Comparison between Fourth-Generation FloTrac/Vigileo System and Continuous Thermodilution Technique for Cardiac Output Estimation after Time Adjustment during Off-Pump Coronary Artery Bypass Graft Surgery: A Retrospective Cohort Study

Figure 1. Patient flow diagram. SQI: signal quality index, PAC: pulmonary artery catheter.

Figure 2. R-squared (R2) values between arterial pressure waveform-based cardiac output estimation (CO-FloTrac) and varying degrees of leading values (values of antegrade time points; 1 to 15 min) of continuous thermodilution cardiac output (COcont). The optimal time adjustment was defined as the delay when R2 was highest within an individual dataset. In cases with persistently low R2 values (<0.1) with any of the adjustments, 6 min of adjustment was applied since no reasonable adjustment could be made based on such low correlation. Each color represents each individual data. The red solid horizontal line indicates R2 = 0.1. Five cases with persistently low R2 values are indicated as dashed lines.

Figure 3. Repeated measures correlation (RM correlation) and root mean squared error (RMSE) between arterial pressure waveform-based cardiac output estimation (CO-FloTrac) and continuous thermodilution cardiac output (COcont) before and after time adjustment. Each color represents each individual dataset.

Figure 4. Continuous thermodilution cardiac output (COcont) and arterial pressure waveform-based cardiac output estimation (CO-FloTrac) of a sample case before (upper panel) and after the time adjustment (lower panel). The plots on the left side are time-series plots and the plots on the right side are scatter plots between the two variables. Note the slightly misaligned peaks and troughs due to the time delays in the COcont values (left upper). The plot on lower left side shows a more optimized alignment after the time adjustment (in this case, 6 min). R-squared (R2) value increased from 0.435 to 0.6333 and root mean squared error (RMSE) decreased from 0.655 to 0.534 L/min after the time adjustment.

Figure 5. Bland–Altman plot for time-adjusted continuous thermodilution cardiac output (COcont, reference method) and arterial pressure waveform-based cardiac output estimation (CO-FloTrac; FloTrac/Vigileo system). Black and red solid lines indicate mean bias and limits of agreement, respectively. Dashed lines indicate 95% confidence intervals of the corresponding parameters. The negative mean bias presented in the figure indicates the overestimation of CO-FloTrac, and 95% of the biases (difference between the two measurements) reside within the presented limits of agreement.

Figure 6. Four-quadrant plot for 3 to 20 min interval change of continuous thermodilution cardiac output (COcont) and arterial pressure waveform-based cardiac output estimation (CO-FloTrac). Each axis indicates interval change of the corresponding parameter. Red rectangles (inner box) indicate the exclusion zone of 10% and deep pink rectangles (outer box) indicate the exclusion zone of 15%. Colored numbers indicate corresponding concordance rates.