Flow measurements in dialysis shunts: lack of agreement between conventional Doppler, CVI‐Q, and ultrasound dilution

• Published in: Nephrology, dialysis, transplantation dialysis, transplantation, 2001-02, Vol.16 (2), p.395-399
• Main Authors: Zanen, Adriaan L., Toonder, Irwin M., Korten, Eunice, Wittens, Cees H. A., Diderich, Philip P. N. M.
• Format: Article
• Language: English
• Subjects: Adult; Anesthesia. Intensive care medicine. Transfusions. Cell therapy and gene therapy; arteriovenous shunt; Biological and medical sciences; Blood Flow Velocity; Cardiovascular system; Catheters, Indwelling; CVI‐Q; Emergency and intensive care: renal failure. Dialysis management; haemodialysis; Humans; Intensive care medicine; Investigative techniques, diagnostic techniques (general aspects); Medical sciences; Renal Dialysis; Ultrasonic investigative techniques; ultrasonography; Ultrasonography - methods; Ultrasonography, Doppler; ultrasound dilution
• ISSN: 0931-0509; 1460-2385
• DOI: 10.1093/ndt/16.2.395

Background. Measuring flow in dialysis shunts is recommended to predict imminent thrombosis. Multiple methods for measuring blood flow are in use. Numerous ultrasound protocols exist which determine volume flow using a conventional Doppler (CD) frequency shift analysis technique. All of these are subject to potentially large errors. Quantitative colour velocity index (CVI‐Q) does not make use of the Doppler equation and is more precise in vitro. Ultrasound dilution (UD) measures access flow during dialysis in a non‐operator‐dependent way. The aim of the present study was to compare these three methods of measuring access flow in vivo for agreement with each other. Methods. In 38 accesses flow was measured by CD, CVI‐Q, and UD. All measurements were done during dialysis. Agreement was determined by intraclass correlation coefficient (ICC=Ri) and Bland–Altman analysis. Results. ICC between UD and CVI‐Q was Ri=0.56. ICC between UD and CD was Ri=0.10, and ICC between CD and CVI‐Q was Ri=0.16. Bland–Altman analysis revealed a bias (mean difference) of −38 ml/min between UD and CVI‐Q, a bias of 1129 ml/min between UD and CD, and a bias of 1167 ml/min between CVI‐Q and CD. Conclusions. CD measurements did not agree with UD or CVI‐Q: much higher values were recorded with the former than with the latter two techniques. The agreement between UD and CVI‐Q measurements is low but reasonable. Caution must be applied in comparing and interpreting values of access flow measured by different techniques.