How accurate are different arterial pressure-derived estimates of cardiac output and stroke volume variation measures in critically ill patients?

Introduction

We compared the cross-correlation between estimates of cardiac output (CO) and left ventricular stroke volume variation (SVV) amongst three commercially available, minimally invasive devices. (LiDCOplus, FloTrac and PiCCO).

Methods

We simultaneously compared continuous and bolus thermodilution CO measures from a pulmonary artery catheter (PAC) with simultaneous estimates of arterial pulse contour-derived CO using the FloTrac^®, LiDCOplus^®and PiCCO^® measured at one time in 20 cardiac surgery patients during the first two postoperative hours. We also compared SVV estimates among the three devices. Mean and absolute values for CO and SVV across all devices were compared by ANOVA and Bland–Altman analysis.

Results

Mean CO values were not different across devices (5.8 ± 1.6 l/min vs 5.9 ± 1.7 l/min vs 5.8 ± 1.6 l/min for PiCCO, LiDCO plus and FloTrac, respectively; P = 0.4). The mean PAC CO (5.8 ± 1.6 l/min) was similar to PiCCO and FloTrac estimated CO values, but less than LiDCO CO values (P < 0.01). Biases between PAC and PiCCO, LiDCO and FloTrac values were 0.19 ± 0.57 l/min, -0.35 ± 0.56 l/min and -0.30 ± 1.56 l/min, respectively, and precision was -1.31 to 0.92 l/min, -1.46 to 0.77 l/min and -2.6 to 2.0 l/min, respectively. LiDCO and FloTrac SVV correlated (r² = 0.58), however, with a bias of -0.40 ± 6.50% and a precision of -13 to 7%; whereas FloTrac and PiCCO SVV were not correlated (r² = ns), with a bias of 4.0 ± 6.0% and a precision of -8 to 16%. LiDCO and PiCCO SVV were also not correlated (r² = ns), with a bias of -5.4 ± 9.0% and a precision of -22 to 17%. Finally, PiCCO and LIDCO pulse pressure variation were correlated (r² = 0.64, P < 0.05), with a bias of 17.0 ± 6.5% and a precision of -10 to 15%.

Conclusion

All three arterial pulse contour analysis devices estimated CO well with a high degree of accuracy and precision. Furthermore, of the two devices that also report pulse pressure variation, both gave similar estimates, whereas SVV estimates correlated well only between LiDCO and FloTrac. The results of prior studies using LiDCO and PiCCO-derived estimates of SVV cannot therefore be compared with each other, nor can absolute values be used to drive similar resuscitation protocols unless independently validated for that catheter.