

0.2

Note: See Table S2 for definitions of interventions required for primary and secondary patency loss using US Renal Data System data. Abbreviations: AVF, arteriovenous fistula; CVC, central venous catheter. a By definition, cohort 1 initiates dialysis therapy with an AVF that is mature and ready to be used (left truncation), explaining the high rate of AVF use and the low rate of secondary patency loss due to AVFs never used.

successfully used, subsequent abandonment of the AVF in the first year (secondary patency loss) was similar between groups (9.5% to 10.6%; pairwise comparisons P < 0.4). Total mean annualized PPPY vascular access costs in the AJKD Vol XX | Iss XX | Month 2018

first year after AVF creation were lowest for patients whose AVFs maintained primary patency in year 1 compared with patients whose AVFs experienced primary or secondary patency loss in year 1 (eg, $8,576 vs $16,428 and 5

Original Investigation $18,989, respectively, in cohort 3; P < 0.001 for both comparisons; Table 3). Patients whose AVFs were not used for HD experienced the highest cost in year 1 (eg, $32,405 in cohort 3; P < 0.001 for all comparisons). These relationships were seen across all 3 study cohorts. Total PPPY vascular access costs in the second year after AVF creation were lower compared to the first year and continue the cost trends observed in the first year. Patients whose AVFs maintained primary patency again experienced lower vascular access costs in year 2 compared with patients whose AVFs experienced primary or secondary patency loss in year 1 or patients who experienced AVF nonuse (Table 3). Patients with secondary patency loss or AVF nonuse had costs that were 3 to 4 times higher than patients who maintained primary patency in year 1 (eg, $5,479 vs $17,350 and $17,000, respectively, in cohort 3; P < 0.001 for both comparisons). Annualized 2.5-year PPPY vascular access costs after AVF creation continue the cost trends observed in the first and second years (Table 3). In other words, successful AVF use and maintenance of primary and secondary patency in year 1 resulted in lower costs over multiple years of follow-up. For example, in cohort 3, annualized costs over 2.5 years of follow-up were $7,871 for patients whose AVFs maintained primary patency in year 1 versus $17,350 for patients whose AVFs experienced secondary patency loss in year 1 versus $31,630 for patients whose AVF was not used (P < 0.001 for both comparisons). Patients with an AVF maturing at HD therapy initiation (cohort 2) had disproportionately higher costs in the first year after AVF creation compared with patients in the other cohorts. Further analysis revealed that patients in this cohort had a higher incidence of AVF creation during an inpatient hospitalization (w33% of patients in cohort 2 compared to w6% of patients in cohorts 1 and 3), with

average inpatient costs of $25,000 to $40,000 compared with outpatient facility costs of $2,500 to $3,500. For all cohorts, vascular access–related median PPPY costs following AVF creation in both the first year and cumulatively over 2.5 years were lowest among patients whose AVFs maintained primary patency in year 1 and highest among patients whose AVFs experienced secondary patency loss in year 1 or were not used (Table S4). Total aggregate Medicare payments for vascular access management among the ESRD population were calculated for 2011, 2012, and 2013 (Table 4). In each year, Medicare paid a total of w$2.8 billion for vascular access–related services. In 2013, there was 50.1% of the total cost for inpatient care related to vascular access procedures or complications (eg, catheter-related bacteremia), 39.6% for invasive imaging and endovascular procedures, 8.6% for open surgical procedures performed on outpatients, 1.1% for anesthesia used in vascular access procedures, and 0.6% for noninvasive diagnostic imaging procedures (Table 4; Item S1). Discussion There has been only 1 study evaluating the costs of vascular access care in the United States13 that focused on costs due to vascular access–associated hospitalizations from complications in CVCs, AVGs, and AVFs. Since this study was published 20 years ago, there has been a paradigm shift in vascular access use because AVFs are now the preferred form of vascular access and incentivized by CMS in Fistula First, Quality Incentive Program, and 5-Star programs.2 The CMS, the principal payer of dialysis services in the United States, has actively promoted use of AVFs with the twin goals of improving health outcomes and lowering costs. Specifically, CMS created the Fistula First Initiative in 2002,2 initially with a target of AVF use

