重症监护病房患者耐碳青霉烯类肠杆菌感染风险预测模型的建立与效果评价

Establishment of risk prediction model for carbapenem resistant Enterobacteriaceae infection in intensive care unit and evaluation of effection

目的研究医院重症监护病房(ICU)耐碳青霉烯类肠杆菌(CRE)感染的危险因素,构建ICU住院患者CRE感染风险预测模型。方法采用回顾性调查和多因素分析方法,了解某医院ICU住院患者CRE感染的危险因素,构建CRE感染风险评分模型。结果模型组共纳入病例4 165例,验证组共纳入病例3 934例。最终纳入风险预测模型变量为ICU住院时间、机械通气、使用抗菌药物,对应风险预测模型分值为2分、5分及4分,分值7~9分为高风险人群。模型组H-L检验P值为0.304,ROC曲线下面积为0.84(95%CI:0.77~0.91,P<0.05)。验证组H-L检验P值为0.524,ROC曲线下面积为0.75(95%CI:0.68~0.82,P<0.05)。结论 CRE感染风险预测模型的判别效度和符合程度均较好,可用于重症监护病房CRE感染风险的主动预测。

Objective To research the possible risk factors of carbapenem-resistant Enterobacteria(CRE) infection in intensive care unit(ICU) of a tertiary general hospital,and to establish the risk prediction model of CRE infection in inpatients in ICU.Methods Retrospective study and multiple-factors analysis were used to construct a CRE risk scoring model and risk assessment model of CRE infection.Results A total of 4 165 cases were included in the model group and 3 934 cases were included in the model verification group.Finally,the risk prediction model variables included:length of stay in ICU,mechanical ventilation,and use of antibiotics.The corresponding risk prediction model scores were 2,5 and 4 points,and the scores were 7 to 9 points for high-risk groups.The H-L test P value of the model group was 0.304,and the area under the ROC curve was 0.84(95%CI:0.77~0.91,P<0.05).The H-L test P value of the prediction effect evaluation group was 0.524,and the area under the ROC curve was 0.75(95%CI:0.68~0.82,P< 0.05).Conclusion The discriminant validity and conformity degree of CRE risk prediction model are good,which can be used to actively predict the risk of CRE infection in ICU.