Acceso Vascular Metanalisis

Original Research
1 Hemodialysis Access Type and Access Patency Loss: An 57

2 58
3 Q1 Observational Cohort Study 59
4 60
Q6 Nicholas S. Roetker, Haifeng Guo, Dena Rosen Ramey, Ciaran J. McMullan, G. Brandon Atkins,
5 61
and James B. Wetmore
6 62
7 63
8 Rationale & Objective: Access patency outcomes for HD. Over 3 years of follow-up, AVG users, Complete author and article 64
9 for arteriovenous fistulas (AVFs) as compared with compared to AVF users, had a higher cumulative information provided before 65
references.
10 arteriovenous grafts (AVGs) in patients receiving incidence of loss of primary unassisted patency 66
11 hemodialysis (HD) who have achieved a func- (87% vs 69%; HR, 1.56; 95% CI, 1.52-1.60), Correspondence to
67
tioning permanent access are not fully explored. loss of primary assisted patency (69% vs 25%; N.S. Roetker (nick.roetker@
12 HR, 3.79; 95% CI, 3.67-3.92), and loss of cdrg.org) 68
13 Study Design: Observational cohort study.
secondary patency (22% vs 10%; HR, 2.03; 95% Kidney Med. 69
14 Setting & Population: Fee-for-service Medicare CI, 1.92-2.16). Stratified analyses revealed XX(XX):100567. Published 70
15 Q2 beneficiaries aged ≥18 years with kidney failure differences by subgroups; in particular, incidence online month xx, xxxx. 71
16 who were newly using a permanent access for of patency loss was higher among patients who doi: 10.1016/ 72
j.xkme.2022.100567
17 maintenance HD from the United States Renal underwent prior interventions to maintain 73
18 Data System (2010-2015). Patients using an oral prefunctional access patency and Black patients. © 2022 by the National 74
anticoagulant were excluded. Kidney Foundation Inc. and
19 Limitations: This analysis focused on outcomes [COPYRIGHT HOLDER] 75
20 Exposure: AVG or AVF. occurring after first successful use of a permanent Published by Elsevier Inc. 76
21 Outcomes: Loss of primary unassisted, primary
access and thus does not inform about risk of All rights reserved. This is 77
patency loss during access maturation. an open access article
22 assisted, and secondary patency. under the CC BY-NC-ND 78
23 Analytical Approach: Outcomes were character-
Conclusions: Among patients with kidney failure license (http:// 79
24 ized using cumulative incidence curves, and HRs
who successfully used a permanent access for creativecommons.org/ 80
HD, patency loss was consistently substantially licenses/by-nc-nd/4.0/).
25 adjusted for sociodemographic and clinical factors 81
higher in those using AVGs compared with AVFs.
26 were estimated for the comparison of AVF versus
New interventions, such as prophylactic drugs, are
82
27 AVG.
needed to improve access longevity and reduce 83
28 Results: The cohort included 60,329 and 17,763 the need for invasive interventions, particularly 84
29 patients newly using an AVF and AVG, respectively, among patients unable to receive a fistula. 85
30 86
31 Maintaining long-term AV access patency frequently 87
32
33
A well-functioning vascular access is essential for the
provision of maintenance hemodialysis (HD) in pa-
tients with kidney failure. An ideal access provides reliable,
requires interventions to overcome thrombosis and other
complications (eg, anatomical problems, stenosis). These
88
89
34 complication-free access to the vascular system while complications are often categorized into 3 types of access 90
35 suiting the clinical circumstances of a patient.1 Due to failure3 by increasing order of severity. First, loss of unas- 91
36 considerably lower rates of complications—including sisted primary patency constitutes complications that are 92
37 infection, stenosis, and thrombosis—a permanent arte- often less severe than thrombosis and require relatively 93
38 riovenous (AV) access such as an arteriovenous fistula simple interventions (eg, angioplasty). Second, loss of 94
39 (AVF) or arteriovenous graft (AVG) is much preferred over assisted primary patency generally involves access thrombosis 95
40 a tunneled central venous catheter for long-term use. events that can be salvaged via prompt thrombectomy. 96
41 Although the AVF has long been considered the optimal Third, loss of secondary patency represents any complication 97
42 access because of its superior lifespan and lower compli- requiring abandonment of the access (eg, irreversible 98
43 cation rate, the AVG can be a suitable alternative with the thrombosis). 99
44 benefits of shorter maturation times, higher primary suc- We sought to characterize the risk of access failure 100
45 cess rates, and accessibility for some who lack adequate outcomes in AV access users by type and patient charac- 101
46 vasculature for a “native” access.2 teristics to help nephrologists and others predict the ex- 102
47 Thrombosis is a major problem among patients with an pected longevity of a permanent access from the time of its 103
48 AV access, particularly grafts, which are inherently first use for HD. Using a nationwide end-stage kidney 104
49 thrombogenic. Nevertheless, a thrombosed AVG is often disease (ESKD) registry, we constructed a cohort of pa- 105
50 easier to declot than a thrombosed AVF.2 If an AV access is tients receiving maintenance HD with a newly functioning 106
51 not salvaged within approximately 48 hours of throm- AV access and compared the rates of primary unassisted, 107
52 bosing, a catheter is often needed to restore temporary primary assisted, and secondary patency loss between 108
53 access. Thus, thrombosis initiates a “countdown” in which those using an AVF and AVG, both overall and in specific 109
54 the health care system must marshal resources and work to patient subgroups. Whereas other work in this area has 110
55 quickly reopen the access. often focused on patients who initiated HD using a 111
56 112
Kidney Med Vol XX | Iss XX | Month 2022 | 100567 1

FLA 5.6.0 DTD XKME100567_proof 23 November 2022 2:33 am ce
Roetker et al
Patients aged <18 years or who received an oral antico-

