Thermodilution cardiac output: comparison between automated and manual injection of indicator

In clinical practice, cardiac output (CO) is usually reported as the average of thermodilution determinations with injection of the thermal indicator performed at end-exhalation. However, an average of multiple determinations with injections equally dispersed throughout the respiratory cycle has been shown to provide the best estimate of mean CO. This study sought to determine the reproducibility of CO determinations obtained with manual injections of indicator solution performed at end-exhalation, compared with those determined by computer-controlled injections equally dispersed throughout the breathing cycle of 27 patients undergoing cardiac operations. Mean CO was calculated by averaging the four determinations obtained with each technique before induction of anesthesia, after induction of anesthesia, after sternotomy, after cardiopulmonary bypass, and after sternal closure. A total of 130 pairs of mean CO estimations were obtained with manual and automated injections. Mean CO values obtained with manual injections were significantly lower than those obtained with the dispersed injection technique (5.0 +/- 1.4 L/min vs 5.3 +/- 1.6 L/min, P = 0.002). The bias between CO values measured with the manual technique was -0.25 +/- 0.47 L/min lower than those obtained with the dispersed technique. The mean relative bias for the group was 7 +/- 18% with 95% confidence intervals of +/- 26%. During mechanical ventilation, the thermodilution technique with manual injection of indicator solution significantly underestimated CO. Variability in the manual injection technique and inappropriate representation of the mean CO by injections timed to occur at end-exhalation contributed to the disparity. These results indicate that the manual technique of determining CO at end-exhalation may not accurately reflect the average CO.