2008--2013年温州市肺结核疫情时空流行病学分析

Epidemiological analysis on the spatiotemporal clustering of pulmonary tuberculosis in Wenzhou from 2008 to 2013

目的分析2008--2013年温州市肺结核疫情的时空分布特征，发现高发聚集区域，探讨聚集原因。方法根据“中国疾病预防控制信息系统”子系统“结核病管理信息系统”2008--2013年温州市肺结核“患者病案管理”模块的患者基本信息，以乡镇级为单位，用全局空间自相关和SaTScan9．3软件对患者登记率进行时间、空间聚集性分析，通过ArcGIS10．1软件呈现肺结核时空聚集区域。结果2008--2013年温州市肺结核登记率从2008年78．23／10万（共登记7136例）降至2013年57．74／10万（共登记5267例）；全局空间自相关和时空扫描分析发现，温州市肺结核疫情存在明显聚集现象，其中城市核心区55个乡镇（街道）为肺结核疫情主要聚集区域，单纯空间扫描分析聚集半径为30．2km，实际病例数为21284例，RR=1．46，LLR=636．96，P〈0．01；时空扫描分析聚集时间为2008--2010年，实际病例数为11580例，RR=1．41，LLR=454．50，P〈0．01；西部部分农村区域为次要聚集区域，分布于5个较小区域，LLR值在283．96至11．57不等，P值均〈0．01。结论时空扫描分析方法可以很好地应用于肺结核疫情高发重点区域分析，结合地理信息系统，能够更加直观、全面地展示疫情聚集区域。

Objective To analyse the spatiotemporai distribution of pulmonary tuberculosis （PTB） cases in Wenzhou from 2008 to 2013, identify the clusters,clarify the related reasons. Methods The main patients informationin tuberculosis case management block was collected through ＂tuberculosis management information system＂ affiliated to the disease prevention and control information system of China. Global spatial autocorrelation and SaTScan 9.3 applications were used to detect and analyse the spatiotemporai clustering of PTB registered rate, at the town level, from 2008 to 2013. The spatiotemporal clustering areas were displayed by ArcGIS 10. 1. Results The PTB incidence rate in Wenzhou decreased from 78. 23 per 100 000 （7136 cases） in 2008 to 57.74 per 100 000 （5267 cases） in 2013. Global spatial autocorrelation with SaTScan showed that there were the obvious clusterings for PTB. 55 towns locating in the central urban district were the main clustering areas for PTB. The pure spatial scan clusterin gradius was 30.2 kin. The true registered number of cases was 21 284, RR= 1.46, LLR= 636.96, P〈 0.01. The spatiotemporal scan aggregating time was from 2008 to 2010. The true registered number of cases was 11 580, LLR= 454. 50, RR= 1.41, P〈0. 01. These condaryclustering areas were 5 small districts in the rural district in westernareas, LLR was variable from 283.96 to 11.57, P〈0. 01. Conclusion The Spatiotemporal autocorrelation and SaTScan methods may serve as efficient tools to detect the clusterings and geospatial hot spots of PTB incidence, and can show PTB clustering areas directly and compreventively, if combined with Geographic Information System.