Factors associated with postoperative occurrence of hypoxemia in patients undergoing total knee replacement surgery and its predictive model creation and validation

This study aims to explore the risk factors of postoperative hypoxemia in patients undergoing total knee replacement surgery and construct and validate the corresponding prediction model. Clinical data of patients who underwent total knee replacement surgery between January 2022 and December 2024 were retrospectively collected in our hospital. Independent risk factors for postoperative hypoxemia were screened using univariate and multivariate logistic regression analyses, and the model was visualized by drawing a nomogram. The diagnostic performance of the model was assessed using the receiver operating characteristic and its area under the curve, while the model fit was evaluated by the Hosmer-Lemeshow goodness-of-fit test. To enhance the reliability of the validation results, internal validation was performed by combining the Bootstrap method and 10-fold cross-validation, and the clinical applicability value of the model was assessed using calibration curve and decision curve analysis. A total of 569 total knee arthroplasty (TKA) patients were included, of which 117 developed postoperative hypoxemia with an incidence rate of 20.56%. The results of univariate and multivariate logistic regression analyses showed that no nerve block was performed (OR = 1.541; 95% CI: 1.083-2.204), age over 65 years (OR = 2.040; 95% CI: 1.323-3.379), and comorbid chronic obstructive pulmonary disease (OR = 2.783; 95% CI: 1.742-5.038), American Society of Anesthesiologists classification greater than II (OR = 1.824; 95% CI: 1.197-2.915), and intraoperative intravenous fluid intake of more than 1500 mL (OR = 1.470; 95% CI: 1.037-2.086) were independent risk factors for postoperative hypoxemia. After subject work characteristic curve analysis and Hosmer-Lemeshow goodness-of-fit test, combined with Bootstrap and 10-fold cross-validation for internal validation, the results showed that the model had good discriminative ability and fitting effect, and the model performance was stable and of high clinical utility. The occurrence of hypoxemia after TKA is influenced by multiple factors, and the nomogram prediction model established in this study demonstrated high accuracy, which is helpful for early clinical identification and intervention in high-risk patients.