113 PLAIN-LANGUAGE SUMMARY agulant within 90 days before the index date were 169
114 In patients with kidney failure receiving hemodialysis, excluded. The latter patients were excluded to allow study 170
115 vascular access complications such as stenosis and of the background risks of patency failure independent of 171
116 thrombosis are common and often require restorative the therapeutic effects of anticoagulation. 172
117 interventions. In the event that a malfunctioning access 173
118 cannot be remediated, it may be necessary to create a Modality and Vascular Access 174
119 brand new access. We examined the risks of needing HD modality was identified using the treatment history 175
120 different types of treatment for access complications in a files. The date of first use of an AV access for HD (ie, the 176
121 cohort of Medicare beneficiaries receiving hemodialysis index date) was identified using the CMS 2728 form or 177
122 who had successfully used an arteriovenous access. We outpatient dialysis claims with a Health Care Common 178
123 Procedure Coding System modifier code of V6 (AVG) or 179
found that the risks of needing an intervention to treat
124 V7 (AVF) but no V5 (catheter) modifier. The vascular 180
125 thrombosis and needing a brand new access placement 181
access modifier codes are reported on at least 1 dialysis
126 were higher among graft users compared with fistula 182
claim per month. Use of an AV access for HD throughout
127 users. New interventions are needed to improve access follow-up was then defined using outpatient dialysis 183
128 outcomes in the hemodialysis population, especially claims occurring after the index date. Claims with Health 184
129 among those not using a fistula. Care Common Procedure Coding System modifier V5 185
130 alone, or with modifier V5 in combination with V6 or V7, 186
131 were considered HD sessions using a catheter. Patients 187
132 catheter and later transitioned to an AV access or condi- were followed as long as they continued to use an AV 188
133 tioned on the presence of pre-ESKD Medicare claims, this access, allowing for periods of temporary catheter use of 189
134 analysis includes all patients receiving HD with a newly no more than 30 consecutive days (ie, patients were 190
135 functioning AV access, whether it was created preemp- censored on reaching 31 consecutive days of catheter use). 191
136 tively or after reaching kidney failure and whether or not Throughout follow-up, patients were categorized as AVF 192
137 Medicare claims were available before the onset of kidney or AVG users according to the access used on the index 193
138 failure.4-6 date. 194
139 195
140 METHODS 196
Outcomes
141 197
142 Data sources The outcomes of interest (illustrated in Fig S1) included 198
143 This study used the following analysis files from the 2010 loss of primary unassisted patency, loss of primary assisted 199
144 to 2015 United States Renal Data System (USRDS) ESKD patency, and loss of secondary patency. Loss of primary 200
145 registry database:7 the End-Stage Renal Disease Medical unassisted patency constituted the first intervention during 201
146 Q3 Evidence Report (Form CMS 2728) and Death Notification follow-up to maintain full function of the access, including 202
147 (CMS 2746) forms, patients, treatment history, and angioplasty, surgery for anatomical or physical complica- 203
148 Medicare Part A (institutional), Part B (physician/suppler), tions, thrombectomy or thrombolysis, new AV access 204
149 and Part D (prescription drug events) claims files. A waiver creation, or access abandonment. Loss of primary assisted 205
150 of informed consent was granted by Hennepin Health- patency was the first intervention to treat access throm- 206
151 care’s institutional review board, and data use agreements bosis, create a new AV access, or abandon the access. 207
152 between the Hennepin Healthcare Research Institute and Finally, loss of secondary patency was defined (only) as 208
153 the National Institute of Diabetes and Digestive and Kidney creation of a new AV access or abandonment of the access. 209
154 Diseases were in place. Subject to the terms of a data use Since follow-up started from the date of first use of the AV 210
155 agreement with the USRDS, qualified individuals can freely access for HD, these outcomes could be considered as 211
156 obtain data used in this analysis from USRDS. representing loss of functional access patency. The algo- 212
157 rithms used to define each patency outcome are presented 213
158 Study design in Table S1. Relevant procedure codes were selected based 214
159 This study used a retrospective cohort design. Patients with on clinical expertise and recent USRDS publications.5,8 215
160 kidney failure receiving HD who were newly using an AV 216
161 vascular access were considered for inclusion. The date of Patient Characteristics 217
162 first use of the AV access for outpatient HD, which could Patient characteristics included age, sex, race, enrollment 218
163 occur on or after the date of kidney failure incidence, was in Medicaid, body mass index (BMI), primary cause of 219
164 defined as the index date. Only patients with continuous kidney failure, catheter use at HD initiation, time since HD 220
165 Medicare fee-for-service and Part D coverage from the date initiation, comorbid conditions, interventions to maintain 221
166 of kidney failure incidence to the index date were prefunctional access patency, and prescription medica- 222
167 included, which ensured that we could observe the full tions. Comorbid conditions were defined using the CMS 223
168 history of modality, vascular access, and medication use. 2728 form in combination with Medicare claims in the 224
2 Kidney Med Vol XX | Iss XX | Month 2022 | 100567

Roetker et al
Table 1. Baseline Characteristics in Patients Receiving Hemodialysis by Access Type

