基于移动百分位数法对渭南市5种传染病的预警研究  

A study on early detection for five communicable diseases by moving percentile method in Weinan City

作　　者：苗美荣[1,5] 周体操 王林江[5] 张艳[5] 樊苏红 陈翠萍 刘玮[4] MIAO Mei-rong;ZHOU Ti-cao;WANG Lin-jiang;ZHANG Yan;FAN Su-hong;CHEN Cui-ping;LIU Wei(Department of Health Monitoring and Occupational Diseases,Weinan Center for Disease Control and Prevention,Weinan Shaanxi,714000,China;Department of Epidemiology,Weinan Center for Disease Control and Prevention,Weinan Shaanxi,714000,China;Department of Epidemic Information,Weinan Center for Disease Control and Prevention,Weinan Shaanxi,714000,China;Office,Weinan Center for Disease Control and Prevention,Weinan Shaanxi,714000,China;Shaanxi Field Epidemiology Training Program,Shaanxi Provincial Center for Disease Control and Prevention,Xi'an Shaanxi,710054,China)

机构地区：[1]渭南市疾病预防控制中心卫生监测与职业病科,陕西渭南714000 [2]渭南市疾病预防控制中心流病科,陕西渭南714000 [3]渭南市疾病预防控制中心疫情信息科,陕西渭南714000 [4]渭南市疾病预防控制中心办公室,陕西渭南714000 [5]陕西省疾病预防控制中心陕西现场流行病学培训项目,陕西西安710054

出　　处：《职业与健康》2020年第4期521-525,529,共6页Occupation and Health

基　　金：陕西省科研基金(2016D093)。

摘　　要：目的应用移动百分位数法进行传染病预警阈值筛选,对5种传染病开展基于渭南市本地疫情的预警研究,使传染病预警系统的应用更高效更灵活。方法利用2012-2016年渭南市5种传染病报告数据创建预警数据库,2017年数据作为验证。选用移动百分位数法建立预警模板,通过比较灵敏度、特异度、阳性预测值、阴性预测值、约登指数并绘制受试者工作特征曲线(ROC曲线),筛选出渭南市5种传染病适合的预警临界值。结果渭南市5种传染病合适预警阈值为:其他感染性腹泻P90、流行性感冒和水痘P85、流行性出血热P75、丙肝P95;预警灵敏度均达到100%,除水痘预警特异度为53%外,其他4种传染病预警特异度均>85%。结论为提高预警系统灵敏度,渭南市其他感染性腹泻、流行性感冒的预警阈值在默认阈值基础上应稍提高,而流行性出血热的预警临界值应稍降低。非预警病种丙肝和水痘,模拟预警模式可考虑时间模型-移动百分位数法。传染病预警临界值应根据不同地区的疫情信息定期进行优化筛选,使传染病预警系统的应用更加灵活高效,适宜于当地的疫情状况。渭南市高发传染病(高于全国平均水平)的预警阈值可适当低于国家平均阈值,提高预警的灵敏度,利于早期识别及控制,降低发病率。Objective To study the premium alert threshold for five kinds of communicable diseases by using the moving percentile alert technique based on the disease information of Weinan,to make the application of infectious disease early warning information system to be more efficient and more flexible. Methods The early warning database was established by using the report data of five communicable diseases in Weinan City from 2012-2016,and was verified by the data in 2017. The moving percentile method was used to create alert template. The sensitivity,specificity,positive predictive value,negative predictive value, Jordan index were compared, and the receiver operating characteristic curve(ROC curve) was drawn,to screen out the premium critical value of early warning for five communicable diseases in Weinan City. Results The appropriate early warning thresholds for five communicable diseases in Weinan City were as follows: other infectious diarrhea was P90,influenza and chicken pox was P85,epidemic hemorrhagic fever was P75 and hepatitis C was P95. All the sensitivity reached 100%. The specificities of four communicable diseases were over 85% except specificity of chicken pox was 53%. Conclusion The warning threshold of other infectious diarrhea and influenza in Weinan City should be slightly increased on the basis of the default threshold to improve the specificity of the warning system,while critical value of early warning of epidemic hemorrhagic fever should be slightly reduced to improve the sensitivity of early warning system. The time model moving percentile method can be considered in the selection of early warning model for non early warning diseases such as hepatitis C and varicella. Early warning threshold of infectious diseases should be set periodically based on the local epidemic information,to make the application of infectious disease early warning information system more flexible and efficient,and suitable for local epidemic situation. It is suggested that the early warning threshold of local high-risk in

关 键 词：传染病 移动百分位数法 预警 阈值 优选 

分 类 号：R181.3[医药卫生—流行病学]