#5089 EVALUATION OF REMOTE MONITORING OF HAEMODIALYSIS VASCULAR ACCESS FOR PREDICTION OF STENOSIS AND THROMBOTIC VASCULAR EVENTS

Abstract

Abstract Background and Aims Early identification of dysfunctional arteriovenous haemodialysis (HD) vascular access (VA) is important for timely referral and intervention. We evaluated the accuracy of remote monitoring technology of VA that uses access flow data routinely collected during HD treatment to predict stenotic/thrombotic vascular events. Method We retrospectively calculated access risk score in a blinded fashion using Vasc-Alert vascular access surveillance technology from all HD treatments sessions in 2 satellite HD units for 12 months. We included in the analysis HD patients dialysing with arteriovenous fistula or graft with available Vascalert data for ≥ 2 months. The Access Risk Score was calculated as average of the scores for every 3 consecutive HD treatments and a high-risk score (HRS) was defined as ≥ 7.1 Using the electronic patient records, we identified patients with significant vascular access events (thrombosis, angiographic stenosis requiring angioplasty or doppler with > 50% stenosis) and without vascular access event. Information for clinically detected malfunctioning fistula was retrieved from the last clinic letter and the last vascular access multidisciplinary meeting notes prior to the vascular event. For the event positive patients, we included in the analysis the Vasc-alert data 2 months prior to the event. For the negative group, we included Vasc-alert data for 5 consecutive months with 1 month follow up. For the analysis we considered HRS positive if ≥2 HRS were generated. Results Out of 141 patients with available Vasc-alert data there were 60 patients dialyzing via a tunneled line. Amongst 81 patients with arteriovenous fistula or graft, 58 had available Vasc-alert data for ≥ 2 months. Out of 12 event positive patients (4 patients with thrombosed access, 6 patients with stenosis requiring angioplasty and 2 patients with >50% on doppler referred and awaiting fistulogram),10 (83%) had ≥2 HRS generated 2 months prior to the vascular event (Median 8, IQR 6.75-8). Out of the 46 patients without vascular events, 15 patients (32.6%) had HRS ≥ 2 and 4 patients had only one HRS score. Patient characteristics by vascular event are presented in Table 1. The sensitivity and specificity of HRS ≥2 for detecting future vascular events were 83.3% and 67.4%, respectively. The positive and negative predictive value of HRS ≥2 was 40% and 93.9% respectively. History of prior access stenosis and clinically detected malfunctioning fistula were significantly associated with vascular access events (P value 0.002, < 0.001 respectively), and HRS≥2 discrete values (P value 0.007 and 0.005 respectively). Within the patients with thrombosed access, 2 patients (50%) detected by HRS were not detected with clinical monitoring. Conclusion Our results suggest that vascular access risk score can be a useful screening tool to assist clinical decision making for VA risk stratification. Prospective studies are required to evaluate its utility in the VA surveillance pathway.