Accuracy of pulse pressure variations for fluid responsiveness prediction in mechanically ventilated patients with biphasic positive airway pressure mode.

The accuracy of pulse pressure variation (PPV) to predict fluid responsiveness using pressure-controlled (PC) instead of volume-controlled modes is under debate. To specifically address this issue, we designed a study to evaluate the accuracy of PPV to predict fluid responsiveness in severe septic patients who were mechanically ventilated with biphasic positive airway pressure (BIPAP) PC-ventilation mode. 45 patients with sepsis or septic shock and who were mechanically ventilated with BIPAP mode and a target tidal volume of 7-8 ml/kg were included. PPV was automatically assessed at baseline and after a standard fluid challenge (Ringer's lactate 500 ml). A 15% increase in stroke volume (SV) defined fluid responsiveness. The predictive value of PPV was evaluated through a receiver operating characteristic (ROC) curve analysis and "gray zone" statistical approach. 20 (44%) patients were considered fluid responders. We identified a significant relationship between PPV decrease after volume expansion and SV increase (spearman ρ = - 0.5, p < 0.001). The area under ROC curve for PPV was 0.71 (95%CI 0.56-0.87, p = 0.007). The best cut-off (based on Youden's index) was 8%, with a sensitivity of 80% and specificity of 60%. Using a gray zone approach, we identified that PPV values comprised between 5 and 15% do not allow a reliable fluid responsiveness prediction. In critically ill septic patients ventilated under BIPAP mode, PPV appears to be an accurate method for fluid responsiveness prediction. However, PPV values comprised between 5 and 15% constitute a gray zone that does not allow a reliable fluid responsiveness prediction.

Machine learning methods to improve bedside fluid responsiveness prediction in severe sepsis or septic shock: an observational study.

BACKGROUND: Passive leg raising (PLR) predicts fluid responsiveness in critical illness, although restrictions in mobilising patients often preclude this haemodynamic challenge being used. We investigated whether machine learning applied on transthoracic echocardiography (TTE) data might be used as a tool for predicting fluid responsiveness in critically ill patients. METHODS: We studied, 100 critically ill patients (mean age: 62 yr [standard deviation: 14]) with severe sepsis or septic shock prospectively over 24 months. Transthoracic echocardiography measurements were performed at baseline, after PLR, and before and after a standardised fluid challenge in learning and test populations (n=50 patients each). A 15% increase in stroke volume defined fluid responsiveness. The machine learning methods used were classification and regression tree (CART), partial least-squares regression (PLS), neural network (NNET), and linear discriminant analysis (LDA). Each method was applied offline to determine whether fluid responsiveness may be predicted from left and right cardiac ventricular physiological changes detected by cardiac ultrasound. Predictive values for fluid responsiveness were compared by receiver operating characteristics (area under the curve [AUC]; mean [95% confidence intervals]). RESULTS: In the learning sample, the AUC values were PLR 0.76 (0.62-0.89), CART 0.83 (0.73-0.94), PLS 0.97 (0.93-1), NNET 0.93 (0.85-1), and LDA 0.90 (0.81-0.98). In the test sample, the AUC values were PLR 0.77 (0.64-0.91), CART 0.68 (0.54-0.81), PLS 0.83 (0.71-0.96), NNET 0.83 (0.71-0.94), and LDA 0.85 (0.74-0.96) respectively. The PLS model identified inferior vena cava collapsibility, velocity-time integral, S-wave, E/Ea ratio, and E-wave as key echocardiographic parameters. CONCLUSIONS: Machine learning generated several models for predicting fluid responsiveness that were comparable with the haemodynamic response to PLR.