225 Characteristic AVF (N = 60,329) AVG (N = 17,763) 281
226 Age (y), n (%)
282
227 18-44 2,838 (4.7%) 771 (4.3%) 283
228 45-64 15,987 (26.5%) 4,310 (24.3%) 284
229 65-74 22,032 (36.5%) 6,026 (33.9%) 285
230 75-84 15,521 (25.7%) 4,981 (28.0%) 286
231 ≥85 3,951 (6.5%) 1,675 (9.4%) 287
232 Female, n (%) 26,734 (44.3%) 10,805 (60.8%) 288
233 Race, n (%) 289
234 White 42,132 (69.8%) 10,008 (56.3%) 290
235 Black 14,746 (24.4%) 6,751 (38.0%) 291
236 Other 3,451 (5.7%) 1,004 (5.7%) 292
237 Medicare/Medicaid dual eligible, n (%) 28,365 (47.0%) 9,600 (54.0%) 293
238 Body mass index in kg/m2, n (%)a 294
239 <18.5 1,796 (3.0%) 847 (4.8%) 295
240 18.5-24.9 18,115 (30.2%) 6,068 (34.4%) 296
241 25-29.9 17,476 (29.2%) 4,750 (27.0%) 297
242 ≥30 22,501 (37.6%) 5,960 (33.8%) 298
243 Primary cause of end-stage kidney disease, n (%) 299
244 Diabetes 31,305 (51.9%) 9,003 (50.7%) 300
245 Hypertension 18,688 (31.0%) 5,772 (32.5%) 301
246 Glomerulonephritis 3,143 (5.2%) 817 (4.6%) 302
247 Cystic kidney disease 894 (1.5%) 182 (1.0%) 303
248 Other 6,299 (10.4%) 1,989 (11.2%) 304
249 Catheter at HD initiation, n (%) 37,632 (62.4%) 12,710 (71.6%) 305
≥6 mo since HD initiation, n (%) 17,703 (29.3%) 4,500 (25.3%)
250 306
Comorbid conditions, n (%)
251 307
Diabetes 45,820 (76.0%) 13,775 (77.5%)
252 308
Hypertension 59,725 (99.0%) 17,634 (99.3%)
253 309
Congestive heart failure 36,007 (59.7%) 11,559 (65.1%)
254 310
Atherosclerotic heart disease 33,464 (55.5%) 10,286 (57.9%)
255 Other cardiac disease 27,655 (45.8%) 8,857 (49.9%)
311
256 Cerebrovascular disease 14,582 (24.2%) 5,539 (31.2%)
312
257 Peripheral vascular disease 26,552 (44.0%) 8,672 (48.8%) 313
258 Chronic obstructive pulmonary disease 20,713 (34.3%) 6,909 (38.9%) 314
259 Atrial fibrillation 9,750 (16.2%) 3,406 (19.2%) 315
260 Liver disease 4,393 (7.3%) 1,452 (8.2%) 316
261 Cancer 8,664 (14.4%) 2,660 (15.0%) 317
262 Interventions to maintain prefunctional access patency, n (%) 318
263 Angioplasty 15,024 (24.9%) 3,350 (18.9%) 319
264 Surgery for anatomical complication 10,748 (17.8%) 2,251 (12.7%) 320
265 Thrombectomy or thrombolysis 2,131 (3.5%) 2,122 (11.9%) 321
266 New access creation (prior failure) 18,991 (31.5%) 7,646 (43.0%) 322
267 Prescription medications, n (%) 323
268 Antiplatelet 9,643 (16.0%) 2,956 (16.6%) 324
269 Statin 28,226 (46.8%) 7,974 (44.9%) 325
270 Antihypertensive 31,035 (51.4%) 9,107 (51.3%) 326
271 Antiarrhythmic 2,104 (3.5%) 696 (3.9%) 327
272 Antidiabetic 23,358 (38.7%) 6,801 (38.3%) 328
273 Abbreviations: AVF, arteriovenous fistula; AVG, arteriovenous graft; HD, hemodialysis. 329
a
Body mass index values missing in n = 579 patients. Percentages are reported with respect to the available sample size for this covariate.
274 330
275 331
276 332
277 year before the index date using the definitions in index date based on the definitions in Table S1. Use of key 333
278 Table S2. Interventions to maintain access patency were prescription medications on the index date were identified 334
279 identified using procedure codes in the year before the using Part D claims. 335
280 336

Roetker et al
Statistical Analysis newly using their AV access for HD in kidney failure, and
337 Descriptive statistics were used to summarize baseline who were not receiving oral anticoagulation, were 393
338 characteristics by vascular access type (AVF and AVG). included in the study. Overall, the mean age was 394
339 Starting from the index date, patients were followed until 68.9 ± 12.3 years, 48.1% were women, and 66.8% were 395
340 the earliest of: date of the outcome of interest; death; kidney White. Patients who were older, women, Black, Medicaid- 396
341 transplant; switch to a catheter access or to peritoneal dial- eligible, normal weight or underweight, catheter users at 397
342 ysis (if longer than 30 days); first oral anticoagulant pre- HD initiation, or who had comorbid conditions were more 398
343 scription; loss of Medicare coverage; August 31, 2015; or 3 likely, on average, to use an AVG (Table 1). Prior in- 399
344 years of follow-up. Cumulative incidence curves were esti- terventions to maintain prefunctional access patency, 400
345 mated for each outcome, separately by access type. The including thrombectomy or thrombolysis and new AV 401
346 cumulative incidence function was used to account for access creation (ie, prior access failure), were more com- 402
347 death as a competing risk.9 Hazard ratios (HRs) comparing mon among AVG users, whereas angioplasty and surgical 403
348 AVG with AVF for each outcome of interest at 3 months, 1 repairs were more common among AVF users (Table 1). 404
349 year, and 3 years of follow-up were estimated using Cox The numbers of previous failed AVFs and AVGs are shown 405
350 proportional hazards models adjusted for the baseline in Table S3. 406
351 covariates in Table 1. Missing values for BMI category were 407
352 multiply imputed with 10 datasets using multinomial lo- Cumulative Incidence of Loss of Access Patency 408
353 gistic regression; results were pooled according to Rubin’s Estimates of cumulative incidence for AVF and AVG users 409
354 rules.10 A sensitivity analysis was also conducted among the for each loss of access patency outcome over the 3 years of 410
355 subgroup of patients with a prior failed AVF. follow-up are presented in Figure 1. Loss of patency was 411
356 To explore whether the associations differed across sub- consistently more common among AVG users. By 3 years, 412
357 groups, cumulative incidence curves and HRs were estimated the probability of needing an intervention for any reason 413
358 with stratification by sex, age, race, BMI, time since HD to maintain full function of the access (ie, loss of primary 414
359 initiation, prior thrombectomy or thrombolysis, prior access unassisted patency) was 87% in AVG users and 69% in AVF 415
360 failure, atrial fibrillation, and antiplatelet medication use. users (Fig 1A). Loss of primary assisted patency, typically 416
361 Interactions were evaluated on the multiplicative scale. constituting the need for an intervention to treat access 417
362 thrombosis, was more than 3-fold higher among AVG 418
363 users (49%) compared to AVF users (15%) by 1 year (Fig 419
364 RESULTS 420
1B). Complete abandonment of the access or new access
365 Patient Characteristics creation (loss of secondary patency) was much less com- 421
366 Construction of the study cohort is shown in Figure S2. A mon than loss of primary patency, but still was more than 422
367 total of 60,329 AVF users and 17,763 AVG users who were 2-fold higher among AVG users (22%) compared to AVF 423
368 424
369 425
370 A Loss of primary unassisted patency B Loss of primary assisted patency
426
1.0
1.0
371 AVG
427
372 428
Cumulative incidence
0.2 0.4 0.6 0.8
0.2 0.4 0.6 0.8
373 AVF AVG

