基于空间点模式分析的疾病分布状态的量化指标研究  

Study on the quantified indices to describe the distributional status of diseases in the spatial point pattern analysis

作　　者：张志杰[1] 彭文祥[1] 周艺彪[1] 陈更新[2] 姜庆五[1] 

机构地区：[1]复旦大学公共卫生学院流行病学教研室公共卫生安全教育部重点实验室,上海200032 [2]安徽省池州市贵池区血吸虫病防治站

出　　处：《中华预防医学杂志》2008年第6期422-426,共5页Chinese Journal of Preventive Medicine

基　　金：国家自然科学基金（30590374）;国家“863”计划（2006AA022402）志谢 R软件Spatstat包的作者Adrian Baddeley教授和挪威经济与商业管理学院（NHH）的Roger Bivand教授对4个指标分析程序的编写给予了很大的帮助

摘　　要：目的从空间点模式分析的角度探讨疾病分布状态的量化指标，为疾病的预防控制提供新的统计学方法。方法基于病例间的距离系统地总结了空间点模式分析中反映疾病分布状态的定量统计指标——G函数、F函数、J函数和K函数。在介绍其基本原理的基础上，研究距离取0到3000m，间隔为50m，使用4个指标对贵池区的急性血吸虫病病例资料进行分析，并在SaTScan软件中用空间移动扫描圆形窗口法进行分析验证。结果贵池区6年共发生83例急性血吸虫病病例，空间点模式分布图显示病例主要分布在长江和秋浦河附近。获得了4个定量统计指标的计算方法，用其对上述急性血吸虫病资料分析发现病例分布具有空间聚集性（G函数和K函数位于95％可信区间的上方，F函数和．J函数位于95％可信区间的下方），与空间移动扫描圆形窗口法的结果一致，后者并发现1个最可能的聚集区域，圆心坐标为（30．65N，117．44E），圆半径为2．69km，相对危险度为12．78（对数似然比=32．80，P=0．0001）。结论量化统计指标的应用不仅解决了传统标点地图不能量化分析的缺点，而且为深入的空间聚集性分析提供了理论依据。Objective To study the quantified indices for describing the distributional status of diseases in the spatial point pattern analysis, and provide the a statistic in disease prevention and control. Methods G function, F function, J function and K function were summarized based on the inter-case distances from the view of spatial point pattern analysis. Through the introduction of the basic principles, these were used to analyze the data of acute schistosomiasis in the Guichi District, Chizhou City, Anhui province,with the study distances being from 0 to 3000 meters with 50-meter intervals. The findings were also validated by means of spatial moving scan window performed in SaTScan software. Results A total of 83 cases of acute schistosomiasis identified in Guichi District, and the point map showed that these cases were mainly distributed around the Yangtze and Qiupu rivers. The computational methods and characteristics of the four quantified indices were obtained. These acute schistosomiasis cases were also explored by using these indices, and the results showed that G and K functions were above 95 % confidence interval. While,F and J functions were below 95% confidence interval. All these four indices showed that spatial clustering existed in the acute cases, which was consistent with the results of spatial moving scan window method. The latter method also found a most likely cluster, the coordinate of the circle center is （30. 65 N, 117.44 E）, radius is 2. 69 km, and relative risk is 12. 78 （LLR = 32. 80, P = 0. 0001 ）. Conclusion The quantified indices to describe the distributional status of diseases have not only solved the obstacle that spatial point pattern map which could only be analyzed qualitatively, but also supplied a theoretical foundation to deepen spatial clustering analysis.

关 键 词：统计学 流行病学 聚集性分析 空间点模式分析 

分 类 号：R686[医药卫生—骨科学]