A predictive model for early postoperative hypoxemia after mitral valve replacement combined with pulmonary arterial hypertension.

Abstract

To establish and validate a novel scoring system on the basis of a nomogram for preoperative prediction of early hypoxemia after mitral valve replacement combined with pulmonary arterial hypertension. For this retrospective cohort study, clinical data from 430 patients with mitral valve disease combined with pulmonary arterial hypertension were collected. Early postoperative hypoxemia was defined as hypoxemia that occurred within 24 hours after operation. Clinical data from 430 patients were subjected to univariate logistic regression analysis (P < .05), and the results were then included in a stepwise multivariate logistic regression analysis (P < .05) to derive independent risk factors for postoperative hypoxemia. All the data from 430 patients were randomly divided into the training and validation cohorts. Nomogram prediction models for postoperative hypoxemia were established using the training cohorts and validated with the validation cohorts. Univariate and stepwise multivariate logistic regression analyses suggested that the systolic pulmonary artery pressure, smoking, age, left atrial end-systolic dimension, and whether patients with accompanying atrial fibrillation underwent maze surgery (we replaced this factor with "atrial fibrillation-maze category" in the article that follows) were independent risk factors. A nomogram prediction model was developed accordingly. The area under the curve values of the training and validation cohorts were 0.838 (95% confidence interval, 0.783-0.892) and 0.799 (95% confidence interval, 0.730-0.869), respectively. The calibration curves were close to the ideal diagonal, and the decision curve analysis indicated a significant net benefit. The risk prediction model developed in this study is a desirable predictor of early postoperative hypoxemia after mitral valve replacement combined with pulmonary arterial hypertension.