Accuracy of dilution techniques for access flow measurement during hemodialysis

Access flow is now widely measured by creating artificial recirculation with the dialysis lines reversed and using dilution methods that sense either ultrasound velocity, electrical impedance, optical, or thermal changes. This study identifies and quantifies factors that influence the accuracy of access flow measurements and recommends ways to reduce these errors. Two major sources of access flow measurement error are identified, arising firstly from the second pass of the indicator by recirculation through the cardiopulmonary system (cardiopulmonary recirculation, CPR), and secondly from changes in venous line blood flow (Qb) and vascular access flow induced by the pressure of venous bolus injections. These errors are considered from theory, by direct measurement of access flow in a sheep model, and by analysis of clinical data. Two extremes for the venous introduction of indicator can be considered in access flow measurements, a slow infusion, which perturbs neither the venous line flow nor access flow but increases the error attributable to the second pass of the indicator by recirculation through cardiopulmonary system, or rapid injection, which eases separation of the second pass of the indicator signal but generates changes in the venous flow and access flow. If CPR is not eliminated, the area added to that of the first pass of indicator ranges up to 40%. Good time resolution could permit the separation of the areas generated by the first and second passage of the indicator. In sheep experiments, injections of 5 or 10 mL into a venous port close to the vascular access caused Qb to change by 20% to 40%. Both the animal experiments and analysis of raw data collected during routine clinical dialysis showed that moving the injection site sufficiently far from the patient, before or into the venous bubble trap, reduced the increase in Qb to only approximately 5% during the critical time when the concentration curve is changing for most tubing brands (Baxter, Belco, Gambro, Hospal, Medisystem, and National Medical Care). Because of the smaller volume of the venous bubble chamber in Cobe tubing (Cobe, Centrysystem 3), this brand showed approximately a 20% increase in Qb. Moving the site of bolus injections to before the bubble trap in the sheep experiments also eliminated the influence of changes in access flow. An additional error in access flow measurement of 20% or more arises from the use of flow reading taken from pump setting rather than a measured flow. The discrepancy between the real flow and pump setting is attributable to needle size, vascular access conditions, or pump calibration. The results show that problems can be minimized by using a dual sensor system that retains the precise timing necessary for separation of access recirculation from CPR; by accurate measurement of dialyzer blood flow; by moving the site of injection to before the venous bubble trap, sufficiently far from the patient, and correcting for any remaining deviations in flow in the venous line concurrent with the dilution curve.