Multiparameter Predictor of Fluid Responsiveness in Cardiac Surgical Patients Receiving Tidal Volumes Less Than 10 mL/kg

Abstract

Introduction. We hypothesize that respiratory variation in the pulmonary artery tracing predicts fluid responsiveness (primary hypothesis) and that inclusion of multiple physiologic waveforms as well as ventilator settings in a predictive model of fluid responsiveness would lead to improvements in the clinical utility of this class of metrics (secondary hypothesis). Methods. Blood pressure tracings were prospectively recorded in 35 patients immediately following cardiac surgery. Fluid bolus administration data, ventilator settings, and cardiac output were recorded prospectively before and after fluid boluses given at the discretion of the treating physician. Results. We observed statistically significant but limited relationships between pulmonic (r² = .26, P = .0052) and systemic (r² = .13, P = .011) pulse pressure variation and changes in cardiac index. A multiparameter estimate of fluid responsiveness, which included respiratory variation in central venous pressure and pulmonary artery pressure, indexed tidal volumes, positive end-expiratory pressure, and mean airway pressure, was also correlated with change in cardiac index (r² = .42, P = .0056). Using the area under the curve (AUC) technique to compare specificity and sensitivity, dynamic indicators (AUC = 0.74, 0.67, and 0.81 for systemic arterial respiratory [pulse pressure] variation, pulmonic arterial respiratory [pulse pressure] variation, and the multiparameter estimate, respectively) outperformed static estimates (0.49 and 0.48 for central venous pressure and pulmonary artery diastolic pressure, respectively). Conclusion. While integration of multiple physiologic waveforms as well as ventilator parameters improves the predictability of fluid responsive metrics in the setting of lung-protective ventilation, the composite index may still be of limited predictive value.