429
374 430
375 431
376 AVF
432
377 433
0.0
0.0
378 0 1 2 3 0 1 2 3
434
379 Years since start of using AV access Years since start of using AV access 435
380 436
381 C Loss of secondary patency
437
1.0
382 438
383 439
0.2 0.4 0.6 0.8
384 440
385 441
386 442
387 443
web 4C=FPO
AVG
388 AVF
444
0.0
389 0 1 2 3
445
390 Years since start of using AV access 446
391 447
Figure 1. Cumulative incidence of loss of access patency. Abbreviation: AV, arteriovenous.
392 448

Roetker et al
A Loss of primary unassisted patency Adjusted

449 Type N events Person-years HR (95% CI) 505
450 3-month 16663 10987 1 (Ref) 506
AVF
451 507
AVG 6226 3000 1.31 (1.27, 1.36)
452 508
1-year 26603 26030 1 (Ref)
453 AVF 509
454 AVG 10249 6120 1.49 (1.45, 1.52) 510
455 3-year AVF 30143 37270 1 (Ref) 511
456 AVG 11299 7252 1.56 (1.52, 1.60) 512
457 513
458 75 125 175
Crude rate (95% CI)
225
514
459 515
460 B Loss of primary assisted patency
Adjusted 516
461 Type N events Person-years HR (95% CI) 517
462 3-month AVF 3855 12635 1 (Ref) 518
463 AVG 4082 3267 3.42 (3.26, 3.58) 519
464 1-year AVF 7133 36526 1 (Ref) 520
465 AVG 6933 7901 3.66 (3.53, 3.79) 521
466 3-year 1 (Ref)
522
AVF 9173 61028
467 523
AVG 8163 10825 3.79 (3.67, 3.92)
468 524
469 10 40 70 100 130 525
470 Crude rate (95% CI) 526
471 527
C Loss of secondary patency
472 Adjusted 528
Type N events Person-years HR (95% CI)
473 529
474 3-month AVF 1605 12907 1 (Ref) 530
475 AVG 938 3663 1.75 (1.61, 1.91) 531
476 1-year AVF 2792 38298 1 (Ref) 532
477 AVG 1626 10200 1.88 (1.76, 2.01) 533
478 3-year AVF 3617 65697 1 (Ref) 534
479 2.03 (1.92, 2.16) 535
web 4C=FPO
AVG 2130 15975

480 536
481 5 10 15 20 25 30 537
Crude rate (95% CI)
482 Access type AVF AVG
538
483 539
484 Figure 2. Association of access type with risk of loss of patency. Hazard ratios are adjusted for age category, sex, race, Medicaid 540
485 enrollment, BMI category, primary cause of end-stage kidney disease, catheter at hemodialysis initiation, time since hemodialysis initi- 541
486 ation, comorbid conditions, interventions to maintain prefunctional patency, and prescription medications. Abbreviations: AVF, arte- 542
487 riovenous fistula; AVG, arteriovenous graft; CI, confidence interval; HR, hazard ratio; Ref, reference. 543
488 544
489 545
490 546
491 users (10%) by 3 years (Fig 1C). Patterns of incidence of AVG users versus AVF users were 1.49 (95% confidence 547
492 loss of access patency were similar among the subgroup of interval [CI], 1.45-1.52) for loss of primary unassisted 548
493 patients with a previous failed AVF (Fig S3). patency (Fig 2A), 3.66 (95% CI, 3.53-3.79) for loss of 549
494 primary assisted patency (Fig 2B), and 1.88 (95% CI, 550
495 Regression Models for Loss of Access Patency 1.76-2.01) for loss of secondary patency (Fig 2C). Asso- 551
496 Crude rates and adjusted HRs comparing AVG versus AVF ciations were similar in the subgroup analysis of patients 552
497 users for each loss of patency outcome through 3 months, with a previous failed AVF (Table S4). 553
498 1 year, and 3 years of follow-up are presented in Figure 2. 554
499 Patency loss was highest in the first 3 months after starting Stratified Cumulative Incidence 555
500 to use an AV access for HD and then plateaued with longer Cumulative incidence estimates for AVF and AVG users 556
501 follow-up. AVG use was consistently associated with stratified by subgroups of interest are presented in 557
502 higher risk of patency loss compared to use of an AVF, Figures 3-5. The most striking pattern was observed for 558
503 independent of the sociodemographic and clinical char- patients who underwent thrombectomy or thrombolysis 559
504 acteristics shown in Table 1. By 1 year, HRs comparing intervention before first use of the AV access for HD, 560

