While invasive thermodilution techniques remain the reference methods for cardiac output (CO) measurement, there is a currently unmet need for non-invasive techniques to simplify CO determination, reduce complications related to invasive procedures required for indicator dilution CO measurement, and expand the application field toward emergency room, non-intensive care, or outpatient settings. We evaluated the performance of a non-invasive oscillometry-based CO estimation method compared to transpulmonary thermodilution. To assess agreement between the devices, we used Bland-Altman analysis. Four-quadrant plot analysis was used to visualize the ability of Mobil-O-Graph (MG) to track CO changes after a fluid challenge. Trending analysis of CO trajectories was used to compare MG and PiCCO® calibrated pulse wave analysis over time (6 h). We included 40 patients from the medical intensive care unit at the Charite - Universitatsmedizin Berlin, Campus Benjamin Franklin between November 2019 and June 2020. The median age was 73 years. Forty percent of the study population was male; 98% was ventilator-dependent and 75% vasopressor-dependent at study entry. The mean of the observed differences for the cardiac output index (COI) was 0.7 l*min(-1*)m(-2) and the lower, and upper 95% limits of agreement (LOA) were -1.9 and 3.3 l*min(-1*)m(-2), respectively. The 95% confidence interval for the LOA was ± 0.26 l*min(-1*)m(-2), the percentage error 83.6%. We observed concordant changes in CO with MG and PiCCO® in 50% of the measurements after a fluid challenge and over the course of 6 h. Cardiac output calculation with a novel oscillometry-based pulse wave analysis method is feasible and replicable in critically ill patients. However, we did not find clinically applicable agreement between MG and thermodilution or calibrated pulse wave analysis, respectively, assessed with established evaluation routine using the Bland-Altman approach and with trending analysis methods. In summary, we do not recommend the use of this method in critically ill patients at this time. As the basic approach is promising and the CO determination with MG very simple to perform, further studies should be undertaken both in hemodynamically stable patients, and in the critical care setting to allow additional adjustments of the underlying algorithm for CO estimation with MG.
