Abstract
Rationale & Objective:
As the proportion of arteriovenous fistulae (AVF) compared to arteriovenous grafts (AVG) in the United States has increased, there has been a concurrent increase in interventions. We sought to explore AVF and AVG maturation and maintenance procedural burden in the first year of hemodialysis.
Study Design:
Observational cohort study.
Setting & Participants:
Patients initiating hemodialysis from July 1, 2012, to December 31, 2014 and having a first time AVF or AVG placement between dialysis initiation and 1 year were identified (N=73,027) using the United States Renal Data System (USRDS).
Predictors:
Patient characteristics.
Outcomes:
Successful AVF/AVG use and intervention procedure burden.
Analytical Approach:
For each group, we analyzed interventional procedure rates during maturation maintenance phases using Poisson regression. We used proportional rate modelling for covariate-adjusted analysis of interventional procedure rates during the maintenance phase.
Results:
During maturation phase, 13,989 of 57,275 (24.4%) patients in the AVF group required intervention, with therapeutic interventional requirements of 0.36 per person (pp). In the AVG group, 2,904 of 15,572 (18.4%) patients required intervention during maturation, with therapeutic interventional requirements of 0.28 pp. During maintenance phase, in the AVF group, 12,732 of 32,115 (39.6%) patients required intervention, with a therapeutic intervention rate of 0.93 per person-year (ppy). During maintenance phase, in the AVG group, 5,928 of 10,271 (57.7%) patients required intervention, with a therapeutic intervention rate of 1.87 ppy. For both phases, intervention rates for AVF tended to be higher on the East Coast, while those for AVG were more uniform geographically.
Limitations:
This study relies on administrative data, with monthly recording of access use.
Conclusions:
During maturation, interventions for both AVF and AVG were relatively common. Once successfully matured, AVF had lower maintenance interventional requirements. During maturation and maintenance phases, there were geographic variations in AVF intervention rates that warrant additional study.
Keywords: Arteriovenous fistula (AVF), Arteriovenous graft (AVG), Endovascular interventions, Vascular access, United States Renal Data System (USRDS)
PLAIN LANGUAGE SUMMARY
To connect to the hemodialysis machine, patients with kidney failure require vascular access such as an arteriovenous fistula or graft. To develop and maintain arteriovenous fistulas and grafts, endovascular interventions are common. We studied the rates of interventions on arteriovenous fistulas and grafts during maturation and maintenance phases in patients new to hemodialysis who had arteriovenous fistulas and grafts placed for the first time during the first year of hemodialysis treatment. We found that during maturation, grafts required fewer interventions. If successfully mature, fistulae had lower interventional requirements. For fistulae, but not grafts, we also noted that there were differences in the number of interventions based on geographic location, with a higher burden on the East Coast.
INTRODUCTION
Although the proportion of patients initiating hemodialysis (HD) with a dialysis catheter remains stable, there’s been a gradual increase in the proportion of AVF in prevalent patients since 2003 (1, 2). A number of patient and provider characteristics impact successful AVF placement (3, 4), with much of the AVF increase occurring in the first HD year (1). However, this increase has not occurred without significant effort. The increased AVF placements has been associated with greater requirement for angioplasties and other interventions (5). This increased intervention rate has been associated with significant cost (6). Alongside these developments, there is increased focus on individualizing the type of dialysis access to patient characteristics, with greater acceptance of arteriovenous grafts (AVG) (7). However, AVF are still thought to have some advantages over AVG, particularly as AVG are thought to require more interventions to maintain patency (8, 9), with the associated health-related quality of life leaning towards AVF (10). Developing protocols for individualization of access type, as well as broader system-based resource allocation, remains a topic of intense research. As the first year following hemodialysis initiation sees the most profound changes in dialysis access, we sought to explore vascular access maturation and maintenance procedural burden in incident HD patients and to explore regional procedural variation.
METHODS
Study Cohort
Using data from the United States Renal Data System (USRDS), we created a cohort of all incident hemodialysis patients from July 1, 2012, through December 31, 2014. The analytic cohort was then limited to patients with first-time AVF or AVG placement between their HD start date and 1 year following HD initiation (Figure 1A). AVF placement was defined by CPT codes 36818–36821 or 36825 in USRDS Medicare claims data; for AVG placement, CPT code 36830 was used. Demographic data was obtained from the USRDS PATIENTS file and the Centers for Medicare & Medicaid Services (CMS) ESRD Medical Evidence Report. The cohort was linked to CROWNWeb, a web-based data collection system implemented across all Medicare-certified dialysis facilities throughout the United States in June 2012, replacing the Standard Information Management System. CROWNWeb incorporates multiple clinical data elements, including monthly vascular access data. Vascular access data for the dates of first and last use of each AVF or AVG placement, until December 31, 2016, were utilized. Interventional procedure data were obtained from USRDS Medicare claims, using a physician-supplier, inpatient, and outpatient claims. The CPT codes shown in Figure 1B were pulled from the claims for 2012 through 2016. As multiple procedures were often performed at the same encounter as part of a single comprehensive therapeutic procedure, a hierarchy to identify the dominant procedure was developed (Figure 1B). These data were merged with patient cohort and follow-up data; any interventional procedure claims that did not fall between the AVF/AVG placement date and the end of follow-up date for a patient were excluded.