Table 3. Vascular Access–Related Per-Patient Per-Year Costs From AVF Creation Based on AVF Outcomes in Year 1 Cohort 1: patients Year 1 Year 2 2.5 y (annualized) Cohort 2: patients Year 1 Year 2 2.5 y (annualized) Cohort 3: patients Year 1 Year 2 2.5 y (annualized)

Maintain Primary Patency Primary Patency Loss initiating HD with a mature AVF $6,442 ± $8,882 $15,009 ± $16,896 $4,279 ± $11,378 $7,403 ± $14,178 $5,560 ± $8,368 $11,761 ± $15,871 initiating HD with a catheter with a maturing AVFb $21,514 ± $35,849 $28,154 ± $25,872 $3,779 ± $10,039 $9,395 ± $23,526 $17,234 ± $36,001 $20,506 ± $23,583 initiating HD with a catheter $8,576 ± $13,994 $16,428 ± $17,536 $5,479 ± $23,764 $8,423 ± $20,446 $7,871 ± $14,183 $13,282 ± $16,942

Secondary Patency Lossa

AVF Nonuse

$18,104 ± $16,653 $13,106 ± $20,304 $16,428 ± $14,865

$30,687 ± $60,013 $13,686 ± $17,432 $29,710 ± $59,959

$28,980 ± $28,137 $15,357 ± $32,506 $24,116 ± $25,318

$55,475 ± $86,397 $15,253 ± $31,403 $51,745 ± $86,782

$18,989 ± $16,721 $17,350 ± $28,477 $17,808 ± $15,764

$32,405 ± $104,025 $17,000 ± $33,334 $31,630 ± $103,941

Note: Values given as mean ± standard deviation. See Item S1 for complete list of all procedures, diagnoses, and codes used to calculate vascular access–related costs. See Table S2 for list of procedures, diagnoses, and codes to identify primary and secondary patency loss. In year 2 of follow-up, a large portion of patients did not have vascular access–related claims (although they were included in the denominator). Annualized per-patient per-year costs for vascular access management were calculated with costs censored at death in each period shown for each cohort by primary and secondary patency loss and AVF nonuse. Abbreviations: AVF, arteriovenous fistula; HD, hemodialysis. a After AVF use. b Cohort 2 comprises a disproportionate number of “crash and burn” patients who underwent AVF creation in the inpatient setting, with average payments of w$25,000 to $40,000 versus w$2,500 to $3,500 (facility costs) for outpatient AVF creation, which is more common among cohorts 1 and 3.

6

AJKD Vol XX | Iss XX | Month 2018

Original Investigation Table 4. Medicare Payments for Vascular Access Management in the ESRD Population Selected Vascular Access Payments Noninvasive diagnostic imaging procedures Open surgical procedures Invasive imaging and endovascular procedures Inpatient admissions Anesthesia associated with VA procedures Total

2011 $15.0

2012 $11.8

2013 $16.8

$260.8 $1,038.2

$223.3 $1,157.2

$246.0 $1,127.9

$1,519.2 $35.0

$1,403.2 $29.7

$1,425.9 $30.1

$2,868.2

$2,825.2

$2,846.7

Note: Amounts given in millions. See Item S1 for a complete list of all vascular access–related procedures, diagnoses, and codes for each category of payments. Costs include all Medicare payments (primary and secondary) for Parts A and B (institutional and physician supplier claims) for both incident and prevalent dialysis patients in each calendar year associated with vascular access creations, interventions, and ancillary costs, including related hospitalizations. Dialysis patients of all ages are included. Costs do not include Medicare payments for vascular access management for predialysis patients or dialysis patients enrolled in health maintenance organizations. Abbreviations: ESRD, end-stage renal disease; VA, vascular access.

