【摘要】 目的 构建预测碳青霉烯类耐药肠杆菌(CRE)院内感染风险的模型。方法 收集某医院2019年6月—2024年6月肠杆菌院内感染患者资料，将患者分为CRE组和碳青霉烯类敏感肠杆菌(CSE)组，构建基于岭回归的风险预测模型，使用Hosmer-Lemeshow方法进行模型验证。结果 共收集350例患者资料，主要感染菌为肺炎克雷伯菌和大肠埃希菌。岭回归法多因素分析显示，连续发热≥7 d、呼吸机使用、导尿管使用、合并其他细菌感染、使用碳青霉烯类抗菌药物、使用抗真菌药物、患有肺部疾病、患有肾脏疾病、手术史为CRE院内感染的独立危险因素(P值均<0.05)。构建的预测模型展现出良好的预测性能，其ROC曲线下的面积(AUC)为0.87,灵敏度为0.82,特异度为0.79。结论 本研究开发的CRE院内感染风险预测模型具有较高的预测准确性，能够为临床实践中CRE院内感染的防控提供理论依据和实用建议。更多还原

【Abstract】 Objective To construct a risk prediction model for nosocomial infection with carbapenem-resistant Enterobacteriaceae(CRE). Methods The data of patients with nosocomial infection of Enterobacteriaceae from June 2019 to June 2024 in a hospital were collected.The patients were divided into CRE group and carbapenem-susceptible Enterobacteriaceae(CSE) group to construct a risk prediction model based on ridge regression, and the model validation was performed using the Hosmer-Lemeshow method. Results The data of 350 patients were collected, with the primary infecting bacteria being Klebsiella pneumoniae and Escherichia coli.Multivariate analysis by ridge regression revealed that continuous fever ≥ 7 d, ventilator use, urinary catheter use, co-infection with other bacteria, carbapenems use, antifungal use, underlying lung disease, kidney disease, and surgical history were independent risk factors for nosocomial infection with CRE(all P values<0.05).The constructed prediction model exhibited good prediction performance, with an area under the ROC curve(AUC) of 0.87,a sensitivity of 0.82,and a specificity of 0.79. Conclusions The CRE nosocomial infection risk prediction model developed in this study features high prediction accuracy and can provide theoretical basis and practical suggestions for the prevention and control of CRE nosocomial infection in clinical practice.更多还原