Hypoxemia is a very common issue in patients undergoing heart valve surgery (HVS), related to poor clinical outcomes. However, studies on severe hypoxemia (SH) after HVS have not been reported. The aims of this study were to identify predictors for SH in patients undergoing HVS and to develop and validate a risk prediction model.
Patients undergoing HVS between 2016 and 2019 in a cardiovascular center were enrolled and were assigned to training and validation sets by a 7:3 ratio. Based on whether patients developed SH, they were divided into two groups. By univariate and multivariate analysis, predictors for SH were identified. Based on the predictors and logistic rule, a nomogram and a risk calculator were generated. The model was evaluated using calibration, discrimination and clinical utility.
The incidence rates of SH, moderate hypoxemia and mild hypoxemia were respectively 2.4, 23.9, and 58.2%. By multivariate analysis, seven independent risk factors for SH after HVS were identified, including body mass index, chronic obstructive pulmonary disease, renal insufficiency, white blood cell count, serum globulin, cardiopulmonary bypass time, and surgical types. The logistic model demonstrated satisfactory discrimination, calibration and clinical utility in both the training and validation sets. A nomogram and a risk calculator based on the logistic model were generated for easy application. Risk stratification was performed and three risk intervals were defined according to the nomogram and clinical practice. In addition, compared to patients without SH, patients with SH had significantly poorer clinical outcomes.
Postoperative hypoxemia was prevalent after HVS, related to poor clinical outcomes. A logistic model including seven independent predictors for SH after HVS were established and validated, which demonstrated satisfactory discrimination, calibration and clinical utility. The results of this study may provide help to individualized risk assessment, early prevention and perioperative management.