基于多因素分析的食管癌术后肺部感染危险因素及抗生素精准使用预测模型构建OA

Objective An analysis was conducted on the risk and protective factors associated with pulmonary infections following esophageal cancer surgery.Subsequently,a nomogram prediction model was developed to offer theoretical guidance for the precise administration of early antibiotic therapy in the postoperative management of esophageal cancer patients.Methods A thorough investigation was performed involving 300 patients who received radical resection for esophageal cancer at our institution from December 2020 to December 2024.Based on the occurrence of postoperative pulmonary infections,the patients were categorized into two groups:the infection group,comprising 114 cases,and the control group,consisting of 186 cases.To identify risk and protective factors associated with postoperative pulmonary infections,both univariate and multivariate logistic regression analyses were employed.Utilize the R programming language to partition the dataset into a training set and a validation set,adhering to an 8:2 split ratio.Subsequently,develop a nomogram prediction model and conduct internal validation of the model.The model's predictive capability was evaluated through ROC analysis,calibration graphs,Hosmer-Lemeshow goodness-of-fit assessment,and clinical utility examination.Results The factors of age,smoking history,diabetes history,chronic obstructive pulmonary disease history,presence of an anastomotic fistula,and the duration of the surgical procedure were identified as independent risk factors for the development of pulmonary infections following esophageal cancer surgery(P＜0.05).Conversely,the postoperative administration of high-grade antibiotics emerged as an independent protective factor,demonstrating a significant association with the incidence of postoperative pulmonary infections in esophageal cancer patients(P＜0.05).Based on this,establish a column chart model for postoperative infection risk;Through ROC curve analysis indicated that within the training set,the area under the curve(AUC)for the model was 0.75(95%CI:0.68～0.83),while the AUC for the validation set was 0.80(95%CI:0.72～0.88),thereby affirming the model's robust predictive capability.The calibration curve training set Hosmer Lemeshow test has a P value of 0.232,and the validation set Hosmer Lemeshow test has a P value of 0.237,both of which are P＞0.05.This prediction model has good accuracy and fitting degree;The decision curve indicates that the model can produce better clinical benefits when the training set is within the risk threshold probability range of(0.12～0.81)and the validation set is within the risk threshold probability range of(0.05～0.83).Conclusion The nomogram model constructed in this study has high predictive power for postoperative lung infection in esophageal cancer,and early use of advanced antibiotics after surgery can effectively prevent infection.