Spatial Clustering Of Pulmonary Tuberculosis Using Spatial Autocorrelation Analysis

ObjectiveTuberculosis(TB) has reemerged as a global public health epidemic in recent year. In2008, nine million new TB cases occurred in worldwide with an incidence of139per100thousands population. According to the World Health Organization(WHO) Report on Global TB Control, China is one of the22high burden countries in the world. The statistics of disease clustering is one of the most important tools for epidemilogists to detect and monitor public health disease patterns. In the recent year, several studies on geographical epidemiology of TB clustering had been published all over the world. In this study, the objectives is using spatial autocorrelation analysis to investigate the spatial patterns of the incidence of pulmonary tuberculosis, to determine where the prevalence high spatial concentration in90countries(districts),in Zhejiang province.MethodsThe data of pulmonary TB cases in90counties (districts) in Zhejiang province from2004to2006were obtained from the TB surveillance system, which monitors TB events among the roughly fifty-one million residents in Zhejiang province. The rate of missing report was2.88%in2004and1.33%in2006respectively. Based on the same county field in digital maps and pulmonary TB incidence database, digital map of Zhejiang province was interrelated with the database of Zhejiang province pulmonary TB incidence to establish Zhejiang geographic information system database. Spatial autocorrelation analysis (Moran's â…  index, Moran scatterplot. Anselin's local indication of spatial association[LISA] and Getis-Ord Gi*statistics) was developed to test for geographical clustering by ArcGIS9.2software.ResultsA total of131671pulmonary TB cases were reported across90countries from2004to2006in Zhejiang province, with the incidence ranging from29.21to569.45per100000population. There were present strong spatial autocorrelation among pulmonary TB incidence rates from2004to2006in Zhejiang province by Moran's â…  analysis and Moran scatterplot (I2004=0.6208,I2005=0.6921,I2006=0.5106, P value0.001). Also, Getis-Ord Gi*statistics indicated that there were high occurrence of pulmonary TB(Getis-Ord Gi>0, P<0.001).Anselin's local indication of spatial association[LISA] for TB incidence in2004identified that these region, Jiangdong, Jiangbei, Beilun, Yingzhou, Haishu district in Nibo city,Yuyao county, Cixi county, Fenghua county, Leqing county, Yongjia county, OUhai, Longwan district in Wenzhou city, Rui-an country, Pingyang country presented statistically significant TB incidence clustering. The spatial clustering analysis by Getis-Ord Gi*statistics identified the most likely significant clustering for high occurrence of TB in Wenzhou region in2004, and the most likely significant clustering for low occurrence of TB in Nibo region in2004.Anselin's local indication of spatial association[LISA] for TB incidence in2005identified that these region, Jiangdong, Yingzhou, Zhenhai district in Nibo city.Yuyao county, Cixi county, Xiaoshan district in Hangzhou city, Yuhuan county, Yongjia county, OUhai, Longwan, Lucheng district in Wenzhou city, Rui-an country, Pingyang country, Qingtian country presented statistically significant TB incidence clustering. The spatial clustering analysis by Getis-Ord Gi*statistics identified the most likely significant clustering for high occurrence of TB in Leqing county, Yuhuan county, Qingtian country, Dongtou county, Yongjia county, Rui-an country, Pingyang country, OUhai, Longwan, Lucheng district in Wenzhou city, Rui-an country, Pingyang country in2005, and the most likely significant clustering for low occurrence of TB in Hangzhou, Jiangxin, Nibo regions in2005.Anselin's local indication of spatial association[LISA] and Getis-Ord Gi*statistics for TB incidence suggested that there were statistically significant hotspots in Wenzhou region, Yuhuan county, Yongjia county. OUhai, Longwan, Lucheng district,Wenzhou city, and there were no statistically significant coldspots in2006.ConclusionIn summary, we used spatial autocorrelation analysis to determine clustering of pulmonary TB from2004to2006in90countries(districts) in Zhejiang province, China. Our study has shown the presence of pulmonary TB clustering. The spatial clustering analysis identified the most likely significant clustering for high occurrence of TB in Wenzhou regions, and the most likely significant clustering for low occurrence of TB in Nibo regions. The results provide an useful information on the prevention and control for pulmonary TB. But the study has some limitions with a short investigating period of time (from2004to2006) and with only the statistically significant clustering of TB. Future research should focus on the space time scan statistic and on the effect of various socio-economic and environmental factors with the high occurrence of TB.