Roetker et al
1.0
1.0
1.0
By sex By age category By race
561 617
562 618
0.8
0.8
0.8
563 619
0.6
0.6
0.6
564 620
565 621
0.4
0.4
0.4
566 AVG, 18-44
AVG, 65-74
AVG, 45-64
AVG, 75-84 622
0.2
0.2
0.2
567 AVG, Male AVG, Female
AVG, 85+
AVF, 45-64
AVF, 18-44
AVF, 65-74
AVG, White
AVG, Other
AVG, Black
AVF, White
623
568 AVF, Male AVF, Female AVF, 75-84 AVF, 85+ AVF, Black AVF, Other
624
0.0
0.0
0.0
569 0 1 2
Years since start of using AV access
3 0 1 2
3 0 1 2
3 625
570 626
571 627
1.0
1.0
1.0
By BMI By time since HD initiation By prior thrombectomy/thrombolysis
572 628
573 629
0.8
0.8
0.8
574 630
0.6
0.6
0.6
575 631
576 632
0.4
0.4
0.4
577 AVG, <18.5 AVG, 18.5-24.9 633
0.2
0.2
0.2
578 AVG, 25-29.9
AVF, <18.5
AVG, 30+
AVF, 18.5-24.9 AVG, 6+ months AVG, <6 months AVG, No AVG, Yes 634
579 AVF, 25-29.9 AVF, 30+ AVF, 6+ months AVF, <6 months AVF, No AVF, Yes
635
0.0
0.0
0.0
580 0 1 2
3 0 1 2
3 0 1 2
3
636
581 637
582 638
1.0
1.0
1.0
By prior maturation failure By history of atrial fibrillation By antiplatelet use
583 639
584 640
0.8
0.8
0.8
585 641
0.6
0.6
0.6
586 642
587 643
0.4
0.4
0.4
588 644
web 4C=FPO
0.2
0.2
0.2
589 AVG, No AVG, Yes AVG, No AVG, Yes AVG, No AVG, Yes 645
590 AVF, No AVF, Yes AVF, No AVF, Yes AVF, No AVF, Yes
646
0.0
0.0
0.0
591 0 1 2
3 0 1 2
3 0 1 2
3
647
592 648
593 Figure 3. Cumulative incidence of loss of primary unassisted patency stratified by patient characteristics. Abbreviations: AV, arterio- 649
594 venous; AVF, arteriovenous fistula; AVG, arteriovenous graft; BMI, body mass index; HD, hemodialysis. 650
595 651
596 652
597 which was highly predictive of loss of primary unassisted BMI (Figs 3-5). Individuals who had experience prior 653
598 and assisted patency (Figs 3-5); notably, in the first year of maturation failure had higher incidence of loss of primary 654
599 using an AV access for HD, individuals with an AVF who unassisted and primary assisted patency but not secondary 655
600 had undergone a prior thrombectomy or thrombolysis had patency. Lastly, incidence of patency loss was similar 656
601 a higher incidence of primary unassisted patency loss than between users and nonusers of antiplatelet medications 657
602 individuals with an AVG who had not undergone a pre- (Figs 3-5). 658
603 vious thrombectomy or thrombolysis. Another prominent 659
604 finding was that Black individuals, particularly AVF users, Stratified Regression Models 660
605 had a higher incidence of loss of primary patency Figures S4-S6 show adjusted HRs, stratified by subgroups 661
606 compared to individuals of White or other race who were of interest, for the comparison of AVG versus AVF for each 662
607 using the same access type (Figs 3-5). Perhaps unexpect- patency loss outcome through 1 year of follow-up. The 663
608 edly, the youngest individuals (those aged 18-44 years) stratified HRs were generally similar to the overall 1-year 664
609 generally had higher incidence of patency loss than older HRs for each outcome. However, there was evidence of 665
610 individuals. Among AVF users, women had a higher interaction in some instances. Most notably, comparing 666
611 incidence of patency loss of all types compared to men AVG users to AVF users, the relative risk of loss of primary 667
612 (Figs 3-5). Conversely, among AVG users, men had a assisted patency was lower among patients with a history 668
613 higher incidence of secondary patency loss (Fig 5). of thrombectomy or thrombolysis (HR, 2.60; 95% CI, 669
614 Furthermore, patients in the obese BMI category (≥30 kg/ 2.33-2.89) than among patients without such a history 670
615 m2) and using an AVG had a higher incidence of each type (HR, 3.79; 95% CI, 3.65-3.94) (P for interaction <0.001). Q4671
616 of patency loss compared to those in other categories of Also, the relative difference in risk of loss of primary 672

Roetker et al
1.0
1.0
1.0
673 AVG, Male
AVF, Male
AVG, Female
AVF, Female
AVG, 18-44
AVG, 85+
AVG, 45-64
AVF, 18-44
AVG, 65-74
AVF, 45-64
AVG, 75-84
AVF, 65-74
AVG, White
AVF, White
AVG, Black
AVF, Black
AVG, Other
AVF, Other
729
674 730
0.8
0.8
0.8
AVF, 75-84 AVF, 85+
675 731
0.6
0.6
0.6
676 732
677 733
0.4
0.4
0.4
678 734
0.2
0.2
0.2
679 735
680 736
0.0
0.0
0.0
681 0 1 2
3 0 1 2
3 0 1 2
3 737
682 738
683 739
1.0
1.0
1.0
684 AVG, <18.5
AVG, 30+
AVG, 18.5-24.9
AVF, <18.5
AVG, 25-29.9
AVF, 18.5-24.9
AVG, 6+ months
AVF, 6+ months
AVG, <6 months
AVF, <6 months
AVG, No
AVF, No
AVG, Yes
AVF, Yes
740
685 741
0.8
0.8
0.8
AVF, 25-29.9 AVF, 30+
686 742
0.6
0.6
0.6
687 743
688 744
0.4
0.4
0.4
689 745
0.2
0.2
0.2
690 746
691 747
0.0
0.0
0.0
692 0 1 2
3 0 1 2
3 0 1 2
3 748
693 749
694 750
1.0
1.0
1.0
695 AVG, No
AVF, No
AVG, Yes
AVF, Yes
AVG, No
AVF, No
AVG, Yes
AVF, Yes
AVG, No
AVF, No
AVG, Yes
AVF, Yes 751
696 752
0.8
0.8
0.8
697 753
0.6
0.6
0.6
698 754
699 755
0.4
0.4
0.4
700 756
web 4C=FPO
0.2
0.2
0.2
701 757
702 758
0.0
0.0
0.0
703 0 1 2
3 0 1 2
3 0 1 2
3
759
704 760
705 Figure 4. Cumulative incidence of loss of primary assisted patency stratified by patient characteristics. Abbreviations: AV, arteriove- 761
706 nous; AVF, arteriovenous fistula; AVG, arteriovenous graft; BMI, body mass index; HD, hemodialysis. 762
707 763
708 764
709 assisted patency and loss of secondary patency comparing to AVGs, particularly in terms of (primary) assisted 765
710 AVG users to AVF users was smaller for women than for patency and secondary patency. 766
711 men (P for interaction <0.001 for each outcome). Our goal was to assist nephrologists and other providers 767
712 in assessing the risk of access failure outcomes in users of 768
713 permanent AV accesses, by access type and patient char- 769
714 DISCUSSION acteristics, from the time of first use of the AV access for 770
715 In this study, we used a large cohort of patients receiving HD. Second, we did not seek to mimic the intention-to- 771
716 maintenance HD from USRDS data to determine absolute treat approach, used by some others, to inform which 772
717 and relative risks of patency loss between AVGs and AVFs. access creation strategy might be best for a given patient. 773
718 We found that, among patients receiving HD who had Executing an intention-to-treat–like approach is somewhat 774
719 successfully used a permanent access, loss of unassisted complex in the USRDS data. The intention-to-treat 775
720 patency at 1 year for an AVG, as compared with an AVF, approach requires explicit consideration of patients who 776
721 was about 1.3-fold higher; loss of primary assisted patency die before dialysis initiation. However, pre-ESKD claims 777
722 was about 3.3-fold higher; and loss of secondary patency are available in the USRDS only for patients who survive to 778
723 was about 2.0-fold higher. After adjustment for a wide initiate dialysis, meaning an analysis of access creation 779
724 range of demographic, socioeconomic, anthropometric, before dialysis initiation would introduce immortal time 780
725 comorbidity, medication, and access history–related fac- bias. Another intention-to-treat approach, which has been 781
726 tors, the modeled HRs demonstrated a similar signal. The used by others, is to follow patients who initiated HD with 782
727 results suggest that, among patients with a functioning a catheter from the time their first AV access was created. 783
728 permanent access, AVFs have superior patency compared However, in this design, those who successfully use a 784

