Accuracy and reproducibility of urea recirculation in detecting haemodialysis access stenosis

Background. There is wide disagreement among studies that have evaluated the accuracy of urea recirculation (UR) in detecting vascular access stenosis. The 3-site method (UR3) has been discredited and replaced by the 2-site method (UR2), but few studies have evaluated UR2. Methods. We compared the accuracies of UR2 and UR3 in detecting stenosis in 59 haemodialysis patients during a 12-month period. All patients were studied without regard to clinical suspicion of stenosis. Stenosis (≤ 50% luminal narrowing) was diagnosed by duplex ultrasound and confirmed by angiography. The reproducibility of UR2 was determined by computing its total standard deviation (SD(TOT)) from measurements during three dialysis sessions over a 15-day period. Results. Stenosis was found in 32% of 124 access studies (mean luminal narrowing = 58%, range = 50%-83%). The mean UR values of stenotic accesses were only slightly higher than non-stenotic accesses for both UR2 (5.6% vs 2.9%, P < 0.01) and UR3 (13.1% vs 11.2%, P = 0.22). An increase in blood pump speed from 300 to 425 ml/min did not improve detection of stenosis by UR2. There were no UR thresholds that could adequately separate the presence of stenosis from its absence. The SD(TOT) of UR2 was 3.%, indicating that a patient's UR2 measurement may vary over a range of 16% (± 2SD(TOT) = ± 8%). Conclusion. Stenosis of the haemodialysis access does not predictably cause recirculation, and the reproducibility of the UR2 measurement is poor.