Frontiers in Cardiovascular Medicine (Oct 2022)

Early identification of delayed extubation following cardiac surgery: Development and validation of a risk prediction model

  • Xia Li,
  • Jie Liu,
  • Zhenzhen Xu,
  • Yanting Wang,
  • Lu Chen,
  • Yunxiao Bai,
  • Wanli Xie,
  • Qingping Wu

DOI
https://doi.org/10.3389/fcvm.2022.1002768
Journal volume & issue
Vol. 9

Abstract

Read online

BackgroundSuccessful weaning and extubation after cardiac surgery is an important step of postoperative recovery. Delayed extubation is associated with poor prognosis and high mortality, thereby contributing to a substantial economic burden. The aim of this study was to develop and validate a prediction model estimate the risk of delayed extubation after cardiac surgery based on perioperative risk factors.MethodsWe performed a retrospective cohort study of adult patients undergoing cardiac surgery from 2014 to 2019. Eligible participants were randomly assigned into the development and validation cohorts, with a ratio of 7:3. Variables were selected using least absolute shrinkage and selection operator (LASSO) logistic regression model with 10-fold cross-validation. Multivariable logistic regression was applied to develop a predictive model by introducing the predictors selected from the LASSO regression. Receiver operating characteristic (ROC) curve, calibration plot, decision curve analysis (DCA) and clinical impact curve were used to evaluate the performance of the predictive risk score model.ResultsAmong the 3,919 adults included in our study, 533 patients (13.6%) experienced delayed extubation. The median ventilation time was 68 h in the group with delayed extubation and 21 h in the group without delayed extubation. A predictive scoring system was derived based on 10 identified risk factors based on 10 identified risk factors including age, BMI ≥ 28 kg/m2, EF < 50%, history of cardiac surgery, type of operation, emergency surgery, CPB ≥ 120 min, duration of surgery, IABP and eGFR < 60 mL/min/1.73 m2. According to the scoring system, the patients were classified into three risk intervals: low, medium and high risk. The model performed well in the validation set with AUC of 0.782 and a non-significant p-value of 0.901 in the Hosmer-Lemeshow test. The DCA curve and clinical impact curve showed a good clinical utility of this model.ConclusionsWe developed and validated a prediction score model to predict the risk of delayed extubation after cardiac surgery, which may help identify high-risk patients to target with potential preventive measures.

Keywords