Predicting Fluid Responsiveness in Critically Ill Patients by Using Combined End-Expiratory and End-Inspiratory Occlusions With Echocardiography

Objectives: First, we aimed at assessing whether fluid responsiveness is predicted by the effects of an end-expiratory occlusion on the velocity-time integral of the left ventricular outflow tract. Second, we investigated whether adding the effects of an end-inspiratory occlusion and of an end-expiratory occlusion on velocity-time integral can predict fluid responsiveness with similar reliability than end-expiratory occlusion alone but with a higher threshold, which might be more compatible with the precision of echocardiography. Design: Diagnostic study. Setting: Medical ICU. Patients: Thirty mechanically ventilated patients in whom fluid administration was planned. Interventions: A 15-second end-expiratory occlusion and endinspiratory occlusion, separated by 1 minute, followed by a 500mL saline administration. Measurements and Main Results: Pulse contour analysis-derived cardiac index and velocity-time integral were measured during the last 5 seconds of 15-second end-inspiratory occlusion and endexpiratory occlusion and after fluid administration. End-expiratory occlusion increased velocity-time integral more in responders than in nonresponders to fluid administration (11% +/- 5% vs 3% +/- 1%, respectively; p < 0.0001), and end-inspiratory occlusion decreased velocity-time integral more in responders than in nonresponders (12% +/- 5% vs 5% +/- 2%, respectively; p = 0.0002). When adding the absolute values of changes in velocity-time integral observed during both occlusions, velocity-time integral changed by 23% +/- 9% in responders and by 8% +/- 3% in nonresponders. Fluid responsiveness was predicted by the end-expiratory occlusion-induced change in velocity-time integral with an area under the receiver operating characteristic curve of 0.938 (0.785-0.989) and a threshold value of 5%. Fluid responsiveness was predicted by the sum of absolute values of changes in velocity-time integral during both occlusions with a similar reliability (area under the receiver operating characteristic curve = 0.973 [0.838-1.000]) but with a threshold of 13%. Both sensitivity and specificity were 93% (68-100%). Conclusions: If consecutive end-inspiratory occlusion and endexpiratory occlusion change velocity-time integral is greater than or equal to 13% in total, fluid responsiveness is accurately predicted. This threshold is more compatible with the precision of echocardiography than that obtained by end-expiratory occlusion alone.