Figure 1.
A. Study flowchart. B. Procedure hierarchy.
Outcomes
Maturation phase was defined as the time period between the date of AVF/AVG placement and date of first use of that access. As previously described (3), successful maturation was defined as documentation of first AVF/AVG use in monthly CROWNWeb reporting of vascular access in use. The term “maturation” was used for both AVF and AVG for the sake of readability, acknowledging that the term is not strictly correct for AVG. If an AVF or AVG was not recorded as used within one year of placement, or another type of vascular access was recorded during that time period, then it was labeled as failing to mature. The follow-up period for an access ended at one year after placement unless any of these events occurred: new vascular access placement, transplant, switch to peritoneal dialysis, death of the patient, or December 31, 2016. Patients with AVF/AVG placements that never had AVF, AVG, or catheter use recorded in CROWNWeb were considered lost to follow-up and excluded from analysis.
Natural (“unassisted”) maturation was defined as the lack of any therapeutic intervention procedures on a placement during maturation phase. Assisted maturation was defined as a placement requiring at least one therapeutic interventional procedure during maturation phase. Placements that matured were followed in CROWNWeb, with the time period between first AVF/AVG use and the end of follow-up defined as maintenance phase. Follow-up ended at the earliest of any of these events: new vascular access placement, kidney transplant, switch to peritoneal dialysis, death, one year after first use, or December 31, 2016.
Statistical Analysis
Interventional procedures rates were analyzed separately for maturation and maintenance phases. During maturation, we made comparisons of the interventional procedure rates per person (pp) between patients with successful access maturation and patients whose access failed to mature. In maintenance phase, comparisons in interventional procedure rates per person-year (ppy) were made between patients with natural maturation and assisted maturation. Comparisons in interventional procedure rates by type of interventional procedure and patient characteristics within maturation and maintenance phases were performed using Poisson regression within AVF and AVG subgroups.
To take into account time-to-event during the maintenance phase, covariate-adjusted analysis of intervention rates during the maintenance phase were modeled using the proportional rates model (11). This model, which is essentially the Cox model analog for recurrent events, has been the strong default for rate modeling since shortly after its development. This approach allows for analysis of time to intervention within the maintenance phase, while also including all intervention data for a patient, rather than using only the first intervention as a typical Cox model would. Correlations among events within-individual are accommodated using a robust (“sandwich”) estimator which treats the patients (as opposed to events) as independent units. Note that, unlike parametric approaches (e.g., Poisson regression), the proportional rates model does not assume that the baseline rate is constant over time; this additional flexibility is a clear advantage. Predictors for this model include assisted maturation (versus natural maturation), the presence of a graft (versus fistula), previous intervention type during the maintenance phase, demographics, patient clinical factors, and geographic information. Demographic characteristics with fewer patients were collapsed into a larger group or, in the case of the pediatric age category, excluded.
IRB Statement and Software
This study was performed under the USRDS Coordinating Center contract with the National Institutes of Health (NIH) National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK); research as part of the contract has been approved by the University of Michigan’s Institutional Review Board (HUM0086162). Because data for USRDS components are collected by federal mandate, there are no individual patient consent requirements. Statistical analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC, USA).
RESULTS
Among the 259,216 incident HD patients, there were 57,275 first-time AVF placements and 15,752 first time AVG placements, with 373 fistula placements and 60 graft placements that were lost to follow-up excluded (Figure 1A). Demographics of the cohort are shown in Table 1. Of the AVF placements, 32,115 of 57,275 (56.1%) were successfully utilized during the study period. For AVG placements, 10,271 of 15,752 (65.2%) were successfully utilized during the study period. Successful maturation was greater for AVFs that had an intervention during the maturation phase when compared to AVFs that had no interventions during this phase (9718 of 13989, 69%, versus 22,397 of 43,286, 52%, respectively, p<0.001). In the AVG group, there was less maturation success for AVGs that had an intervention during the maturation phase, when compared to AVGs that had no interventions during the maturation phase (1,698 of 2,904, 58%, vs 8,573 of 12,848, 67%, respectively, p<0.001). With respect to follow-up time, 69.6% in the AVF group and 57.6% in the AVG group were followed for the full 365 days post first access use. Median time to maturation was 113 days (IQR 92–177 days) for AVF and 56 days (IQR 36–77 days) for AVG.
