Evaluation of a new arterial pressure-based cardiac output device requiring no external calibration

BACKGROUND: Several techniques have been discussed as alternatives to the intermittent bolus thermodilution cardiac output (CO(PAC)) measurement by the pulmonary artery catheter (PAC). However, these techniques usually require a central venous line, an additional catheter, or a special calibration procedure. A new arterial pressure-based cardiac output (CO(AP)) device (FloTrac™, Vigileo™; Edwards Lifesciences, Irvine, CA, USA) only requires access to the radial or femoral artery using a standard arterial catheter and does not need an external calibration. We validated this technique in critically ill patients in the intensive care unit (ICU) using CO(PAC )as the method of reference. METHODS: We studied 20 critically ill patients, aged 16 to 74 years (mean, 55.5 ± 18.8 years), who required both arterial and pulmonary artery pressure monitoring. CO(PAC )measurements were performed at least every 4 hours and calculated as the average of 3 measurements, while CO(AP )values were taken immediately at the end of bolus determinations. Accuracy of measurements was assessed by calculating the bias and limits of agreement using the method described by Bland and Altman. RESULTS: A total of 164 coupled measurements were obtained. Absolute values of CO(PAC )ranged from 2.80 to 10.80 l/min (mean 5.93 ± 1.55 l/min). The bias and limits of agreement between CO(PAC )and CO(AP )for unequal numbers of replicates was 0.02 ± 2.92 l/min. The percentage error between CO(PAC )and CO(AP )was 49.3%. The bias between percentage changes in CO(PAC )(ΔCO(PAC)) and percentage changes in CO(AP )(ΔCO(AP)) for consecutive measurements was -0.70% ± 32.28%. CO(PAC )and CO(AP )showed a Pearson correlation coefficient of 0.58 (p < 0.01), while the correlation coefficient between ΔCO(PAC )and ΔCO(AP )was 0.46 (p < 0.01). CONCLUSION: Although the CO(AP )algorithm shows a minimal bias with CO(PAC )over a wide range of values in an inhomogeneous group of critically ill patients, the scattering of the data remains relative wide. Therefore, the used algorithm (V 1.03) failed to demonstrate an acceptable accuracy in comparison to the clinical standard of cardiac output determination.