Roetker et al
0.40
0.40
0.40
785 AVG, Male
AVF, Male
AVG, Female
AVF, Female
AVG, 18-44
AVG, 65-74
AVG, 45-64
AVG, 75-84
AVG, White
AVG, Other
AVG, Black
AVF, White
841
786 AVG, 85+ AVF, 18-44 AVF, Black AVF, Other 842
0.30
0.30
0.30
AVF, 45-64 AVF, 65-74
787 843
AVF, 75-84 AVF, 85+
788 844
0.20
0.20
0.20
789 845
790 846
0.10
0.10
0.10
791 847
792 848
0.00
0.00
0.00
793 0 1 2
3 0 1 2
3 0 1 2
3 849
794 850
795 851
0.40
0.40
0.40
796 AVG, <18.5
AVG, 25-29.9
AVG, 18.5-24.9
AVG, 30+
AVG, 6+ months
AVF, 6+ months
AVG, <6 months
AVF, <6 months
AVG, No
AVF, No
AVG, Yes
AVF, Yes
852
797 AVF, <18.5 AVF, 18.5-24.9 853
0.30
0.30
0.30
AVF, 25-29.9 AVF, 30+
798 854
799 855
0.20
0.20
0.20
800 856
801 857
0.10
0.10
0.10
802 858
803 859
0.00
0.00
0.00
804 0 1 2
3 0 1 2
3 0 1 2
3
860
805 861
806 862
0.40
0.40
0.40
807 AVG, No
AVF, No
AVG, Yes
AVF, Yes
AVG, No
AVF, No
AVG, Yes
AVF, Yes
AVG, No
AVF, No
AVG, Yes
AVF, Yes 863
808 864
0.30
0.30
0.30
809 865
810 866
0.20
0.20
0.20
811 867
812 868
web 4C=FPO
0.10
0.10
0.10
813 869
814 870
0.00
0.00
0.00
815 0 1 2
3 0 1 2
3 0 1 2
3
871
816 872
817 Figure 5. Cumulative incidence of loss of secondary patency stratified by patient characteristics. Abbreviations: AV, arteriovenous; 873
818 AVF, arteriovenous fistula; AVG, arteriovenous graft; BMI, body mass index; HD, hemodialysis. 874
819 875
820 876
821 fistula or graft at HD initiation are excluded. To overcome techniques have improved over time, allowing salvage of 877
822 these limitations, we selected an approach that is akin to an access that would otherwise have been abandoned. Un- 878
823 “as-treated” analysis, whereby risks of patency outcomes surprisingly, analyses following patients from the time of 879
824 are modeled conditional on having achieved a functioning AVF or AVG creation have reported a much higher inci- 880
825 fistula or graft. Thus, our design allows us to include all AV dence of secondary patency loss (eg, 41% for AVFs and 881
826 access users (as opposed to only the subset who have pre- 43% for AVGs).6 This is explained by the high rate of 882
827 ESKD claims or who initiate HD with a catheter). Use of unsuccessful access maturation. Conversely, one study 883
828 any future access-preserving intervention, such as a novel even reported worse secondary patency outcomes in AVF 884
829 pharmacologic agent, would almost certainly be explored users.11 885
830 in prevalent patients dialyzing with a permanent access, Our subgroup analyses revealed notable findings. We 886
831 meaning that our design is particularly relevant in a sce- found that loss of primary and secondary patency among 887
832 nario where foundational data on access survival is AVF users was higher in women than in men, in concor- 888
833 required. dance with a previous analysis.12 Conversely, among AVG 889
834 Although our study design has some unique aspects, users, we found that loss of secondary patency was higher 890
835 some comparisons can still be drawn with other studies. in men compared to women, in contrast with the previous 891
836 We found a lower incidence of access abandonment at 1 study, which found little difference by sex.12 In general 892
837 year (4% for AVFs and 10% for AVGs) than did an earlier concordance with other work, however, we generally 893
838 analysis of USRDS data by Lee et al5 (18% for AVFs and found that Black, as compared with White, patients fared 894
839 24% for AVGs), which was also conditioned on successful more poorly across all patency types, with the exception of 895
840 first use of the AV access. It is possible that thrombectomy secondary patency in AVG users, where outcomes were 896

