Respiration-related variations in central venous pressure as predictors of fluid responsiveness in spontaneously breathing patients.

Abstract

The hemodynamic parameters used to accurately predict fluid responsiveness (FR) in spontaneously breathing patients (SB) require specific material and expertise. Measurements of the central venous pressure (CVP) are relatively simple and, importantly, are feasible in many critically ill patients. We analyzed the accuracy of respiration-related variations in CVP (vCVP) to predict FR in SB patients, and examined the optimization of its measurement using a standardized, deep inspiratory maneuver. We performed a monocentric, prospective, diagnostic evaluation. Spontaneously breathing patients in intensive care units with a central venous catheter were prospectively included. vCVP was measured while the patient was spontaneously breathing, both with (vCVP-st) and without (vCVP-ns) a standardized inspiratory maneuver, and calculated as: Minimum inspiratory v-wave peak pressure - Maximum expiratory v-wave peak pressure. A passive leg raising-induced increase in the left ventricular outflow tract velocity-time integral ≥10% defined FR. Among 63 patients, 38 (60.3%) presented FR. vCVP-ns was not significantly different between responders and non-responders (-4.9 mmHg [-7.5; -3.1] vs. -4.1 mmHg [-5.4; -2.8], respectively; p = 0.15). vCVP-st was lower in responders than non-responders (-9.7 mmHg [-13.9; -6.2] vs. -3.6 mmHg [-10.6; -1.6], respectively; p = 0.004). A vCVP-st < -4.7 mmHg predicted FR with 89.5% sensitivity, a specificity of 56.0%, and an area under the ROC curve of 0.72 [95% CI: 0.58; 0.86] (p = 0.004). When a central venous catheter is present, elevated values for vCVP-st may be useful to identify spontaneously breathing patients unresponsive to volume expansion. Nevertheless, the necessity of performing a standardized, deep-inspiration maneuver may limit its clinical application.