Fig. 2

ROC curve analyses for the development of postoperative MACE. The mean PAP (A) and mean CVP (B) predicted development of postoperative MACE with moderate (AUC 0.726; 95% CI, 0.645 to 0.808) and low accuracy (AUC 0.616; 95% CI, 0.523 to 0.708), respectively. C. A multivariate ROC curve analysis was performed using mean PAP and model 1 (NYHA class, preoperative beta-blocker use, left ventricular ejection fraction, emergency surgery, intra-aortic balloon pump use, mean CVP). The solid line shows model 1, while the dashed line shows the addition of mean PAP to model 1. A multivariate ROC curve analysis showed that adding the mean PAP [mmHg] to model 1 achieved a better predictive value than model 1 alone in predicting MACE, with an AUC of 0.794, which is an increase of 0.059 (P = 0.036). The continuous net reclassification improvement and integrated discrimination improvement were 0.510 (95% CI, 0.169 to 0.851; P = 0.003) and 0.0071 (95% CI, -0.020 to 0.034; P = 0.603), respectively. AUC Area under the receiver operating characteristic curve, CI Confidence interval, CVP Central venous pressure, MACE Major adverse cardiovascular events, NYHA New York Heart Association, PAP Pulmonary artery pressure, ROC Receiver operating characteristic