Evaluation of Several Methods for Computing Stroke Volume from Central Aortic Pressure

Abstract

Six pulse-contour methods for estimating stroke volume from a single central aortic blood pressure were evaluated in 8 dogs and 17 patients. In the dogs, wide variations in stroke volume, measured with an electromagnetic flowmeter, were obtained by pacing the heart at various rates during a control period and during several pharmacologic interventions. Good correlations existed between measured stroke volume and most estimators when the data from each intervention were analyzed separately. However, regression analysis revealed considerable variation in the individual slopes and intercepts, and thus a poor correlation was obtained when all data for one dog were combined in a single analysis. Similar evaluations were carried out in two groups of patients in whom the pressure-gradient technique was used to measure stroke volume. In the group with minimal variations in hemodynamic status, the correlations between estimated and true stroke volume were reasonably good. In the patients having a wide range of hemodynamic conditions, considerable variation in both slopes and intercepts was observed, and the combined correlation coefficients were generally poor. Although pulse contour methods of estimating stroke volume may work reasonably well over a range of stroke volumes when the variation is induced by a single perturbing agent, none of these methods perform adequately when the variation is induced by multiple perturbing agents; thus their clinical usefulness is markedly limited.