碳青霉烯类耐药肠杆菌科细菌血流感染危险因素及其列线图模型的建立

Risk factors and nomogram model for the bloodstream infection caused by carbapenem-resistant Enterobacteriaceae

目的 探讨碳青霉烯类耐药肠杆菌科细菌血流感染危险因素及其预测模型的建立。方法 回顾性分析2020年1月-2021年12月徐州医科大学附属连云港医院120例碳青霉烯类耐药肠杆菌科细菌(CRE)血流感染患者的临床相关资料,并选取同期碳青霉烯类敏感肠杆菌科细菌(CSE)血流感染病例80例为其对照组。分析CRE组患者病原菌及其检出科室,并比较两组年龄、住重症监护室(ICU)时间、抽血培养时Pitt菌血症评分等临床资料,采用单因素分析比较CRE和CSE两组间差异,多因素Logistic回归分析CRE血流感染发生的危险因素,开发列线图预测CRE血流感染的风险,采用受试者工作(ROC)曲线、标准曲线、决策曲线检测该模型对CRE血流感染发生的预测效能。结果 肺炎克雷伯菌、大肠埃希菌和阴沟肠杆菌是CRE最主要的菌株,ICU、泌尿外科和神经外科是最主要的检出科室;Logistic回归分析结果显示,高龄、慢性阻塞性肺疾病、手术引流、腹腔内感染、住ICU时间、抽血培养时Pitt菌血症评分是CRE血流感染的独立危险因素,训练集预测模型的曲线下面积(AUC)值为0.710(95%CI:0.589~0.831),敏感度为62.90%,特异度为90.90%;验证集模型ROC曲线显示AUC值为0.818(95CI:0.622~1.000),灵敏度为72.70%,特异度为94.10%。结论 高龄、慢性阻塞性肺疾病、手术引流、腹腔内感染、住ICU时间、抽血培养时Pitt菌血症评分是CRE血流感染的独立危险因素,经列线图建立的风险预测模型能较好地预测CRE血流感染的发生风险。

OBJECTIVE To explore the risk factors for the bloodstream infection caused by carbapenem-resistant Enterobacteriaceae and to establish the prediction model.METHODS Clinical data of 120 patients with bloodstream infection caused by carbapenem-resistant Enterobacteriaceae(CRE)in Lianyungang Hospital Affiliated to Xuzhou Medical University from Jan 2020 to Dec 2021 were retrospectively analyzed,and 80 patients with bloodstream infection by carbapenem-sensitive Enterobacteriaceae bacteria(CSE)during the same period were selected as the control group.The pathogenic bacteria isolated from patients and departments in the CRE group were analyzed,and the clinical data of the two groups were compared,such as age,ICU stay time,Pitt bacteremia score during blood culture.Univariate analysis was used to compare the differences in CRE and CSE between the two groups,and multivariate logistic regression analysis was used to analyze the risk factors for CRE bloodstream infection.Receiver operating characteristic(ROC)curve,standard curve and decision curve were used to test the prediction efficiency of this model for CRE bloodstream infection occurrence.RESULTS Klebsiella pneumoniae,Escherichia coli and Enterobacter cloacae were the main strains of CRE,and ICU,urologic surgery and neurosurgery were the main departments where the positive specimens came from.Multivariate logistic regression analysis showed that old age,chronic obstructive pulmonary disease,surgical drainage,intra-abdominal infection,ICU stay time,Pitt bacteremia score during blood culturation were independent risk factors for CRE bloodstream infection.The area under the curve(AUC)of the training set was 0.710(95%CI:0.589-0.831),with the sensitivity of 62.90%,and the specificity of 90.90%.The ROC curve of validation set model showed that the AUC value was 0.818(95CI:0.622-1.000),with the sensitivity of 72.70%,and the specificity of 94.10%.CONCLUSION Advanced age,chronic obstructive pulmonary disease,surgical drainage,intrabitoneal infection,ICU stay,Pitt bacteremia score during blood culture were independent risk factors for CRE bloodstream infection.The risk prediction model established by the graph could well predict the risk of CRE bloodstream infection.