Sun Jian-wei, Xu Bian-li, Chen Hao-min, Kang Kai, Huang Li-li
Henan Provincial Center for Disease Control and Prevention, Zhengzhou, China.
Zhonghua Liu Xing Bing Xue Za Zhi. 2011 Oct;32(10):1026-9.
To demonstrate and evaluate five different methods in the determination of temporal clustering on infectious diseases. The incidence rates of bacillary dysentery in Jinshui district, Zhengzhou city, Henan province from 2008 to 2010 were analyzed by 5 different methods-Cluster Analysis, Runs Test, Negative Binomial Distribution, Circular Distribution and Concentration Ratio. Through Cluster Analysis, data showed that the epidemic period was from May to Sept. with August as the peak. Runs Test confirmed a cluster of month-incidence in 2008 and 2009 (P < 0.05) and a random distribution in 2010. The Concentration Ratio showed a weakened seasonal incidence cluster to a certain extent by M from 2008 to 2010. The Circular Distribution demonstrated an inclining cluster of time (P < 0.01) and it was on July 11(th) and 29(th), as well on August 24(th) in 2008, 2009 and 2010. In terms of day-incidence, the Negative Binomial Distribution presented a cluster in 2008 and 2010, but with no significant difference in 2009. The five above said methods could flexibly be used in determining the temporal clustering of infectious disease at different occasions.
为了演示和评估五种不同的方法来确定传染病的时间聚集性。采用聚类分析、游程检验、负二项分布、圆形分布和集中度比五种不同方法,对河南省郑州市金水区2008年至2010年细菌性痢疾的发病率进行了分析。通过聚类分析,数据显示流行期为5月至9月,8月为高峰。游程检验证实2008年和2009年的月发病率存在聚集性(P<0.05),2010年呈随机分布。集中度比显示2008年至2010年季节性发病聚集性在一定程度上有所减弱。圆形分布显示时间呈倾斜聚集性(P<0.01),在2008年、2009年和2010年的7月11日和29日以及8月24日。就日发病率而言,负二项分布显示2008年和2010年存在聚集性,但2009年无显著差异。上述五种方法可灵活用于不同场合传染病时间聚集性的确定。