应用累积和模型探测北京市猩红热流行起始时间的预警分析

Early detection on the onset of scarlet fever epidemics in Beijing, using the Cumulative Sum

应用2005--2011年北京市猩红热报告病例数据，探讨累积和（CUSUM）模型在探测猩红热流行起始时间的应用价值。采用ExCel软件建立C1-MILD（C1）、C2-MEDIUM（C2）和C3-ULTRA（C3）模型，比较不同参数组合下C1～C3模型的Youden（YD）指数和检出时间（DT），筛选出各模型的最优参数组合和最佳模型，应用2011年猩红热报告病例数验证在最优参数组合下C1～C3模型的预警效果。结果表明，C1的最优参数组合为k=0．5，H=20；C2的最优参数组合为k=0．7，H=20；C3的最优参数组合为k=1．1，H=20。在最优参数组合下，C1的YD指数为83．0％，DT平均值为0．64周；C2的YD指数为85．4％，DT平均值为1．27周；C3的YD指数为85．1％，DT平均值为1．36周。三种模型中C1的预警功效最好。C1～C3模型均在猩红热流行开始4周内发出预警信号。表明CUSUM模型在探测猩红热流行起始时间时具有较好的功效。

Based on data related to scarlet fever which was collected from the Disease Surveillance Information Reporting System in Beijing from 2005 to 2011, to explore the efficiency of Cumulative Sum （CUSUM） in detecting the onset of scarlet fever epidcmics. Models as C1-MILD （C 1 ）, C2-MED1UM （C2） and C3-ULTRA （C3） were used. Tools for evaluation as Youden＇ s index and detection time were calculated to optimize the parameters and optimal model. Data on 2011 scarlet fever surveillance was used to verify the efficacy of these models. C1 （k=0.5, H=2σ）, C2 （k=0.7, H=2σ）, C3 （k= 1.1, H=2σ） appeared to be the optima/parameters among these models. Youden＇ s index of C1 was 83.0% and detection time being 0.64 weeks, Youden＇ s index of C2 was 85.4% and detection time being 1.27 weeks, Youden＇ s index of C1 was 85.1% and detection time being 1.36 weeks. Among the three early warning detection models, C1 had the highest efficacy. Three models all triggered the signals within 4 weeks after the onset of scarlet fever epidemics. The early warning detection model of CUSUM could be used to detect the onset of scarlet fever epidemics, with good efficacy.