Roetker et al
similar in Black and White patients.13 Patients with the derived from USRDS data, our patency taxonomy, and our
897 highest BMIs generally had worse AVG patency than pa- use of monthly modifier codes as reported by dialysis fa- 953
898 tients with lower BMIs; other studies have also shown an cilities to Medicare, which we used to improve the accu- 954
899 association between obesity and poor patency out- racy of classifying vascular access during follow-up. 955
900 comes.11,14 Generally, we did not observe differences in In conclusion, among patients undergoing maintenance 956
901 patency between users and nonusers of antiplatelet agents. HD who have successfully used a permanent HD access, 957
902 Similarly, in a clinical trial of patients with newly created AVGs were associated with nearly 4-fold higher risk of 958
903 AVFs, clopidogrel was ineffective at promoting successful having primary assisted patency loss and approximately 2- 959
904 maturation; nevertheless, in contrast with our study, the fold higher risk of secondary patency loss compared with 960
905 trial found that clopidogrel reduced early AVF thrombosis.15 AVFs over long-term follow-up (1-3 years). Interventions, 961
906 However, the results of that trial did not address whether such as novel pharmacologic approaches, are needed to 962
907 clopidogrel improves AVF patency over the long term after improve access patency in patients receiving HD, particu- 963
908 successful first use of the permanent access—which was the larly patients who are unable to receive a fistula access. 964
909 conceptual framework of our present study. 965
910 The Fistula First Breakthrough Initiative and the 2006 966
911 Kidney Disease: Improving Global Outcomes (KDIGO) SUPPLEMENTARY MATERIAL 967
912 Vascular Access Guidelines16 lionized a fistula-based Supplementary File (PDF) 968
913 approach, which was interpreted by many as meaning Figure S1: Schematic of study outcomes. AV, arteriovenous 969
914 that grafts should be considered merely as a final recourse Figure S2: Construction of the study cohort 970
915 when attempting to secure a permanent access. Doubts Figure S3: Cumulative incidence of loss of access patency in the 971
916 have been expressed about the original fistula-centric subgroup of patients with a previous failed AVF 972
917 approach, appropriately in our view, by important publi- Figure S4: Hazard ratios, comparing AVG users versus AVF users, 973
918 cations.17-19 Indeed, the goals of the initiative were sub- for one-year risk of loss of primary unassisted patency, stratified by 974
919 sequently clarified and the initiative itself renamed.20 The patient characteristics 975
920 main challenge to a fistula-centric approach is the danger Figure S5: Hazard ratios, comparing AVG users versus AVF users, 976
921 of nonmaturation of the fistula and subsequent need for for one-year risk of loss of primary assisted patency, stratified by 977
922 prolonged central venous catheter use.21 When fistula patient characteristics 978
923 nonmaturation is considered, the relative benefits of fis- Figure S6: Hazard ratios, comparing AVG users versus AVF users, 979
924 tulas, relative to grafts, become less strong. As such, our for one-year risk of loss of secondary patency, stratified by patient 980
925 results should not be interpreted as supporting a fistula characteristics 981
926 first–based approach. Rather, our goal was to compare Table S1: Algorithms Used to Identify Loss of Vascular Access 982
927 outcomes in grafts to fistulas in patients who had achieved Patency 983
928 a functioning permanent access. Such a comparison is a Table S2: Comorbid Condition Definitions 984
929 useful component of the fistula-versus-graft debate, and Table S3: Numbers of Previous Failed AVFs and AVGs by Access 985
930 the incidence curves we generated may be useful to a Type 986
931 nephrologist rounding in a dialysis facility who is advising Table S4: Association of Access Type With Risk of Loss of Patency 987
932 a patient currently using an AV access on potential future in the Subgroup of Patients With a Previous Failed AVF 988
933 outcomes or contemplating, for example, use of a new 989
934 therapy designed to preserve access patency. ARTICLE INFORMATION 990
935 Our study has important limitations. First, as noted Authors’ Full Names and Academic Degrees: Nicholas S. Roetker,
991
936 above, our study was deliberately designed to examine the PhD, MS, Haifeng Guo, MS, Dena Rosen Ramey, ABD, Ciaran J. 992
937 question of outcomes after first successful use of a per- McMullan, MB, BCh, BAO, G. Brandon Atkins, MD, PhD, and 993
938 manent access and was not designed to inform the ques- James B. Wetmore, MD, MS. 994
939 tion of which access might be most suitable from initial Authors’ Affiliations: Chronic Disease Research Group, Hennepin 995
940 creation in, for example, patients with late-stage chronic Healthcare Research Institute, Minneapolis, Minnesota (NSR, HG, 996
JBW); Merck & Co, Inc, Kenilworth, New Jersey (DRR, CJM,
941 kidney disease approaching dialysis. Such an approach, 997
GBA); Division of Nephrology, Hennepin County Medical Center
942 which cannot explicitly account for primary non- and Department of Medicine, University of Minnesota, Minneapolis, 998
943 maturation of an AVF (such as one placed before or soon Minnesota (JBW). 999
944 after initiation of HD), may make outcomes of AVGs Address for Correspondence: Nicholas S. Roetker, PhD, MS, 1000
945 appear relatively worse than AVFs, creating a possible Chronic Disease Research Group, Hennepin Healthcare Research 1001
946 “bias” against AVGs. Our findings cannot, therefore, Institute, 701 Park Ave, Suite S2.100, Minneapolis, MN 55415. 1002
947 address which access type might be most suitable as an Email: nick.roetker@cdrg.org 1003
948 initial attempt at access creation. Second, residual con- Authors’ Contributions: Research idea and study design: NSR, 1004
DRR, CJM, GBA, JBW; data acquisition: NSR, HG; data analysis/
949 founding is likely present, despite our extensive attempts at 1005
interpretation: NSR, HG, DRR, CJM, GBA, JBW; statistical
950 covariate control. Third, our analysis is based on claims, analysis: NSR, HG; supervision or mentorship: DRR, JBW. Each 1006
951 which induces some degree of imprecision. These limita- author contributed important intellectual content during manuscript 1007
952 tions may be counterbalanced by our use of a large cohort drafting or revision and accepts accountability for the overall work 1008

