摘要
目的构建ICU患者多重耐药菌感染风险预测模型,帮助早期筛查高风险患者。方法采用回顾性研究方法选取2017年10月至2019年4月入住ICU的多重耐药菌感染患者共606例,并按照1∶1比例随机选取同时期内非多重耐药菌感染患者606例。通过Logistic回归分析构建预测模型并绘制模型的列线图。收集2019年5月入住综合ICU的265例患者进行外部验证。结果最终纳入入院方式(OR=6.420)、手术(OR=3.385)、抗生素使用>4 d(OR=1.398)、使用糖皮质激素(OR=2.091)、原发性肺部感染(OR=6.847)、低蛋白血症(OR=1.383)进入Logistic回归方程。评价模型效果:ROC曲线下面积为0.849,灵敏度85.8%,特异度71.6%,约登指数为0.574,所对应的最佳临界值为0.465;模型校准度曲线与参考线吻合较好。模型实际应用正确率为82.26%。结论本研究构建的预测模型对ICU患者多重耐药菌感染风险预测具有较好准确度及区分度,可早期识别高风险患者并实施感染防控。
Objective To construct the risk prediction model for multi-drug resistance organism(MDRO)infection of ICU patients and help early screening of high-risk patients.Method Choose 606 patients infected by MDRO admmitted in ICU from Oct.2017 to Apr.2019 by retrospective study.Choose 606 patients infected by other organisms at the same period randomly according to the ratio of 1∶1.Construct the prediction model by Logistic regression analysis and draw the nomogram.Collect 265 patients admmitted in comprehensive ICU in May 2019 and perform external validation.Result Admission pattern(OR=6.420),surgery(OR=3.385),using antibiotics>4 d(OR=1.398),using glucocorticoids(OR=2.091),primary pulmonary infection(OR=6.847)and hypoalbuminemia(OR=1.383)are included into Logistic regression equation finally.Evaluation on effect of model:the area under ROC curve is 0.849,the sensitivity is 85.8%,the specificity is 71.6%,the Youden index is 0.574,the corresponding optimal cut-off point is 0.465.The calibration curve is close to the reference line of the model.The correct rate of model in practical application is 82.26%.Conclusion The prediction model constructed in this research has a nice accuracy and discrimination on predicting MDRO infection of ICU patients,which can recognize high-risk patients in early stage and help perform infection prevention and control.
作者
李茜
王丽竹
朱祎容
邵清
向艳
郭晶晶
Li Qian;Wang Lizhu;Zhu Yirong;Shao Qing;Xiang Yan;Guo Jingjing(The Second Affiliated Hospital Zhejiang University School of Medicine,Hangzhou Zhejiang 310009,China)
出处
《护理与康复》
2023年第2期29-34,共6页
Journal of Nursing and Rehabilitation
基金
浙江省医药卫生科技计划项目(面上项目),编号2020KY147。