in 40% of prevalent HD patients; subsequently, the goal was increased to 66%. In our study using contemporary national data from Medicare and CMS claims to examine the rate and timing of vascular access events and the Medicare costs from the time of AVF surgical creation based on various scenarios of AVF function in the first year after creation, we found that among patients initiating HD therapy using a CVC with or without a maturing AVF, nearly half (46%) never used their AVFs for dialysis within 6 months of AVF creation, resulting in continuing use of a CVC and the potential for a second vascular access creation surgery. Similar to our findings, other studies indicate that w50% of AVFs never become usable for HD,14-20 although the Hemodialysis Fistula Maturation (HFM) Study found that a lower proportion, approximately a quarter of all AVFs, were not successfully used for dialysis despite undergoing procedures to assist maturation.21 Differences in study design, patient demographics, and end point definitions make comparison with these studies difficult; for example, the HFM Study was conducted at 7 academic centers and enrolled a patient population that was younger and included a lower percentage of females and predialysis patients, but a higher proportion of African Americans.21 Moreover, in our study, among those with an AVF that was used for HD, many (40%-63%) required interventions to facilitate use and a large majority (71%86%) required an intervention to maintain AVF patency in the first year following AVF creation. PPPY vascular access costs in the 2.5 years after surgical creation were twice as high for patients with an AVF that required an intervention in the first year after creation, more than twice as high for patients with an AVF that was abandoned in the first year after creation, and 4 times as high for patients with an AVF that was not used compared with patients with an AVF that did not require an intervention in the first year. Overall in 2013, AJKD Vol XX | Iss XX | Month 2018

Medicare paid a total of $2.8 billion for vascular access–related services. AVFs that require an intervention to promote AVF use require more interventions to maintain patency,5,6 have shorter secondary patency, and as shown in this study, have substantially greater costs. For patients initiating HD therapy with a CVC, the duration of CVC dependence is increased if the AVF fails to become usable for HD or requires a salvage procedure to promote use,5 and is further prolonged if the AVF is abandoned. Previously published literature has demonstrated that dialysis with a CVC is associated with numerous adverse outcomes, including substantially elevated rates of infection, including bacteremia, osteomyelitis, and endocarditis.22-27 These complications are 5 to 8 times higher than with an AVF and result in increased costs from hospitalizations.22,23,26 Moreover, many AVFs develop complications that cannot be amenable to salvage procedures, at which point the AVF is abandoned and a new access is required. Our study demonstrates that AVF nonuse, primary patency loss, and secondary patency loss substantially increase the overall costs of vascular access care. Per-patient vascular access costs in the first year after surgical creation were twice as high for patients with AVF primary patency loss compared with patients who maintained primary patency. Costs in the first year were 4 times as high for patients whose AVFs were not used, which reflects costs associated with interventions to promote AVF maturation, new access surgeries, and complications associated with prolonged exposure to CVCs. Vascular access costs were significantly higher in the second year for patients who experienced AVF primary or secondary patency loss in year 1 or who never used their AVFs for HD compared with patients who maintained primary patency in year 1. Although the costs of vascular access management in AVF patients are highest in the first year and lower in the second year, during 2.5 years of follow-up, the significant differences in costs based on AVF outcomes continued. This suggests that AVF failure in the first year after surgical creation can result in higher costs for multiple years. Our study presents novel data on total vascular access costs within the ESRD program. Total 2013 Medicare feefor-service expenditures for vascular access in the ESRD population were $2.85 billion, constituting w12% of all ESRD Medicare expenditures for HD patients. This figure excludes expenditures for HD patients who were covered by other payers (eg, Medicare Advantage and commercial). This figure also excludes patient obligations such as copays and deductibles, which accounted for 13% of total costs in the Medicare ESRD population in 2012.28 Thus, vascular access management in patients with ESRD costs an estimated $5 billion annually in total direct expenditures by public and private payers and patients, a figure that excludes vascular access expenditures in all pre-ESRD patients. Our study has several major strengths. First, it evaluates a national cohort of all Medicare-insured dialysis patients 7