TABLE 1:
Demographics
| Total | AVF | AVG | |
|---|---|---|---|
| Total | 73,027 | 57,275 | 15,752 |
| Age | |||
| 0–21 | 159 | 147 | 12 |
| 22–44 | 5,870 | 4,827 | 1,043 |
| 45–64 | 23,381 | 18,893 | 4,488 |
| 65–74 | 21,515 | 17,081 | 4,434 |
| 75+ | 22,102 | 16,327 | 5,775 |
| Sex | |||
| Male | 40,259 | 33,393 | 6,866 |
| Female | 32,768 | 23,882 | 8,886 |
| Race | |||
| White | 48,134 | 39,135 | 8,999 |
| Black/African American | 20,977 | 14,990 | 5,987 |
| American Indian or Alaska Native | 782 | 697 | 85 |
| Asian | 2,306 | 1,774 | 532 |
| Native Hawaiian or Pacific Islander | 674 | 563 | 111 |
| Other or Multiracial | 154 | 116 | 38 |
| Hispanic Ethnicity | |||
| Hispanic | 10,617 | 8,753 | 1,864 |
| Non-Hispanic | 62,311 | 48,452 | 13,859 |
| Pre-ESRD Nephrology Care | |||
| Yes | 39,139 | 30,249 | 8,890 |
| No | 33,888 | 27,026 | 6,862 |
| Diabetes | |||
| Yes | 47,150 | 36,978 | 10,172 |
| No | 25,877 | 20,297 | 5,580 |
| Primary Cause of ESRD | |||
| Diabetes | 35,668 | 28,092 | 7,576 |
| Hypertension | 23,291 | 18,037 | 5,254 |
| Glomerulonephritis | 4,160 | 3,331 | 829 |
| Cystic kidney | 670 | 520 | 150 |
| Other urologic | 995 | 777 | 218 |
| Other cause | 6,071 | 4,818 | 1,253 |
| Unknown cause | 1,746 | 1,383 | 363 |
| Network | |||
| (01 CT) Net. of New England | 426 | 317 | 109 |
| (02 NY) Net. of N.Y. | 2,163 | 1,726 | 437 |
| (03 NJ) Trans-Atlantic R. C. | 3,445 | 2,701 | 744 |
| (04 PA) ESRD Net. Org. #4 | 2,915 | 2,249 | 666 |
| (05 VA) Mid Atlantic R. C. | 2,724 | 2,117 | 607 |
| (06 NC) Southeastern Kidney Council | 4,385 | 3,380 | 1,005 |
| (07 FL) ESRD Net. of Florida | 6,670 | 5,176 | 1,494 |
| (08 MS) Network 8 | 4,780 | 3,587 | 1,193 |
| (09 IN) Tri-state R. N. | 4,627 | 3,575 | 1,052 |
| (10 IL) Renal Net. of Illinois | 5,768 | 4,548 | 1,220 |
| (11 MN) Renal Net. of Upper Midwest | 3,238 | 2,487 | 751 |
| (12 MO) ESRD net. #12 | 4,710 | 3,738 | 972 |
| (13 OK) ESRD net. #13 | 2,933 | 2,332 | 601 |
| (14 TX) Net. of Texas | 3,482 | 2,744 | 738 |
| (15 CO) Inter-mountain ESRD net. | 7,653 | 5,996 | 1,657 |
| (16 WA) Northwest Renal Net. | 2,914 | 2,459 | 455 |
| (17 N-CA) Trans-Pacific ESRD Net. | 1,788 | 1,460 | 328 |
| (18 S-CA) Southern California Net. | 3,210 | 2,586 | 624 |
Maturation Phase
Among patients with AVF, 13,989 of 57,275 (24.4%) required intervention during maturation, with overall interventional requirements of 0.51 procedures per patient (pp) and therapeutic interventional requirements of 0.36 pp (Table 2A). Among patients with AVG, 2,904 of 15,752 (18.4%) required intervention during maturation, with overall interventional requirements of 0.35 pp (p<0.001 vs AVF) and therapeutic interventional requirements of 0.28 pp (p<0.001 vs AVF). Requirements for a new catheter, catheter exchange, or catheter intervention were somewhat different, with 0.20 pp for the AVF group and 0.17 pp for AVG group (p<0.001). We also sought to describe demographic differences in those requiring intervention during maturation (Supplemental Tables 1 – 2). Patient subgroup analyses had similar results to the overall cohort. However, there was notable geographic variation for both matured and failed AVF, with a trend towards greater interventional utilization in ESRD Networks in the Eastern Seaboard, whereas AVG intervention use during maturation was more similar across US networks.
Table 2A:
Maturation Patients and Procedures
| AVF | AVG | Pc | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Failed AVF Maturation | Successful AVF Maturation | Pa | AVF Total | Failed AVG Maturation | Successful AVG Maturation | Pb | AVG Total | ||
| Patients | 25,160 | 32,115 | n/a | 57,275 | 5,481 | 10,271 | n/a | 15,752 | n/a |
| Patients with Interventions | 4,271 | 9,718 | n/a | 13,989 | 1,206 | 1,698 | n/a | 2,904 | n/a |
| Interventional Procedures | 9,853 (0.39 pp) | 19,323 (0.6 pp) | <0.001 | 29,176 (0.51 pp) | 2,514 (0.46 pp) | 2,974 (0.29 pp) | <0.001 | 5,488 (0.35 pp) | <0.001 |
| Diagnostic Fistulogram Only | 3,305 (0.13 pp) | 5,247 (0.16 pp) | <0.001 | 8,552 (0.15 pp) | 478 (0.09 pp) | 610 (0.06 pp) | <0.001 | 1,088 (0.07 pp) | <0.001 |
| Any Therapeutic Intervention (Excludes isolated fistulogram) | 6,548 (0.26 pp) | 14,076 (0.44 pp) | <0.001 | 20,624 (0.36 pp) | 2,036 (0.37 pp) | 2,364 (0.23 pp) | <0.001 | 4,400 (0.28 pp) | <0.001 |
AVF failed maturation vs. AVF successful maturation
AVG failed maturation vs. AVG successful maturation
AVF total vs. AVG total
pp = per person
Of the subset of AVF that successfully matured, 9,718 of 32,115 (30.3%) required interventions during the maturation phase (“assisted maturation”), for 0.60 pp, with therapeutic interventions accounting for 0.44 pp (Table 2A). For the subset of AVG that were successfully matured, only 1,698 of 10,271 (16.5%) of AVG patients required intervention for maturation, with 0.29 pp for any intervention (p<0.001 vs AVF) and 0.23 pp for therapeutic interventions (p<0.001 vs AVF). For the subset of patients with failed AVG, intervention rates were higher than for successful AVG, with the opposite seen for the subsets of failed versus successful AVF. Rates of specific intervention types are summarized in Table 2B. Angioplasties occurred more frequently in AVF, while thrombectomies occurred more frequently in AVG, regardless of maturation status. Angioplasty and therapeutic intervention rates were higher in successful versus failed AVF, while the reverse relationship was true for the AVG group. In contrast, thrombectomies in particular were carried out much more often in those with either AVF or AVG maturation failure, presumably without long term success.
