构建个体化预测乳腺癌患者术后医院感染的Nomogram模型

Construction of the Nomogram Model for Individualized Prediction of Postoperative Nosocomial Infection in Breast Cancer Patients

目的探究乳腺癌患者术后发生医院感染的危险因素,并建立个体化预测Nomogram模型。方法回顾性分析我院收治的136例乳腺癌术后患者临床资料。比较术后发生医院感染和未发生医院感染患者的一般资料及疾病情况,采用单因素和多因素Logistics回归分析发生医院感染的危险因素,建立Nomogram预测模型并计算C指数,应用验证集以受试工作特征(ROC)曲线评估模型的预测效能。结果136例乳腺癌患者术后有24例患者发生了医院感染,发生率为17.65%。Logistic回归分析显示,患者年龄≥60岁、使用抗菌药物不合理、白细胞计数<4×10^(9)/L、血清白蛋白<30g/L、血糖水平>6.1mmol/L、APACHEⅡ评分≥20分及化疗是患者发生医院感染的独立危险因素(P<0.05)。发生医院感染Nomogram模型采用重复Bootsrap法内部验证AUC为0.901,提示分区度良好,且观测值与预测值一致性良好。结论患者年龄≥60岁、使用抗菌药物不合理、白细胞计数<4×10^(9)/L、血清白蛋白<30g/L、血糖水平>6.1mmol/L、APACHEⅡ评分≥20分及化疗是患者发生医院感染的独立危险因素,在此因素基础上构建的Nomogram模型可准确预测医院感染的发生风险,帮助早干预早治疗,促进预后。

Objective To explore the risk factors of postoperative nosocomial infection in breast cancer patients,and establish an individualized predictive Nomogram model.Methods A total of 136 postoperative breast cancer patients admitted to our hospital from January 2017 to October 2020 were selected as the research objects and their clinical data was retrospectively analyzed.The general information and disease conditions of patients with and without nosocomial infection after operation were compared.The risk factors for nosocomial infection were analyzed by single and multivariate logistic regression.The Nomogram prediction model was established and the C index was calculated.The operating characteristic(ROC)curve was used to evaluate the predictive performance of the model.Results There were 24 cases(17.65%)patients with breast cancer after operation developed nosocomial infection.Logistic regression analysis showed that age≥60 years,unreasonable use of antibiotics,WBC<4×10^(9)/L,serum albumin<30 g/L,blood glucose level>6.1 mmol/L,APACHEⅡscore≥20 and chemotherapy were independent risk factors for nosocomial infection(P<0.05).The Nomogram model for nosocomic infection verified by repeated Bootsrap method,and the internal AUC was 0.901,indicating that the partitioning degree was good,and the observed and predicted values were in good agreement.Conclusion Patients’age≥60 years,unreasonable use of antibiotics,WBC<4×10^(9)/L,serum albumin<30 g/L,blood glucose level>6.1 mmol/L,APACHEⅡscore≥20 and chemotherapy were all independent risk factors for nosocomial infection.Based on these factors,the Nomogram model could accurately predict the risk of nosocomial infection,help early intervention and early treatment,and promote prognosis.