Original Investigation available from the USRDS; thus, results are widely representative of the elderly US HD population. A second strength is the ability to determine whether an AVF has become usable for dialysis through evaluation of vascular access modifier codes (V codes) reported monthly by every dialysis unit. Our study also has several limitations. First, aggregate vascular access costs were calculated for each calendar year for the ESRD population, but (1) could not be calculated for predialysis procedures/costs because these data were incomplete according to USRDS officials, and (2) use of the USRDS database provided information only for vascular access procedures/costs for patients who initiate dialysis therapy. Data for procedures and costs for patients who did not initiate dialysis therapy were not available to assess the potentially substantial costs that may have been incurred when fistulas were created but never used before death, because chronic kidney disease did not progress to ESRD, or because of the choice of conservative management in lieu of initiating dialysis therapy. Second, given our database constraints, we do not have information about why such a large proportion of fistulas (nearly half) were never used. Identifying the causes of the high levels of fistula nonuse (eg, failed maturation and patient refusal) would be important to mitigate the large economic impact found in this analysis. Third, the analysis does not include data for Medicare health maintenance organization patients or those with non-Medicare claims such as private insurance, although the large majority of elderly patients with ESRD use the Medicare fee-for-service program. Fourth, our study does not directly compare costs between AVFs and AVGs. We believe this is an important question for additional analysis because studies comparing clinical outcomes and costs associated with AVGs would help in determining whether different patient selection for permanent access could affect outcomes or costs. Last, costs were only captured from the perspective of the third-party payer; societal costs are anticipated to be much higher (lost time from work, family, etc), and costs from the patient perspective would also be considerable. This study suggests that AVFs that experience dysfunction in the first year after creation, especially those that are never used for HD, result in substantially higher long-term costs. There remains an unmet clinical need for improving outcomes and reducing avoidable costs after AVF surgical creation. Supplementary material Item S1: Codes associated with vascular access management among ESRD patients. Table S1: Definition and sample size based on AVF outcomes in year 1 for 3 cohorts by AVF status at the time of first dialysis. Table S2: List of interventions used to identify, from administrative data, (a) primary patency loss in first year and (b) secondary patency loss. Table S3: Use and primary patency outcome measures for the 3 study cohorts.

8

Table S4: Vascular access-related PPPY median costs after AVF creation based on AVF outcomes in year 1 for 4 cohorts according to AVF status at the time of first dialysis.

