Objective: To evaluate the accuracy of a new continuous cardiac output monitor (one based on the thermodilution principle) in critically ill patients.
Design: Criterion standard study.
Setting: Multidisciplinary intensive care unit in a university hospital.
Patients: Fourteen critically ill patients, with different diseases, requiring pulmonary artery catheterization.
Interventions: In two patients with a left ventricular assist system, a defined, sudden 1 L/min change in cardiac output was carried through to evaluate the in vivo response time of the continuous cardiac output monitoring system. In the remaining 12 patients, cardiac output was altered by varying the dose of catecholamines, by volume loading, or by varying the level of sedation. In four patients, a rapid infusion of cold saline was given through a central venous catheter to test the performance of the system under these conditions.
Measurements and main results: Cardiac output was monitored continuously. A total of 163 (13 to 18 per patient) bolus determinations of cardiac output were performed, using the conventional thermodilution technique and simultaneously using the indocyanine green dye dilution technique. The range of cardiac output was 3.8 to 15.6 L/min. The results of the continuous thermodilution method were compared with the results of the bolus thermodilution and the dye dilution methods, respectively. The mean difference (bias) +/- SD of differences (precision) was 0.35 +/- 1.01 L/min for continuous vs. bolus thermodilution and 0.34 +/- 1.01 L/min for continuous thermodilution vs. indocyanine green dye dilution. Linear regression (correlation) analyses were y = 0.95x + 0.76 (r2 = .91) for continuous and bolus thermodilution and y = 0.93x + 0.87 (r2 = .91) for continuous thermodilution and dye dilution. The 75% in vivo response time was 10.5 mins. The infusion of cold isotonic saline led to erroneous continuous cardiac output values. When the conventional bolus thermodilution and dye dilution techniques were compared, mean difference was -0.01 +/- 0.54 L/min and the results of linear regression analyses were y = 0.97x + 0.22 (r2 = .97).
Conclusions: Continuous cardiac output measurement using the thermodilution technique is reasonably accurate and is reliable and applicable in routine clinical practice, and therefore may add to patient safety. However, the response time is too slow for the immediate detection of acute changes in cardiac output. Some clinical conditions such as the rapid infusion of cold solutions can interfere with the continuous cardiac output measurement. Conventional bolus thermodilution and indocyanine green dye dilution methods showed good agreement and can be used interchangeably.