Table 2B:
Specific Maturation Procedures
| AVF | AVG | Pc | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Failed AVF Maturation | Successful AVF Maturation | Pa | AVF Total | Failed AVG Maturation | Successful AVG Maturation | Pb | AVG Total | ||
| Angioplasty | 3,377 (0.14 pp) | 9,284 (0.29 pp) | <0.001 | 12,661 (0.22 pp) | 502 (0.09 pp) | 743 (0.07 pp) | <0.001 | 1,245 (0.08 pp) | <0.001 |
| Thrombectomy | 1,340 (0.05 pp) | 830 (0.03 pp) | <0.001 | 2,170 (0.04 pp) | 1,207 (0.22 pp) | 1,318 (0.13 pp) | <0.001 | 2,525 (0.16 pp) | <0.001 |
| Thrombectomy only | 234 (0.01 pp) | 90 (0 pp) | <0.001 | 324 (0.01 pp) | 255 (0.05 pp) | 222 (0.02 pp) | <0.001 | 477 (0.03 pp) | <0.001 |
| Percutaneous Thrombectomy | 846 (0.03 pp) | 579 (0.02 pp) | <0.001 | 1,425 (0.02 pp) | 776 (0.14 pp) | 920 (0.09 pp) | <0.001 | 1,696 (0.11 pp) | <0.001 |
| Revision + Thrombectomy | 260 (0.01 pp) | 161 (0.01 pp) | <0.001 | 421 (0.01 pp) | 176 (0.03 pp) | 176 (0.02 pp) | <0.001 | 352 (0.02 pp) | <0.001 |
| Revision | 1,556 (0.06 pp) | 3,276 (0.1 pp) | <0.001 | 4,832 (0.08 pp) | 231 (0.04 pp) | 139 (0.01 pp) | <0.001 | 370 (0.02 pp) | <0.001 |
| Stent | 275 (0.01 pp) | 686 (0.02 pp) | <0.001 | 961 (0.02 pp) | 96 (0.02 pp) | 164 (0.02 pp) | 0.5 | 260 (0.02 pp) | 0.8 |
AVF failed maturation vs. AVF successful maturation
AVG failed maturation vs. AVG successful maturation
AVF total vs. AVG total
pp = per person
Maintenance Phase
Once mature, the profiles of interventions during the maintenance phase were much different than during maturation, inverting the procedural burden of the AVF and AVG groups. For AVF patients, 12,732 of 32,115 (39.6%) required intervention during the maintenance phase of the study period, with 1.29 ppy for any intervention and 0.93 ppy for therapeutic interventions (Table 3A). In contrast, 5,928 of 10,271 (57.7%) of AVG patients required intervention, with 2.18 ppy for any intervention during the study period (p<0.001 vs AVF) and 1.87 ppy for therapeutic interventions (p<0.001 vs AVF). For both types of vascular access, those AVF and AVG that required intervention during maturation continued to have a higher ongoing procedural burden compared to patients from the same access type. Requirements for a new catheter, catheter exchange, or catheter intervention were also different, with 0.11 ppy for the AVF group and 0.22 ppy for AVG group (p<0.001). Specific intervention types are detailed in Table 3B. Angioplasty rates were not different between the AVF and AVG groups. However, there were profound differences in thrombectomy rates, with lower rates for AVF in both subgroups (natural and assisted maturation). We sought to determine if there were demographic differences in those requiring post-maturation intervention (Supplemental Tables 3 – 4). AVF patients had higher repeated intervention rates in the ESRD Networks covering the Eastern Seaboard, although variation was common throughout the country. AVG intervention rates were somewhat more uniform geographically.
