Analysis of risk factors and establishment of a risk prediction model for cardiothoracic surgical intensive care unit readmission after heart valve surgery in China: A single-center study.

Background: Valvular heart disease is one of the most frequent and challenging heart diseases worldwide. The incidence of complications and cardiothoracic surgical intensive care unit (CSICU) readmission after cardiac valve surgery is high. Because CSICU readmission is costly and adversely impacts the quality life, reducing the risk of CSICU readmission has become one of the main focuses of health care.

Objective: To explore the risk factors for CSICU readmission and to establish a risk prediction model for CSICU readmission in heart valve surgical patients.

Methods: A total of 1216 patients who had undergone cardiac valvular surgery between January 2016 and August 2017 at the First Affiliated Hospital of Sun Yat-sen University were assigned as the development and validation data sets. Data from 824 patients in the development data set were retrospectively analyzed to identify potential risk factors with univariate analysis. Multivariate logistic regression was used to determine the independent risk factors of CSICU readmission, which served as the basis for our prediction model. The calibration and discrimination of the model were assessed by the Hosmer-Lemeshow (H-L) test and the area under the receiver operating characteristic (ROC) curve, respectively.

Results: Six preoperative variables (age ≥ 65, previous chronic lung disease, prior cardiac surgery, left ventricular ejection fraction (LVEF) ≤ 40%, 40% < LVEF ≤ 50%, and New York Heart Association (NYHA) classification III/IV), two intraoperative variables (multiple valve repair/replacement and cardiopulmonary bypass time ≥ 180 min), and five postoperative variables (cardiac arrest, acute respiratory distress syndrome, pneumonia, deep sternal wound infection, and renal failure) were independent risk factors of CSICU readmission. Our risk prediction model, which was established based on the above-mentioned risk factors, had robust discrimination and calibration in both the development and validation data sets.

Conclusion: The prediction model established in our study is a simple, objective, and accurate scoring system, which can be used to predict the risk of CSICU readmission and assist researchers with designing intervention strategies to prevent CSICU readmission.