Background: Vascular access surveillance by ultrasound dilution blood flow rate (Qa) measurement is widely recommended; however, optimal criteria for detecting stenosis and predicting thrombosis in arteriovenous fistulae (AVFs) are still not clearly defined.
Methods: In a blinded trial, we evaluated the accuracy of single Qa measurement, Qa adjusted for mean arterial pressure (Qa/MAP), and decrease in Qa over time (dQa) in detecting stenosis and predicting thrombosis in an unselected population of 120 hemodialysis subjects with native forearm AVFs (91 AVFs, located at the wrist; 29 AVFs, located at the midforearm). All AVFs underwent fistulography, which identified greater than 50% stenosis in 54 cases.
Results: Receiver operating characteristic curve analysis showed that dQa, Qa, and Qa/MAP have a high stenosis discriminative ability with similar areas under the curve (AUCs), ie, 0.961 +/- 0.025, 0.946 +/- 0.021, and 0.912 +/- 0.032, respectively. In the population as a whole, optimal thresholds for stenosis were Qa less than 750 mL/min alone and in combination with dQa greater than 25% (efficiency, 90%); however, the best threshold depended on anastomotic site; it was Qa less than 750 mL/min for an AVF at the wrist and Qa less than 1,000 mL/min for an AVF in the midforearm. Qa was the best predictor of incipient thrombosis (AUC, 0.981 +/- 0.013) with an optimal threshold at less than 300 mL/min (efficiency, 94%). Pooled intra-assay and interassay variation coefficients were 8.2% for MAP, 7.9% for Qa, and 11.2% for Qa/MAP.
Conclusion: Our study shows that ultrasound dilution Qa measurement is a reproducible and highly accurate tool for detecting stenosis and predicting thrombosis in forearm AVFs. Neither Qa/MAP nor dQa improve the diagnostic performance of Qa alone, although its combination with dQa increases the test's sensitivity for stenosis.