Thermodilution is unreliable in veno-venous extracorporeal membrane oxygenation (VV-ECMO). Systemic oxygenation depends on recirculation fractions and ratios of ECMO flow to cardiac output. In a prospective in vitro simulation, we assessed the diagnostic accuracy of a modified thermodilution technique for recirculation and cardiac output. We hypothesized that this method provided clinically acceptable precision and accuracy for cardiac output and recirculation.


Two ECMO circuits ran in parallel, one representing a VV-ECMO, the second representing native heart, lung and circulation. Both circuits shared the right atrium. Extra limbs for recirculation and pulmonary shunt were added. We simulated ECMO flows from 1 to 2.5 l/min and cardiac outputs from 2.5 to 3.5 l/min with recirculation fractions (0 – 80%) and pulmonary shunts. Thermistors in both ECMO-limbs and the pulmonary artery measured the temperature changes induced by cold bolus injections into the arterial ECMO-limb. Recirculation fractions were calculated from the ratio of the areas under the temperature curve (AUC) in the ECMO-limbs and from partitioning of the bolus volume (flow based). With known partitioning of bolus volumes between ECMO and pulmonary artery, cardiac output was calculated. High precision ultrasonic flow probes served as reference for Bland-Altman plots and linear mixed-effect models.


Accuracy and precision for both the recirculation fraction based on AUC (bias -5.4 %; limits of agreement (LoA) -18.6 to 7.9 %) and flow based (bias -5.9 %; LoA -18.8 to 7.0 %) are clinically acceptable. Calculated cardiac output for all recirculation fractions was accurate, but imprecise (RecirculationAUC: Bias 0.56 L/min; LoA -2.27 to 3.4 L/min; RecirculationFLOW: Bias 0.48 L/min; LoA -2.22 to 3.19 L/min). Recirculation fraction increased bias and decreased precision.


Adapted thermodilution for VV-ECMO allows simultaneous measurement of recirculation fraction and cardiac output and may help optimize patient management with severe respiratory failure.