基于空间聚集性与时空扫描的肺结核流行特征分析  

A clustering study on pulmonary tuberculosis based on pure spatial clustering and spatiotemporal scanning

作　　者：李婷[1] 何金戈[1] 杨长虹[2] 李运葵[1] 王丹霞[1] 陈闯[1] 吴建林[3] LI Ting HE Jin-ge YANG Chang-hong LI Yun-kui WANG Dan-xia CHEN Chuang WU Jian-lin(Tuberculosis Preven- tion and Control Department, Sichuan Center for Disease Control and Prevention, Chengdu 610041, China)

机构地区：[1]四川省疾病预防控制中心结核病预防控制所,成都610041 [2]四川省疾病预防控制中心公共卫生信息所,成都610041 [3]四川省疾病预防控制中心办公室,成都610041

出　　处：《中国防痨杂志》2016年第12期1032-1040,共9页Chinese Journal of Antituberculosis

基　　金：四川省科技厅应用基础研究项目（20LIJY0066）

摘　　要：目的探讨2011—2015年四川省肺结核发病的时空分布特征。方法基于中国疾病预防控制信息系统传染病报告信息管理系统中四川省184个县、市、区2011—2015年肺结核疫情数据和人口数据，建立地理信息数据库，导入OpenGeoda12．0和SaTScan9．4．1软件，进行全局、局部空间自相关分析和时空扫描聚类分析，确定肺结核发病时空热点区域，通过ArcGIS10．2软件进行可视化。结果2011—2015各年度四川省肺结核报告发病率分别为77．62／10万（63040例）、76．45／10万（62325例）、74．51／10万（60781例）、68．65／10万（56485例）和67．13／10万（54645例），呈明显的聚集性分布，各年度全局空间自相关系数（Moran指数I值）分别为0．46、0．50、0．52、0．56和0．63，均具有统计学意义（P值均-0．001）。局部空间自相关分析结果表明，2011—2015年分别有18、15、15、15、14个地区处于“高-高”（HH）区域，连续5年处于HH区域的有8个，全部位于甘孜州。时空扫描分析表明，2011-2015年四川省肺结核疫情存在八级聚集区域，共涉及81个地区，对数似然比（10g-likelihoodratio，LLR）为24．65-3012．78，P值均〈0．01。其中，一级聚类区域位于2013--2014年甘孜州的16个县和阿坝州的7个县，LL尺值为3012．78，RR值为3．18，P〈0．01。结论四川省肺结核疫情存在明显的时空聚集特征，在川西高原少数民族地区、凉山州彝族地区、秦巴山区尤为显著，应实施针对性强的区域结核病防治策略和政策支持。Objective To explore the spatial and temporal distribution of pulmonary tuberculosis （PTB） in Sichuan province. Methods Geographic information data were from the incidence data of PTB and demographic data reported in the China disease prevention of infectious disease reporting information management system in 184 counties of Sichuan province from 2011 to 2015. Global indication of spatial autocorrelation （GISA）, local indication of spatial autocorrelation （LISA） and spatial-temporal clustering analysis were conducted with software OpenGeoda 1.2.0 and SaTScan 9.4.1 in order to determine high risk areas of PTB, and it could be visualized by software ArcGIS 10.2. Results From 2011 to 2015, the incidence rates of PTB in Sichuan were 77.62/100 000 （63 040 cases）, 76.45/100 000 （62 325 cases）, 74.51/100 000 （60 781 cases）, 68.65/100 000 （56 485 cases） and 67.13/100 000 （54 645 cases）, respectively. There was obvious clustering for the incidence rate in each year and the global spatial autocorrelation coefficient Moran＇s I was statistically significant （from 2011 to 2015, the values were 0.46, 0.50, 0.52, 0.56 and 0.63, respectively, all P=0. 001）. Local autocorrelation analysis showed that there were 18, 15, 15, 15 and 14 counties in the ＂high-high＂ （HH） region from 2011 to 2015, respectively. There were 8 counties all located in Ganzi in the HH region for 5 consecutive years. Spatial temporal clustering analysis showed that there were 8 clustering districts of the epidemic of PTB in Sichuan, covering a total of 81 counties, log-likelihood ratio （LLR） values ranged from 24.65 to 3012.78, all P〈0.01. The most likely clustering was in 16 counties of Ganzi and 7 counties of Aba from 2013 to 2014 （LLR = 3012.78, RR = 3.18, P〈0.01）. Conclusion Obvious spatial temporal clustering of PTB distribution was found in Sichuan, especially in ethnic minority areas in the Western Sichuan Plateau, Liangshan Yi area and Qin Ba mountain area, where TB control strategies and polic

关 键 词：结核 肺 时空聚类分析 传染病控制 

分 类 号：R181.22[医药卫生—流行病学]