综合重症监护病房患者多重耐药菌医院感染风险评估模型构建

Risk assessment model construction of nosocomial infection of multi-drug resistant organism in comprehensive intensive care unit patients

目的构建综合重症监护病房(intensive care unit,ICU)患者多重耐药菌医院感染的风险评估模型。方法对2017年1月—2019年12月入住综合ICU超过48 h的患者进行回顾性分析,按照3∶1的比例将患者随机分为建模组和验证组(随机种子为20200108),对建模组的数据进行危险因素识别,构建logistic回归模型,根据相关系数对各危险因素进行赋值,建立感染风险评分模型,利用受试者工作特征曲线(receiver operating characteristic curve,ROC)评价模型的预测准确度,代入验证组的数据再次进行评价。结果3年内入住ICU超过48 h的患者共950例,多重耐药菌医院感染的发病率为12.42%(118例)。将单因素分析有统计学意义(P<0.05)的14个自变量纳入二元logistic回归分析,性别、感染前综合ICU住院天数、心脑血管疾病、感染前手术次数、留置导尿管、使用呼吸机、使用营养支持为7个独立危险因素,对各变量赋值计算,进行风险分层,得分8~14分为高风险组,1~7分为中风险组,-4~0分为低风险组。建模组模型评价显示,Hosmer-Lemeshow检验P=0.32,ROC下面积为0.870,敏感度为84.3%,特异度为80.0%。结论所构建的模型对综合ICU患者多重耐药菌者医院感染风险有较好的拟合度与判别度,可用于高危人群的识别,进行早期预防与控制。

Objective To construct a risk assessment model for nosocomial infection of multi-drug resistant organism in comprehensive intensive care unit(ICU)patients.Methods Retrospective analysis was conducted on the patients who were admitted to comprehensive ICU longer than 48 hours from January 2017 to December 2019.The patients were randomly divided into the modeling group and the validation group according to the ratio of 3∶1(random seed was 20200108).Risk factor identification was carried out on the data of the modeling group.The logistic regression model was constructed,and each risk factor was assigned according to correlation coefficient.Risk score model of infection was established by the receiver operating characteristic curve(ROC)to evaluate the prediction accuracy.The data substituted into the validation group was evaluated again.Results A total of 950 patients were admitted to ICU over 48 hours in three years.The incidence of nosocomial infection of multi-drug resistant organism was 12.42%(118 cases).A total of 14 independent variables with a single factor analysis(P<0.05)were included in the binary logistic analysis.Gender,comprehensive length of stay in ICU before infection,cardiovascular and cerebrovascular diseases,number of operations before infection,indwelling catheter,use of ventilator and nutrition support were seven independent risk factors.Value of each variable was assigned and risk was stratified.The score was 8-14 for the high risk group,1-7 score for the medium risk group and-4-0 for the low risk group.Model evaluation of the modeling group showed that in hosmer-lemeshow test,P value was equal to 0.32,the area under the ROC was 0.870,the sensitivity was 84.3%,and the specificity was 80.0%.Conclusion The constructed model has a good fit and discrimination for the risk of nosocomial infection in ICU patients with multi-drug resistant organism,which can be used to identify high-risk population for the early prevention and control.