The multivariable prognostic models for severe complications after heart valve surgery

Area of Science:

• Cardiovascular Surgery
• Medical Informatics
• Predictive Analytics

Background:

• Heart valve surgery carries risks of severe complications such as low cardiac output syndrome (LCOS), acute kidney injury requiring hemodialysis (AKI-rH), and multiple organ dysfunction syndrome (MODS).
• Accurate prediction of these complications is crucial for patient management and improving surgical outcomes.

Purpose of the Study:

• To develop and validate multivariable prognostic models for predicting severe complications after heart valve surgery.
• To compare the predictive performance of models using only preoperative risk factors versus those including both preoperative and intraoperative factors.

Main Methods:

• Retrospective development of multivariate logistic regression models using data from 930 patients (2014-2015) and validation on 713 patients (2016-2017).
• Utilized the least absolute shrinkage and selection operator (LASSO) for model development.
• Assessed model discriminative ability using receiver operating characteristic (ROC) curves and calculated area under the curves (AUCs).

Main Results:

• Models incorporating both preoperative and intraoperative risk factors (PIRF models) demonstrated significantly higher predictive accuracy than models using only preoperative factors (PRF models).
• AUCs for PIRF models predicting LCOS, AKI-rH, and MODS were 0.821, 0.780, and 0.774, respectively, compared to lower AUCs for PRF models.
• Key intraoperative factors identified included auxiliary cardiopulmonary bypass time and combined tricuspid valve replacement.

Conclusions:

• Incorporating intraoperative risk factors substantially enhances the predictive power of prognostic models for severe complications following heart valve surgery.
• These improved models can aid in better risk stratification and clinical decision-making for patients undergoing heart valve procedures.