Roetker et al
by ensuring that questions pertaining to the accuracy or integrity of hemodialysis in the United States. J Vasc Surg. 2018;68(4):
1009 any portion of the work are appropriately investigated and resolved. 1166-1174. 1060
7. Johansen KL, Chertow GM, Foley RN, et al. US Renal Data
1010 Support: This study was funded by Merck Sharp & Dohme Corp, a 1061
subsidiary of Merck & Co, Inc, Kenilworth, New Jersey. The authors System 2020 Annual Data Report: epidemiology of kidney
1011 employed by the funder (DRR, CJM, GBA) had a role, jointly with the disease in the United States. Am J Kidney Dis. 2021;77 1062
1012 other coauthors, in the study design, interpretation of the study (4)(suppl 1):A7-A8. doi:10.1053/j.ajkd.2021.01.002 1063
1013 findings, drafting of the manuscript, and the decision to submit the 8. Thamer M, Lee TC, Wasse H, et al. Medicare costs associated 1064
1014 manuscript for publication. with arteriovenous fistulas among US hemodialysis patients. 1065
1015 Financial Disclosure: NSR receives grant/research support to Am J Kidney Dis. 2018;72(1):10-18. 1066
1016 Chronic Disease Research Group from Amgen, the Centers for 9. Kalbfleisch JD, Prentice RL. The Statistical Analysis of Failure 1067
Time Data. Wiley; 1980.
1017 Disease Control and Prevention, and the National Institutes of 1068
Health (NIH). JBW has been a consultant for Aurinia and Reata 10. Rubin DB. Multiple Imputation for Nonresponse in Surveys.
1018 and receives grant/research support to Chronic Disease Research Wiley; 1987. 1069
1019 Group from NIH (National Institute of Diabetes and Digestive and 11. Copeland T, Lawrence P, Woo K. Outcomes of initial hemodi- 1070
1020 Kidney Diseases), OPKO, Merck, Relypsa, Genentech, Bristol alysis vascular access in patients initiating dialysis with a 1071
1021 Q5 Myers Squibb, and Acadia. DRR, CJM, and GBA are employed by tunneled catheter. J Vasc Surg. 2019;70(4):1235-1241. 1072
1022 the funder of this study. HG has no relevant financial interests. 12. Arhuidese IJ, Faateh M, Meshkin RS, Calero A, Shames M, 1073
1023 Acknowledgements: The authors thank Chronic Disease Research Malas MB. Gender-based utilization and outcomes of autoge- 1074
Group colleague Anna Gillette for manuscript editing. nous fistulas and prosthetic grafts for hemodialysis access.
1024 1075
Ann Vasc Surg. 2020;65:196-205.
1025 Disclaimer: The data reported here have been supplied by the
13. Arhuidese IJ, Aji EA, Muhammad R, Dhaliwal J, Shukla AJ, 1076
United States Renal Data System (USRDS). The interpretation
1026 and reporting of these data are the responsibility of the author(s) Malas MB. Racial differences in utilization and outcomes of 1077
1027 and in no way should be seen as an official policy or interpretation hemodialysis access in the United States. J Vasc Surg. 1078
1028 of the U.S. government. 2020;71(5):1664-1673. 1079
1029 Peer Review: Received March 4, 2022. Evaluated by 2 external peer 14. Arhuidese IJ, Holscher CM, Elemuo C, Parkerson GR, 1080
1030 reviewers, with direct editorial input from the Statistical Editor, an Johnson BL, Malas MB. Impact of body mass index on out- 1081
comes of autogenous fistulas for hemodialysis access. Ann
1031 Associate Editor, and the Editor-in-Chief. Accepted in revised form 1082
September 11, 2022. Vasc Surg. 2020;68:192-200.
1032 15. Dember LM, Beck GJ, Allon M, et al. Effect of clopidogrel on 1083
1033 early failure of arteriovenous fistulas for hemodialysis: a ran- 1084
1034 domized controlled trial. JAMA. 2008;299(18):2164-2171. 1085
1035 REFERENCES 16. Vascular Access 2006 Work Group. Clinical practice guide- 1086
1036 1. Lok CE, Huber TS, Lee T, et al. National Kidney F. KDOQI lines for vascular access. Am J Kidney Dis. 2006;48(suppl 1): 1087
1037 clinical practice guideline for vascular access: 2019 update. S176-S247. 1088
Am J Kidney Dis. 2020;75(4)(suppl 2):S1-S164. 17. Lee T, Ullah A, Allon M, et al. Decreased cumulative access
1038 1089
2. Brown RS. Barriers to optimal vascular access for hemodialy- survival in arteriovenous fistulas requiring interventions to pro-
1039 sis. Semin Dial. 2020;33(6):457-463. mote maturation. Clin J Am Soc Nephrol. 2011;6(3):575-581. 1090
1040 3. Sidawy AN, Gray R, Besarab A, et al. Recommended stan- 18. Dember LM. Fistulas first–but can they last? Clin J Am Soc 1091
1041 dards for reports dealing with arteriovenous hemodialysis ac- Nephrol. 2011;6(3):463-464. 1092
1042 cesses. J Vasc Surg. 2002;35(3):603-610. 19. Allon M. Vascular access for hemodialysis patients: new data 1093
1043 4. Bylsma LC, Reichert H, Gage SM, et al. Clinical outcomes of should guide decision making. Clin J Am Soc Nephrol. 1094
1044 arteriovenous access in incident hemodialysis patients with 2019;14(6):954-961. 1095
Medicare coverage, 2012-2014. Am J Nephrol. 2019;49(2): 20. ESRD National Coordinating Center. Fistula first catheter last.
1045 1096
156-164. Accessed October 28, 2021. https://esrdncc.org/en/fistula-
1046 5. Lee T, Qian J, Thamer M, Allon M. Tradeoffs in vascular access first-catheter-last/ 1097
1047 selection in elderly patients initiating hemodialysis with a cath- 21. Allon M, Imrey PB, Cheung AK, et al. Relationships between 1098
1048 eter. Am J Kidney Dis. 2018;72(4):509-518. clinical processes and arteriovenous fistula cannulation and 1099
1049 6. Arhuidese IJ, Orandi BJ, Nejim B, Malas M. Utilization, patency, maturation: a multicenter prospective cohort study. Am J Kid- 1100
1050 and complications associated with vascular access for ney Dis. 2018;71(5):677-689. 1101
1051 1102
1052 1103
1053 1104
1054 1105
1055 1106
1056 1107
1057 1108
1058 1109
1059 1110


Acceso Vascular Metanalisis

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Acceso Vascular Metanalisis

Uploaded by

Copyright:

Available Formats

Original Research

1 Hemodialysis Access Type and Access Patency Loss: An 57

Kidney Med Vol XX | Iss XX | Month 2022 | 100567 1

Patients aged <18 years or who received an oral antico-

2 Kidney Med Vol XX | Iss XX | Month 2022 | 100567

Table 1. Baseline Characteristics in Patients Receiving Hemodialysis by Access Type

Kidney Med Vol XX | Iss XX | Month 2022 | 100567 3

0.2 0.4 0.6 0.8

373 AVF AVG

4 Kidney Med Vol XX | Iss XX | Month 2022 | 100567

A Loss of primary unassisted patency Adjusted

AVG 2130 15975

Kidney Med Vol XX | Iss XX | Month 2022 | 100567 5

6 Kidney Med Vol XX | Iss XX | Month 2022 | 100567

Kidney Med Vol XX | Iss XX | Month 2022 | 100567 7

8 Kidney Med Vol XX | Iss XX | Month 2022 | 100567

Kidney Med Vol XX | Iss XX | Month 2022 | 100567 9

10 Kidney Med Vol XX | Iss XX | Month 2022 | 100567

You might also like