Article Information Authors’ Full Names and Academic Degrees: Mae Thamer, PhD, Timmy C. Lee, MD, Haimanot Wasse, MD, MPH, Marc H. Glickman, MD, Joyce Qian, MA, Daniel Gottlieb, MBA, Scott Toner, MBA, and Timothy A. Pflederer, MD. Authors’ Affiliations: Medical Technology & Practice Patterns Institute, Bethesda, MD (MT, JQ); Department of Nephrology, University of Alabama at Birmingham, Birmingham, AL (TCL); Division of Nephrology, Rush University Medical Center, Chicago, IL (HW); Hancock Jaffe Laboratories, Irvine, CA (MHG); Proteon Therapeutics, Inc, Waltham, MA (DG, ST); and Illinois Kidney Disease and Hypertension Center, Peoria, IL (TAP). Address for Correspondence: Timothy A. Pflederer, MD, 200 E Pennsylvania Ave, Ste 212, Peoria, IL 61603. E-mail: tap@ renalcareassoc.com Authors’ Contributions: Research idea and study design: MT, TAP, ST, DG, TCL, HW, MHG; data acquisition: MT; data analysis/ interpretation: MT, TAP, ST, DG, TCL, HW, MHG, JQ; statistical analysis: MT, JQ; supervision or mentorship: TAP. Each author contributed important intellectual content during manuscript drafting or revision and accepts accountability for the overall work by ensuring that questions pertaining to the accuracy or integrity of any portion of the work are appropriately investigated and resolved. Support: This study was funded by Proteon Therapeutics, Inc, which is developing an investigational therapy for HD vascular access. Proteon staff, listed in the next section, served as co-authors and contributed to the research idea and study design, data analysis/ interpretation, writing the manuscript, and the decision to submit the manuscript for publication. Financial Disclosure: Dr Lee is supported by an American Society of Nephrology (ASN) Carl W. Gottschalk Scholar Grant, University of Alabama at Birmingham Nephrology Research Center Anderson Innovation Award, University of Alabama at Birmingham Center for Clinical and Translational Science Multidisciplinary Pilot Award (1UL1TR001417-01), grant 1R43DK109789-01 from National Institutes of Diabetes, Digestive and Kidney Diseases, and grant 1I01BX003387-01A1 from a Veterans Affairs Merit Award. Dr Lee is a consultant for Proteon Therapeutics and Merck, serves as a counselor for the American Society of Diagnostic and Interventional Nephrology (ASDIN), and serves as a member of the National Kidney Foundation Kidney Disease Outcomes Quality Initiative (NKF-KDOQI) vascular access guideline work group. Dr Wasse is a consultant for Proteon Therapeutics and CTI-Clinical Trial Services, serves as President-Elect for the ASDIN, is on the Executive Council of the ASDIN, and serves as the Co-Chair of the Kidney Health Initiative (KHI) Vascular Access Clinical Outcomes Workgroup. Dr Glickman is currently Chief Medical Officer for Hancock Jaffe labs; he is retired from active vascular surgical practice and serves as consultant and member of the NKF-KDOQI guideline work group and as consultant and member to the ASN in the KHI. Dr Glickman is a consultant to Merit Medical and TVA Medical and Proteon Therapeutics. Drs Thamer and Qian are employees of Medical Technology and Practice Patterns Institute. Mr Gottlieb is an employee of and has stock ownership and options in Proteon Therapeutics, Inc, which funded this study. Mr Toner is an employee of and has stock ownership and options in Proteon Therapeutics, Inc, which funded this study. Dr Pflederer is a consultant for Proteon Therapeutics, serves on the Medical Advisory Board of Fresenius Kidney Care, is an advisor to the American Medical Association Current Procedural Trends Committee, and serves on the Renal Physicians Association board of directors.

AJKD Vol XX | Iss XX | Month 2018

Original Investigation Peer Review: Received July 10, 2017. Evaluated by 2 external peer reviewers, with direct editorial input from a Statistics/Methods Editor, an Associate Editor, and the Editor-in-Chief. Accepted in revised form January 8, 2018.

References 1. Ravani P, Palmer SC, Oliver MJ, et al. Associations between hemodialysis access type and clinical outcomes: a systematic review. J Am Soc Nephrol. 2013;24(3):465-473. 2. Lynch JR, Wasse H, Armistead NC, McClellan WM. Achieving the goal of the Fistula First Breakthrough Initiative for prevalent maintenance hemodialysis patients. Am J Kidney Dis. 2011;57(1):78-89. 3. Pisoni RL, Young EW, Dykstra DM, et al. Vascular access use in Europe and in the United States: results from the DOPPS. Kidney Int. 2002;61(1):305-316. 4. CROWNWeb. The ESRD National Patient Registry and Quality Measure Reporting System. http://mycrownweb.org/. Accessed September 28, 2017. 5. Harms JC, Rangarajan S, Young CJ, Barker-Finkel J, Allon M. Outcomes of arteriovenous fistulas and grafts with and without intervention prior to successful use. J Vasc Surg. 2016;64(1): 155-162. 6. Lee T, Ullah A, Allon M, et al. Decreased cumulative access survival in arteriovenous fistulas requiring interventions to promote maturation. Clin J Am Soc Nephrol. 2011;6(3):575-581. 7. Saran R, Robinson B, Abbott KC, et al. US Renal Data System 2016 annual data report: epidemiology of kidney disease in the United States. Am J Kidney Dis. 2017;69(3)(suppl 1): S1-S688. 8. Allon M, Robbin ML. Increasing arteriovenous fistulas in hemodialysis patients: problems and solutions. Kidney Int. 2002;62(4):1109-1124. 9. Chiulli LC, Vasilas P, Dardik A. Superior patency of upper arm arteriovenous fistulae in high risk patients. J Surg Res. 2011;170(1):157-164. 10. Pflederer TA, Kwok S, Ketel BL, Pilgram T. A comparison of transposed brachiobasilic fistulae with nontransposed fistulae and grafts in the Fistula First era. Semin Dial. 2008;21(4): 357-363. 11. Solid CA, Collins AJ, Ebben JP, et al. Agreement of reported vascular access on the medical evidence report and on Medicare claims at hemodialysis initiation. BMC Nephrol. 2014;15(1):30. 12. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP. STROBE Initiative. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement: guidelines for reporting observational studies. Int J Surg. 2014;12(12):1495-1499.

