摘要
目的研究移动百分位数控制图法(简称控制图法)在北京市西城区3种重点传染病中的预警效果,为传染病预警工作提供参考和借鉴。方法收集2011-2017年北京市西城区3种重点传染病的周发病数据,建立以观察周和前后各2周(A方法)或其前后各1周(B方法)作为基线数据的两种控制图模型,参考"流行"标准,通过比较灵敏度、特异度、约登指数和受试者工作特征曲线(ROC曲线),选出最优预警模型和预警界值。结果对北京市西城区的其他感染性腹泻、手足口病和水痘进行预警测试,控制图A方法的最优预警界值分别为P_(85)、P_(65)和P_(80),ROC曲线下面积(AUC值)分别为0.998、0.919和0.982;控制图B方法的最优预警界值分别为P_(85)、P_(70)和P_(80),AUC值分别为0.996、0.932和0.977,同病种两种控制图法的AUC值差异有统计学意义(P<0.05)。结论控制图法具有简单实用和预警功效高的特点,可以有效应用于西城区3种重点传染病的预警工作。
Objective To evaluate the early warning effects of the mobile percentile control chart method(Control chart method) in predicting three kinds of major infectious diseases so as to provide reference for early warning against the infectious diseases.Methods Weekly incidence data of three infectious diseases in Xicheng District of Beijing from 2011 to 2017 were collected for modeling, two early warning models were established on the basis of observation week and each swing for two weeks(method A) or one week(method B) according to the principles of Control chart method, the optimal model and thresholds were selected through comparison of sensitivity,specificity, Youden’ s index and receiver operating characteristic curve(ROC) with reference to the "prevalence" criteria. Result Early warning tests for other infectious diarrhea,hand-foot-mouth disease and varicella showed the alert thresholds of method A were P85,P65 and P80,the values of area under the ROC curve(AUC) were 0.998,0.919 and 0.982,respectively. The alert thresholds of method B were P85,P70 and P80, the values of AUC were 0.996, 0.932 and 0.977, respectively. The test results of the AUC of method A and the method B for the same disease were statistically significantly different(P<0.05).Conclusion Control chart method has the characteristics of simple and high early warning efficiency and can be used for early warning against three major infectious diseases in Xicheng district of Beijing.
作者
崔策
张燕
宋静
CUI Ce;ZHANG Yan;SONG Jing(Xicheng District Center for Disease Control and Prevention,Beijing 100120,China)
出处
《预防医学情报杂志》
CAS
2019年第5期446-450,共5页
Journal of Preventive Medicine Information
关键词
预警模型
移动百分位数控制图法
受试者工作特征曲线
rarly warning model
mobile percentile control chart method
receiver operating characteristic curve