Table 3A:
Maintenance Patients and Procedures
| AVF | AVG | Pc | |||||||
|---|---|---|---|---|---|---|---|---|---|
| AVF Natural Maturation | AVF Assisted Maturation | Pa | AVF Total | AVG Natural Maturation | AVG Assisted Maturation | Pb | AVG Total | ||
| Patients | 22,397 | 9,718 | n/a | 32,115 | 8,573 | 1,698 | n/a | 10,271 | n/a |
| Patients with Interventions After Successful Maturation | 7,445 | 5,287 | n/a | 12,732 | 4,763 | 1,165 | n/a | 5,928 | n/a |
| Interventional Procedures | 19,656 (1.07 ppy) | 14,407 (1.82 ppy) | <0.001 | 34,063 (1.29 ppy) | 12,977 (2 ppy) | 3,750 (3.17 ppy) | <0.001 | 16,727 (2.18 ppy) | <0.001 |
| Diagnostic Fistulogram Only | 6,542 (0.36 ppy) | 3,122 (0.39 ppy) | <0.001 | 9,664 (0.37 ppy) | 1,896 (0.29 ppy) | 488 (0.41 ppy) | <0.001 | 2,384 (0.31 ppy) | <0.001 |
| Any Therapeutic Intervention (Excludes isolated fistulogram) | 13,114 (0.71 ppy) | 11,285 (1.42 ppy) | <0.001 | 24,399 (0.93 ppy) | 11,081 (1.71 ppy) | 3,262 (2.75 ppy) | <0.001 | 14,343 (1.87 ppy) | <0.001 |
AVF natural maturation vs. AVF assisted maturation
AVG natural maturation vs. AVG assisted maturation
AVF total vs. AVG total
ppy = per person year
Table 3B:
Maintenance Patients and Procedures
| AVF | AVG | Pc | |||||||
|---|---|---|---|---|---|---|---|---|---|
| AVF Natural Maturation | AVF Assisted Maturation | Pa | AVF Total | AVG Natural Maturation | AVG Assisted Maturation | Pb | AVG Total | ||
| Angioplasty | 9,368 (0.51 ppy) | 8,442 (1.07 ppy) | <0.001 | 17,810 (0.68 ppy) | 4,098 (0.63 ppy) | 1,190 (1 ppy) | <0.001 | 5,288 (0.69 ppy) | 0.3 |
| Thrombectomy | 1,863 (0.1 ppy) | 1,431 (0.18 ppy) | <0.001 | 3,294 (0.13 ppy) | 6,089 (0.94 ppy) | 1,826 (1.54 ppy) | <0.001 | 7,915 (1.03 ppy) | <0.001 |
| Thrombectomy only | 136 (0.01 ppy) | 106 (0.01 ppy) | <0.001 | 242 (0.01 ppy) | 612 (0.09 ppy) | 177 (0.15 ppy) | <0.001 | 789 (0.1 ppy) | <0.001 |
| Percutaneous Thrombectomy | 1,528 (0.08 ppy) | 1,241 (0.16 ppy) | <0.001 | 2,769 (0.11 ppy) | 5,140 (0.79 ppy) | 1,568 (1.32 ppy) | <0.001 | 6,708 (0.87 ppy) | <0.001 |
| Revision + Thrombectomy | 199 (0.01 ppy) | 84 (0.01 ppy) | 0.9 | 283 (0.01 ppy) | 337 (0.05 ppy) | 81 (0.07 ppy) | 0.03 | 418 (0.05 ppy) | <0.001 |
| Revision | 837 (0.05 ppy) | 400 (0.05 ppy) | 0.09 | 1,237 (0.05 ppy) | 218 (0.03 ppy) | 65 (0.05 ppy) | <0.001 | 283 (0.04 ppy) | <0.001 |
| Stent | 1,046 (0.06 ppy) | 1,012 (0.13 ppy) | <0.001 | 2,058 (0.08 ppy) | 676 (0.1 ppy) | 181 (0.15 ppy) | <0.001 | 857 (0.11 ppy) | <0.001 |
AVF natural maturation vs. AVF assisted maturation
AVG natural maturation vs. AVG assisted maturation
AVF total vs. AVG total
ppy = per person year
Intervention Location
The type of facility performing these interventions was analyzed. During maturation (Figure 2A), outpatient hospital-based intervention was more common. Office/independent clinics (i.e., free-standing interventional units) were the second most likely location. For maintenance phase (Figure 2B), the facility profile changed. For that phase, facilities designated as office/independent clinic were the clear majority, with the hospital-setting comprising a large minority of sites where maintenance procedures were performed, suggestive of differential intervention referral control. These relationships were similar for AVF and AVG.
Figure 2.
A. Place of service during vascular access maturation, graphed by percent of total procedures. B. Place of service during vascular access maintenance, graphed by percent of total procedures.
Intervention Rate Modeling
Rate ratios (RR) for the maintenance phase, estimated through proportional rates modeling, are depicted in Figure 3. For example, the RR for AVG (ref: fistula), estimated at 1.41 represents a 41% higher intervention rate for AVG relative to AVF (all other covariates being equal). There was a higher RR of intervention for AVG and for any dialysis access requiring intervention during maturation, as well as for AVF or AVG that required previous thrombectomy, stent, angioplasty, or revision. While the disadvantage of older age was maintained, gender differences were much less clinically profound. The advantage of the West Coast was still present during maintenance phase, with lower RR for intervention, while the Eastern Seaboard had a higher RR for intervention. As a sensitivity analysis, we separately analyzed AVF and AVG (Supplementary Figures 1–2), with similar results. Additionally, there was no meaningful change in outcome if the number of maturation interventions was added to the model, except for the intervention variable itself (RR 1.31; 95%CI 1.29, 1.32).
Figure 3.
Proportional rates model of maintenance interventions.
