Prospective Analysis of Arteriovenous Fistula Performance in the Context of Competing Risks

Abstract

Key Points:

1. Among 257 newly created arteriovenous fistulas, primary nonfunction occurred in 49%, and only 55% were ultimately used for dialysis.

2. Loss of arteriovenous fistula patency was lower when competing risks were accounted for compared with conventional Kaplan–Meier analysis.

3. We present icon-array plots that summarize our data and may be used a decision aid for patients in the future.

Background: Many patients with newly created arteriovenous fistulas (AVFs) may die before the AVF is needed for hemodialysis. However, formal competing risks (CRs) frameworks are rarely used to report AVF patency, which may lead to biased estimates. We sought to identify the proportion of newly created AVF experiencing primary nonfunction and describe long-term patency using a CR framework.

Methods: We conducted a prospective observational study in 257 adults with newly created AVF in Alberta, Canada. The primary outcome was primary nonfunction. Secondary outcomes included loss of primary patency, loss of assisted primary patency, and loss of secondary functional patency. Results were presented using icon-array plots to form the basis for future decision aids.

Results: Participants were 63.0% male, with mean age 62.3 years and median follow-up 18.5 months (range, 0.02–180 months). Of 257 participants, 50 could not be assessed for function or primary nonfunction, usually because of death. Of the remaining 207, 102 (49.3%) had primary nonfunction, and function was ultimately established for 142 (68.6%). Thus, only 142 of the 257 participants (55.3%) ultimately used the AVF for hemodialysis. High rates of CRs led to biased results from Kaplan–Meier analyses of lost patency. When accounting for CRs, loss of primary patency among AVFs with established function was 36.6%, 65.5%, and 66.2%, at 1, 3, and 5 years, respectively.

Conclusions: Only 55% of fistulas were ultimately used for hemodialysis when accounting for CRs and primary nonfunction. These results and the icon-array plots may inform discussions surrounding vascular access options for patients.

Graphical abstract

Figure 1

Icon-array plots showing clinically relevant outcomes. (A) Initial outcomes for all participants (n=257). (B) Subsequent outcomes for those with established AVF function (n=142). (A) AVF never required for hemodialysis because of participants switching or choosing another modality to initiate KRT (i.e., kidney transplant, chronic PD, or conservative care). (B) Timing is based on time elapsed since the date of established function. *This category includes fistulas that are functional at 1 year and those that were censored before 1 year. Censoring reasons include death, modality switch (kidney transplant, chronic PD, recovery of function), or moving away. AVF arteriovenous fistula; PD, peritoneal dialysis.

Figure 2

Icon-array plots showing proportion of newly created AVFs that were usable within a certain time after access creation. Analysis included all AVFs where function could be assessed (n=207).

Figure 3

Plots showing cumulative incidence of outcomes of interest on the basis of both KM and CR analyses. (A) Loss of primary patency (B) Loss of assisted primary patency (C) Loss of secondary patency. N=142: including only AVF with established function. N=257: analyses including all newly created AVFs. CIF, cumulative incidence function; CR, competing risk; KM, Kaplan–Meier.