Agreement of surgeon’s choice of autologous arteriovenous fistula with a vascular access selection app and correlation with 1-year outcomes

Abstract

Objective:

The purpose of this study was to compare surgeons’ choice of autologous arteriovenous fistula (AVF) to that recommended by a vascular access selection app and investigate impact on outcomes.

Design:

Single center retrospective study.

Methods:

The “My Vascular Access” app was used to evaluate the appropriateness of autologous accesses created in our institution between January and August 2021. Patients were then stratified into two groups (Group R = Recommended; Group NR = Non-recommended), based on their appropriateness ranking, and 1-year outcomes compared. The primary endpoint was access maturation.

Results:

The cohort consisted of 113 subjects with a mean age of 64 ± 15 years, of which 13% were in pre-dialysis. Preoperative vein mapping revealed a forearm cephalic vein diameter of 2.4 ± 1.3 mm, basilic vein of 4.2 ± 1.7 mm, and radial artery measured 2.4 ± 0.8 mm. Radiocephalic fistulas were created in 53 cases (47%). The app rated 61% of AVFs (n = 69) as appropriate, 22% inappropriate, and 10% intermediate, but eight accesses—all brachiobasilic—were not ranked by the app. The cumulative access maturation rate for the entire cohort was 65% ± 5%, 76% ± 4%, and 78% ± 4% at 3 months, 6 months, and 1-year respectively, without significant difference between groups (i.e. 1-year maturation rate of 80% vs 78%; p = 0.55). Recommended accesses were less likely to receive reinterventions aimed at access maturation (20% vs 39%, p = 0.041) and less likely to encounter access thrombosis during follow-up (7% vs 22%; p = 0.027). In multivariate analysis, female sex was the only variable independently associated with unassisted maturation failure.

Conclusion:

Only 60% of autologous AVFs created by vascular surgeons aligns with recommendations from the “My Vascular Access” app. Recommended AVFs are associated with lower risk of access thrombosis and maturation-assist interventions, but no significant difference in access functionality at 1-year in our cohort.