DISCUSSION
Interventions for both AVF and AVG were common during the maturation phase, with fewer interventions for AVG. Once mature, AVF had lower intervention requirements. Similar to Lee et al (12), we found that AVF and AVG requiring interventional assistance for maturation continued to require more interventions to maintain patency. In addition to the impact of procedural burden on both patient and policy, there were clinically significant differences in the long-term outcomes for AVF and AVG that can impact access choice. For context, Arhuidese et al (13) have shown 5-year secondary AVF patency rates at 5 years of 44.8% and 48.6% for diabetic and nondiabetic patients, respectively, which contrasted with secondary AVG patency rates of 34.1% and 36.8%, respectively. Coupled with a 4.2% prevalence of AVG infection necessitating excision, AVF were favored in their analysis. That said, the procedural burden of AVF or AVG should be interpreted in the patient-specific context of the durability and life expectancy of the access, expected access-specific complications, and likely future access needs.
We observed somewhat higher procedure rates for AVFs than for AVGs during the maturation phase for patients over 65 years old and for patients 45–64 years old. However, during maintenance phase, AVFs in all age groups required fewer interventions compared to AVGs. Similarly, Lee et al (14) found that older patients (≥67 years old) with an AVF required more interventions to make the AVF functional compared to similar patients with an AVG, with an associated longer catheter dependence time. Ko et al (15) found that incident patients >80 years of age still benefited from an AVF. Arhuidese et al (16) found that elderly patients with a life expectancy greater than 4 months benefited from an AVF. When targeted to patients with reasonable anatomy, AVFs can result in fewer overall procedures for patients when considering both the maturation and maintenance phases of AVF care regardless of age group. However, there are patients who may not be an appropriate candidate for one or the other type of access. For example, a patient at high risk of steal syndrome may not be a good AVG candidate even though veins for AVF may be suboptimal. Similarly, patients without adequate veins or other physiological issues that prevent successful AVF placement or maintenance may be AVG candidates.
From a cost perspective, there have been a number of studies about choice of AVF versus AVG that use different assumptions and are somewhat difficult to reconcile (17–21), as they usually have incomplete capture of comprehensive morbidity and cost. In our study, the need for new catheter placement, catheter exchange, or intervention exacerbated existing procedural burden. Central interventions required for catheter-caused central stenoses should be otherwise captured by the procedural hierarchy (Figure 1B). However, this additional catheter time is also coupled with additional time at risk for catheter-related sepsis. While this study did not examine that issue, such septic complications can significantly increase the cost and morbidity of vascular access that may not be offset by the AVFs’ lower long-term infection risk.
We found substantial US geographic variation during both phases of care, with higher rates of interventions on the East Coast. The geographic differences we found for AVF placement and maturation may correspond to known regional differences in population densities, access to care, and travel distance to providers based on geographic location (1–4, 22, 23), making it difficult to make inferences about quality of care. Despite this limitation, one plausible explanation for geographic variation in procedural burden is that surgeons may choose to place AVF at higher risk of failure if there is readily available access to quality interventions. Further research is needed to elucidate the most productive practice patterns that minimize unnecessary patient interventional burden.
Our results provide useful information for proper targeting of resource allocation as a better understanding of the effective use of different interventional approaches develop (24). Additionally, we found that interventions during the maturation phase were more commonly performed in a hospital setting, while maintenance interventions more likely occurred at free-standing interventional centers. Given that the surgeons placing the access are more likely to be centered around a hospital and are likely either doing the interventions themselves or driving the referrals for interventions during the maturation phase, this finding is not surprising. Similarly, dialysis units and nephrologists are more likely to manage the ordering of maintenance phase interventions, so it is similarly not surprising that the proportion of interventions performed outside of a hospital setting become the majority setting during the maintenance phase.
This study has limitations. There is a selection bias for which patients have AVF and AVG placed, with further confounding based on local surgeon and nephrology preference. As AVG use expands following the 2019 KDOQI vascular access guideline updates (7), AVG placed in a broader population may have better outcomes, at least in the context of the lifetime of that particular access. Extrapolation of historical crude interventional rates to this expanded population may result in erroneous inferences. Due to database limitations, we limited analysis to incident dialysis patients with first access placed during the first year. There may be differences in broader cohorts, particularly for those with AVF or AVG placed pre-dialysis. AVF and AVG inherently have different success rates for, and times to, maturation—which may bias the maintenance phase results and may also have significant impact catheter complication rates. Furthermore, we made a dichotomy for maturation versus maintenance phase based on monthly CROWNWeb determination of cannulation using a functional definition due to data source limitations. The selection bias, when combined with the analytic approach, essentially results in an “as treated,” rather than an “intent to treat,” analytic approach—making this study more useful for resource requirement predictions, rather than providing guidance on choice of vascular access placement for an individual patient at the time of operation. While data entry could also be prone to error, we have previously found good correlation between the CMS Medical Evidence Form 2728 and CROWNWeb in incident hemodialysis patients (25). CROWNWeb provides monthly data, so it is conceivable that the lag between use and data entry, or sporadic short-term access use or nonuse, could introduce bias. However, month-to-month consistency of these data are reasonably good (3). Additionally, the sample size is large enough that it likely affects both groups equally. We only followed the patients for 1 year following placement. We picked this timeframe to utilize the most recent data available while at the same time allowing for similar follow-up for the entire cohort. It is certainly possible, and possibly even likely given other studies with earlier cohorts that allowed longer follow-up (6, 12), that maintenance phase interventions become more frequent with time.
