摘要
目的构建神经外科医院感染(院感)风险的预警模型并用于医院感染监测软件,为筛查高风险人群提供依据。方法收集某三甲医院2018年1月至2021年6月神经外科病例资料,按7∶3随机分建模组(2597例)和验证组(1113例),分析院感的影响因素,建立风险预测和评分模型进行实时预警。结果共收集神经外科3710例,院感率11.6%。建模组38个因素中院感率较高的5种因素是住院天数≥15 d、碳青霉烯类、呼吸机使用8~14 d、用药≥4种和气管切开;多因素logistic回归分析显示,住院天数(X_(1))≥15 d(OR=9.89)、发热(X_(3),OR=4.57)、喹诺酮类(X_(4))(OR=1.93)、限制级用药(X_(5),OR=2.61)和用药种类(X6)≥4种(OR=2.74),是院感的独立危险因素,而脑出血(X_(2))的OR=0.47;预警模型:Y=-5.01+2.29X_(1)-0.75X_(2)+1.52X_(3)+0.66X_(4)+0.96X_(5)+1.01X6,模型拟合度较好。当危险预警值≥6分者为院感高风险患者;模型的预测效果较好(ROC曲线下面积为0.805)。结论按神经外科特点建立院感风险预警模型,与院感信息系统相结合,能尽早预警出高风险患者,将感控关口前移。
Objective To build an risk warning model for nosocomial infection in neurosurgery department and apply it in nosocomial infection monitoring software,and to provide evidence for high-risk population screening.Methods Data of neurosurgery cases were collected in a Grade-A tertiary hospital between January 2018 and June 2021.Patients were randomized into modeling group(2597 cases)and verification group(1113 cases)based on a ratio of 7∶3.Influencing factors of nosocomial infection were identified,which were then used to establish risk prediction and scoring model for the purpose of real-time warning prediction.Results A total of 3710 cases were collected in neurosurgery department.The incidence rate of nosocomial infection was 11.6%.Among 38 factors in the modeling group,five factors were associated with high incidence rate of nosocomial infection,including hospital stay≥15 days,use of carbapenems,respirator use for 8-14 days,use≥4 kinds of drugs and tracheotomy.Multivariate logistic regression analysis showed that length of hospital stay(X_(1))≥15 days(OR=9.89),fever(X_(3),OR=4.57),use of quinolones(X_(4))(OR=1.93),use of restricted grade medication(X_(5),OR=2.61)and use≥4 kinds of drugs(X6)(OR=2.74)were independent risk factors for nosocomial infection,while the OR was 0.47 for cerebral hemorrhage(X_(2)).Early warning model:Y=-5.01+2.29X_(1)-0.75X_(2)+1.52X_(3)+0.66X_(4)+0.96X_(5)+1.01X6,showed a good level of model fit.High risk population of nosocomial infection was defined by having risk warning score≥6 points.The prediction ability of warning model was good(area under ROC curve was 0.805).Conclusion Risk warning model of nosocomial infection can be established based on the characteristics of neurosurgery department,in combination with nosocomial infection information system.Model may help to identify high risk patients timely and therefore could make infection control earlier.
作者
黄菊
肖瑜
杨坚娥
刘慧楹
胡洁芬
HUANG Ju;XIAO Yu;YANG Jiane;LIU Huiying;HU Jiefen(Zhongshan Affiliated Chinese Medicine Hospital of Guangzhou University of Chinese Medicine,Zhongshan,Guangdong 528401,China)
出处
《海峡预防医学杂志》
CAS
2022年第5期5-8,共4页
Strait Journal of Preventive Medicine
关键词
医院感染
神经外科
院感监测
院感预警
统计预测
Nosocomial Infection
Neurosurgery
Nosocomial Monitoring
Nosocomial Warning
Statistical Forecast
