Objective: The Modelflow method computes cardiac output (CO) from arterial pressure (CO-MFao) by simulating a non-linear three-element Windkessel model of aortic input impedance. We present a novel technique to apply the Modelflow method to the left ventricular pressure (Plv) signal, to obtain an estimation of CO (CO-MFlv).
Methods: We extended the model by simulating the aortic valve as a resistance placed in series to the characteristic impedance. In our model the valve resistance is infinite in diastole, while it has a patient related value during systole. Twenty one patients with heart failure were studied with a combined pressure-conductance catheter positioned in the left ventricle and a micromanometer in the aorta. CO changes were induced during temporary sequential atrio-biventricular pacing with different atrio-ventricular and inter-ventricular intervals. After a single calibration we compared CO-MFlv with CO-MFao (516 measurement series) and with CO estimated by the conductance catheter volume technique (CO-cond) (267 series).
Results: CO ranged from 2.42 to 7.59 l/min with an overall mean of 4.36 (1.38) l/min. Comparing the two Modelflow methods, the bias was -0.04 (0.36) l/min, with limits of agreement of -0.77 and 0.70 l/min, and a coefficient of variation of 8.4%. Of the 516 changes in CO from baseline values, 112 were greater than 0.5 l/min and 110 of these (98%) were scored in the same direction by both methods. For comparisons between CO-MFlv and CO-cond the bias was -0.10 (0.49), with limits of agreement of -1.12 and 0.90 l/min, and a coefficient of variation of 12.5%.
Conclusions: Cardiac output estimates by the modelflow method from aortic pressure and left ventricular pressure are interchangeable in patients without mitral and aortic abnormalities. After an initial calibration, CO-MFlv presents near zero bias and an adequate precision.