CPT code-based billing errors are not uncommon for other procedural specialties (26), making it reasonable to suspect that coding errors are similarly present in interventions on dialysis access. Some of the CPT codes are specific to dialysis access, while some are not. In most cases, we’ve coupled such codes to the 36147 code, which is specific for the fistulogram performed on dialysis access with the majority of endovascular interventions for the study period. It is certainly possible that there were interventions performed under alternative guidance, such as ultrasound. Also, we chose to approach the interventions on any given day or encounter as a single intervention, with a hierarchy determining the dominate procedure (Figure 1B). Since some CPT codes can indicate both small and large interventions, this hierarchy may falsely classify some procedures as dominate. However, multiple interventions are often done in a single setting, so this seemed the best approach to address database limitations.
In summary, we found that during maturation, interventions for both AVF and AVG were common. Once successfully matured, AVF had lower maintenance interventional requirements. It is worth noting that interventions in and of themselves are not necessarily a problem, as long as they are logistically and budgetarily accounted for--and patient and system expectations match. If long-term permanent access is expediently established and maintained, such interventions can be very much worthwhile. Finally, geographic variations were observed in both phases, which suggests that future studies geared to understanding this variation could help determine the most efficient approach for establishing dialysis access, while also addressing disparities in access to care. This study supplies a more detailed understanding of the overall procedural burden for AVF and AVG in the United States.
Supplementary Material
Acknowledgements:
April F. Wyncott, MPH, MBA, and Vivian H. Kurtz, MPH, provided regulatory compliance and administrative support for this project.
Support:
This study was performed under the USRDS Coordinating Center contract with the NIH NIDDK and the work was conducted by the Coordinating Center. At the time of this project, the USRDS Coordinating Center was located at the University of Michigan Kidney Epidemiology and Cost Center, in partnership with Arbor Research Collaborative for Health, Ann Arbor, MI. The USRDS director was Rajiv Saran, MBBS, MD, MRCP, MS, Professor of Medicine and Epidemiology at the University of Michigan, and the co-deputy directors were Vahakn B. Shahinian, MD, MS, Associate Professor of Medicine at the University of Michigan, and Bruce M. Robinson, MD. The NIDDK project officers were Kevin C. Abbott, MD, MPH, and Lawrence Y.C. Agodoa, MD. This study was also partially funded in part by R01-DK070869 from the NIH NIDDK. The funder had no role in the study design, data collection, data analysis, data interpretation, or writing of the report.
Financial Disclosure:
Kenneth J. Woodside has received research support from Laminate Medical Technologies and Astellas Pharma US, is a consultant to Laminate Medical Technologies, and is on the medical advisory board of Nephrodite. Kaitlyn J. Repeck, Purna Mukhopadhyay, and Ronald L. Pisoni are employees of Arbor Research Collaborative for Health, which administers the Dialysis Outcomes and Practice Patterns Study (DOPPS). For details, see https://www.dopps.org/AboutUs/Support.aspx. Rajiv Saran serves as a consultant for Kyowa-Kirin Co., has been on advisory boards for Reata Pharmaceuticals and AstraZeneca, and has received travel support and honoraria from Chugai Pharmaceuticals. The remaining authors declare that they have no relevant financial interests.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Disclaimer: The interpretation and reporting of these data are the responsibility of the authors and in no way should be seen as an official policy or interpretation of the U.S. government.
REFERENCES
- 1.Saran R, Robinson B, Abbott KC, Agodoa LYC, Bragg-Gresham J, Balkrishnan R, et al. US Renal Data System 2018 Annual Data Report: Epidemiology of Kidney Disease in the United States. Am J Kidney Dis. 2019;73(3 Suppl 1):S1–S772. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Pisoni RL, Zepel L, Port FK, Robinson BM. Trends in US Vascular Access Use, Patient Preferences, and Related Practices: An Update From the US DOPPS Practice Monitor With International Comparisons. Am J Kidney Dis. 2015;65(6):905–15. [DOI] [PubMed] [Google Scholar]
- 3.Woodside KJ, Bell S, Mukhopadhyay P, Repeck KJ, Robinson IT, Eckard AR, et al. Arteriovenous Fistula Maturation in Prevalent Hemodialysis Patients in the United States: A National Study. Am J Kidney Dis. 2018;71(6):793–801. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Shahinian VB, Zhang X, Tilea AM, He K, Schaubel DE, Wu W, et al. Surgeon Characteristics and Dialysis Vascular Access Outcomes in the United States: A Retrospective Cohort Study. Am J Kidney Dis. 2020;75(2):158–66. [DOI] [PubMed] [Google Scholar]
