Construction and validation of a risk prediction model for intraoperative hypothermia in children undergoing thoracoscopic lobectomy.

Abstract

To develop a predictive model to assess the risk of peri-operative hypothermia occurrence in children undergoing thoracoscopic lobectomy and validate its effectiveness. 172 children who underwent thoracoscopic lobectomy were randomly divided into a training set of 124 cases and a validation set of 48 cases in a 7:3 ratio. The study found a 33.1% incidence of intraoperative hypothermia (IPH) in children undergoing thoracoscopic lobectomy. Logistic regression analysis was performed using SPSS 26.0 to screen influencing factors, establish a risk prediction model, and draw the receiver-operating characteristic curve. Univariate analysis identified anesthesia-skin incision time, operating-room temperature, humidity, basal body temperature, end body temperature, and body temperature half an hour after anesthesia as influencing factors. Multivariate logistic regression revealed anesthesia-skin incision time (odds ratio (OR) = 1.595), operating-room humidity (OR = 4.094), and body temperature half an hour after anesthesia (OR = 112.595) as independent predictors. The nomogram model demonstrated an excellent discrimination with area under the curve (AUC) of 0.989 (95% confidence interval (CI): 0.976-1.000), maximum Youden index of 0.94, sensitivity of 1, specificity of 0.94, and cutoff value of 0.195. The Hosmer-Lemeshow test (χ₂ = 1.751, P = 0.195) and bootstrap resampling (consistency coefficient = 0.947) confirmed the model's goodness of fit, internal consistency, and stability. Validation set results are similar to those in the training set, with an AUC of 0.989 (95% CI: 0.969-1.000), a sensitivity of 1, a specificity of 0.929, and a calibration curve error of 0.032 (<0.05), indicating high predictive accuracy. These findings suggest nomogram is a robust tool for predicting IPH in the pediatric thoracoscopic lobectomy.