AJKD Vol XX | Iss XX | Month 2018

13. Feldman HI, Kobrin S, Wasserstein A. Hemodialysis vascular access morbidity. J Am Soc Nephrol. 1996;7(4):523-535. 14. Dember LM, Beck GJ, Allon M, et al. Effect of clopidogrel on early failure of arteriovenous fistulas for hemodialysis: a randomized controlled trial. JAMA. 2008;299(18):2164-2171. 15. Huijbregts HJ, Bots ML, Wittens CH, et al. Hemodialysis arteriovenous fistula patency revisited: results of a prospective, multicenter initiative. Clin J Am Soc Nephrol. 2008;3(3): 714-719. 16. Singh P, Robbin ML, Lockhart ME, Allon M. Clinically immature arteriovenous hemodialysis fistulas: effect of US on salvage. Radiology. 2008;246(1):299-305. 17. Asif A, Cherla G, Merrill D, Cipleu CD, Briones P, Pennell P. Conversion of tunneled hemodialysis catheter-consigned patients to arteriovenous fistula. Kidney Int. 2005;67(6): 2399-2406. 18. Asif A, Roy-Chaudhury P, Beathard GA. Early arteriovenous fistula failure: a logical proposal for when and how to intervene. Clin J Am Soc Nephrol. 2006;1(2):332-339. 19. Beathard GA, Arnold P, Jackson J, Litchfield T; Physician Operators Forum of RMS Lifeline. Aggressive treatment of early fistula failure. Kidney Int. 2003;64(4):1487-1494. 20. Turmel-Rodrigues L, Mouton A, Birmele B, et al. Salvage of immature forearm fistulas for haemodialysis by interventional radiology. Nephrol Dial Transplant. 2001;16(12):2365-2371. 21. Cheung AK, Imrey PB, Alpers CE, et al; Hemodialysis Fistula Maturation Study Group. Intimal hyperplasia, stenosis, and arteriovenous fistula maturation failure in the Hemodialysis Fistula Maturation Study. J Am Soc Nephrol. 2017;28(10): 3005-3013. 22. Lee T, Barker J, Allon M. Tunneled catheters in hemodialysis patients: reasons and subsequent outcomes. Am J Kidney Dis. 2005;46(3):501-508. 23. Oliver MJ, Rothwell DM, Fung K, Hux JE, Lok CE. Late creation of vascular access for hemodialysis and increased risk of sepsis. J Am Soc Nephrol. 2004;15(7):1936-1942. 24. Dhingra RK, Young EW, Hulbert-Shearon TE, Leavey SF, Port FK. Type of vascular access and mortality in U.S. hemodialysis patients. Kidney Int. 2001;60(4):1443-1451. 25. Pastan S, Soucie JM, McClellan WM. Vascular access and increased risk of death among hemodialysis patients. Kidney Int. 2002;62(2):620-626. 26. Nassar G, Ayus J. Infectious complications of the hemodialysis access. Kidney Int. 2001;60(1):1-13. 27. Ajay R, William M, Debi K, Cariad E. Catheter access for hemodialysis defines higher mortality in late-presenting dialysis patients. Ren Fail. 2010;32(10):1183-1188. 28. Saran R, Li Y, Robinson B, et al. US Renal Data System 2014 annual data report: epidemiology of kidney disease in the United States. Am J Kidney Dis. 2015;66(1)(suppl 1):S1-S306.

9