Reproducibility of thermodilution cardiac output determination in critically ill patients: Comparison between bolus and continuous method

Objective. A semi-continuous thermodilution method (CCO) was recently developed to measure cardiac output with less risk of bacterial contamination, fluid overload, and user-induced errors than the classical bolus technique (BCO). Previous comparison between these two methods showed negligible bias. However, large limits of agreement suggest that the two methods are not interchangeable. We hypothesized that this poor agreement may be due to differences in reproducibility. Methods. In 23 critically ill patients, 369 paired measurements of CCO and BCO were compared (range of cardiac outputs: 2.8 to 16 L/min). The reproducibility of BCO and CCO methods was evaluated on a sample of 205 and 209 determinations, respectively. Results. The comparison between the CCO and the BCO methods confirmed previous results: i.e., small bias (-0.39 L/min) and large limits of agreement (-2.06 to +1.28 L/min). Reproducibility showed no bias for either the CCO or the BCO method. Limits of reproducibility agreement between repeated determinations were approximately 50% less for CCO than for BCO method: respectively -0.87 to +0.82 L/min for the CCO method and -1.56 to +1.37 L/min for the BCO method. Consequently the threshold necessary to ascertain that the difference-between two measurements was not due to the internal variability of the method (3 x SEM) was 0.39 for the CCO method and 0.75 L/min for the BCO method. Conclusion. Differences in reproducibility may explain the poor agreement between the CCO and BCO methods. The better reproducibility of the CCO method allows the detection of smaller variations in cardiac output and suggests the superiority of this new method.