ICU患者多重耐药菌感染危险因素及列线图模型构建

Risk factors and nomogram model construction of multi-drug resistance organism(MDRO)infection in ICU patients

目的分析ICU患者感染多重耐药菌(Multi-drug resistance organism,MDRO)的危险因素,并构建列线图模型。方法回顾性分析新疆医科大学第一附属医院2020年1月-2023年12月入住ICU的383例MDRO感染患者临床资料,按照1︰1比例随机选取本院同时期入住ICU的非MDRO感染患者383例的临床资料作对照。通过Logistic回归分析确立独立危险因素,建立模型方程,使用R软件绘制列线图模型。结果单因素分析结果显示,MDRO感染患者与非MDRO感染患者入院类型、手术、简化急性生理学评分Ⅱ(SAPSⅡ)评分、急性生理与慢性健康评分Ⅱ(APACHEⅡ)评分、骨科疾病、使用激素类药物、气管切开、机械通气、中心静脉置管、抗生素联用、抗生素使用时间、白介素-6水平比较,差异均有统计学意义(P<0.05)。多因素分析结果显示,抗生素使用时间、使用激素类药物、入院类型、手术、SAPSⅡ评分、气管切开是ICU患者MDRO感染的独立危险因素(P<0.05)。基于以上6种因素构建列线图模型,经Hosmer-Lemeshow检验,χ^(2)=4.857,P=0.773;模型校准曲线与理想曲线吻合良好;受试者工作特征(ROC)曲线显示6种因素联合预测ICU患者MDRO感染的曲线下面积(Area under curve,AUC)为0.829[95%CI:(0.800~0.858)],最佳截断值为0.548,灵敏度为0.708,特异度为0.841。结论本研究构建的列线图模型拥有良好的临床实用价值,可为临床医务人员早期识别ICU患者MDRO感染风险及制定相应的防控措施提供参考。

Objective:The risk factors of multi-drug resistance organism(MDRO)infection in ICU patients were analyzed,and a nomogram model was constructed.Methods:The clinical data of 383 MDRO infected patients admitted to ICU in the hospital from January 2020 to December 2023 were retrospectively analyzed,and the clinical data of 383 non-MDRO infected patients admitted to ICU in the same period were randomly selected as controls according to a ratio of 1∶1.The model equation was constructed by Logistic regression analysis,and the nomogram model was drawn by R software.Results:Single factor analysis showed that admission type,surgery,simplified acute physiology scoreⅡ(SAPSⅡ),acute physiology and chronic health scoreⅡ(APACHEⅡ),orthopedic disease,hormone drug use,tracheotomy,mechanical ventilation,central venous catheterization,antibiotic combination,the time of antibiotic use and interleukin-6,the differences were statistically significant(P<0.05).The results of multivariate analysis showed that the time of antibiotic use,hormone drug use,damission type,operation,SAPSⅡscore and tracheotomy were independent risk factors for MDRO infection in ICU patients(P<0.05).A nomogram model was constructed based on the above 6 factors,and tested by Hosmer-Lemeshow,χ^(2)=4.857,P=0.773;The model calibration curve was in good agreement with the ideal curve.The receiver operating characteristic(ROC)curve showed that the area under curve(AUC)of 6 factors combined to predict MDRO infection in ICU patients was 0.829[95%CI:(0.800~0.858)],the optimal cut-off value was 0.548,the sensitivity was 0.708,and the specificity was 0.841.Conclusion:The nomogram model constructed in this study has good clinical practical value,which can provide reference for clinical medical personnel to identify the risk of MDRO infection in ICU patients and formulate corresponding prevention and control measures.