- 5.Beathard GA, Urbanes A, Litchfield T. Changes in the Profile of Endovascular Procedures Performed in Freestanding Dialysis Access Centers over 15 Years. Clin J Am Soc Nephrol. 2017;12(5):779–86. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Thamer M, Lee TC, Wasse H, Glickman MH, Qian J, Gottlieb D, et al. Medicare Costs Associated With Arteriovenous Fistulas Among US Hemodialysis Patients. Am J Kidney Dis. 2018;72(1):10–8. [DOI] [PubMed] [Google Scholar]
- 7.Lok CE, Huber TS, Lee T, Shenoy S, Yevzlin AS, Abreo K, et al. KDOQI Clinical Practice Guideline for Vascular Access: 2019 Update. Am J Kidney Dis. 2020;75(4 Suppl 2):S1–S164. [DOI] [PubMed] [Google Scholar]
- 8.Arhuidese IJ, Orandi BJ, Nejim B, Malas M. Utilization, patency, and complications associated with vascular access for hemodialysis in the United States. J Vasc Surg. 2018;68(4):1166–74. [DOI] [PubMed] [Google Scholar]
- 9.Harms JC, Rangarajan S, Young CJ, Barker-Finkel J, Allon M. Outcomes of arteriovenous fistulas and grafts with or without intervention before successful use. J Vasc Surg. 2016;64(1):155–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Domenick Sridharan N, Fish L, Yu L, Weisbord S, Jhamb M, Makaroun MS, et al. The associations of hemodialysis access type and access satisfaction with health-related quality of life. J Vasc Surg. 2018;67(1):229–35. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Lin DY, Wei LJ, Yang I, Ying Z. Semiparametric regression for the mean and ratefunctions of recurrent events. J R Stat Soc Series B Stat Methodol. 2000;62(4):711–30. [Google Scholar]
- 12.Lee T, Qian JZ, Zhang Y, Thamer M, Allon M. Long-Term Outcomes of Arteriovenous Fistulas with Unassisted versus Assisted Maturation: A Retrospective National Hemodialysis Cohort Study. J Am Soc Nephrol. 2019;30(11):2209–18. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Arhuidese IJ, Purohit A, Elemuo C, Parkerson GR, Shames ML, Malas MB. Outcomes of Autogenous Fistulas and Prosthetic Grafts for Hemodialysis Access in Diabetic and Non-Diabetic Patients. J Vasc Surg. 2020;72(6):2088–96. [DOI] [PubMed] [Google Scholar]
- 14.Lee T, Qian J, Thamer M, Allon M. Tradeoffs in Vascular Access Selection in Elderly Patients Initiating Hemodialysis With a Catheter. Am J Kidney Dis. 2018;72(4):509–18. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Ko GJ, Rhee CM, Obi Y, Chang TI, Soohoo M, Kim TW, et al. Vascular access placement and mortality in elderly incident hemodialysis patients. Nephrol Dial Transplant. 2020;35(3):503–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Arhuidese IJ, Cooper MA, Rizwan M, Nejim B, Malas MB. Vascular access for hemodialysis in the elderly. J Vasc Surg. 2019;69(2):517–25 e1. [DOI] [PubMed] [Google Scholar]
- 17.Brooke BS, Griffin CL, Kraiss LW, Kim J, Nelson R. Cost-effectiveness of repeated interventions on failing arteriovenous fistulas. J Vasc Surg. 2019;70(5):1620–8. [DOI] [PubMed] [Google Scholar]
- 18.Hall RK, Myers ER, Rosas SE, O’Hare AM, Colon-Emeric CS. Choice of Hemodialysis Access in Older Adults: A Cost-Effectiveness Analysis. Clin J Am Soc Nephrol. 2017;12(6):947–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Al-Balas A, Lee T, Young CJ, Kepes JA, Barker-Finkel J, Allon M. The Clinical and Economic Effect of Vascular Access Selection in Patients Initiating Hemodialysis with a Catheter. J Am Soc Nephrol. 2017;28(12):3679–87. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Chang JT, Li IH, Shih JH, Chien WC, Pan KT, Wu ST, et al. Health Care Service Utilization After Various Vascular Access Selections. J Surg Res. 2019;244:166–73. [DOI] [PubMed] [Google Scholar]
- 21.Al-Jaishi AA, Liu AR, Lok CE, Zhang JC, Moist LM. Complications of the Arteriovenous Fistula: A Systematic Review. J Am Soc Nephrol. 2017;28(6):1839–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.O’Hare AM, Johansen KL, Rodriguez RA. Dialysis and kidney transplantation among patients living in rural areas of the United States. Kidney Int. 2006;69(2):343–9. [DOI] [PubMed] [Google Scholar]
- 23.Moist LM, Bragg-Gresham JL, Pisoni RL, Saran R, Akiba T, Jacobson SH, et al. Travel time to dialysis as a predictor of health-related quality of life, adherence, and mortality: the Dialysis Outcomes and Practice Patterns Study (DOPPS). Am J Kidney Dis. 2008;51(4):641–50. [DOI] [PubMed] [Google Scholar]
- 24.Swaminathan S, Sommers BD, Thorsness R, Mehrotra R, Lee Y, Trivedi AN. Association of Medicaid Expansion With 1-Year Mortality Among Patients With End-Stage Renal Disease. JAMA. 2018;320(21):2242–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Mukhopadhyay P, Pearson J, Woodside KJ, Bell S, Pisoni RL, Schaubel DE, et al. Vascular access at dialysis initiation in the United States Renal Data System (USRDS): strong agreement between CMS 2728 medical evidence form and CROWNWeb. J Am Soc Nephrol. 2015;26 Suppl:286A. [Google Scholar]
- 26.Duszak R, Blackham WC, Kusiak GM, Majchrzak J. CPT coding by interventional radiologists: a multi-institutional evaluation of accuracy and its economic implications. J Am Coll Radiol. 2004;1(10):734–40. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.