Study On The Spatial Clustered Distribution Of Tuberculosis And Its Influential Factors In China

BackgroundTuberculosis (TB), known as "white plague", is a severe chronic disease which is caused by Mycobacterium tuberculosis.According to World Health Organization, worldwidely there are around9million new tuberculosis annualy and about3million people die of TB. China is one of the22TB high-burden countries in the world and the number of tuberculosis patient ranks2nd. Previous studies on tuberculosis and other infectious diseases were limited to simple description of incidence of the disease, ignoring the geographic correlation, not doing a study on the basis of the quantitative level of the spatial distribution of diseases. However, as an infectious disease, tuberculosis is related to the local environment, population, climate, and so on. Because of its infectivity and universality, the occurrence, development and prevalence of TB is a spatial phenomenon. Several studies showed that the spatial distribution of TB was nonrandom and clustered. Therefore, it is necessary to explore the hotspots of TB from the quantitative level with the use of spatial epidemiology and statistics of the analysis, as well as space, time and space-time cluster. This reflects dynamic distribution characteristics on space and its influential factors in order to meet the needs of TB prevention and control work.ObjectivesThe main objectives of this research:exploring the spatial distribution of TB incidence with spatial autocorrelation, space-time scan statistic and spatial regression analysis;exploring where and when the new cases happen, that is to say, exploring the space, time and spatial-temporal hot spot to know the cluster of TB,and analysis influential factors of incidence rate, aim to support the theory evidence for the policy of disease prevention and control and the reference for the same study.MethodologyIn this study,the distribution of TB incidence is described in2005-2012, based on GIS.The global spatial autocorrelation and local spatial autocorrelation were implemented in ArcGISl0.0software,in order to determine the pattern of spatial clustering. The space-time scan statistic was performed using SaTScan software. And further spatial regression analysis was realized with OpenGeoDa software, used to analysis influential factors of incidence rates.Results1. From2005-2012, a total of8693449TB cases have been reported. The incidence rates were96.31/100000,86.23/100000,88.55/100000,88.52/100000,81.09/100000,74.27/100000,71.09/100000and70.62/100000in the eight years.2. Global Moran’s/were0.360121,0.358639,0.373570,0.320409,0.362954,0.339992,0.407360and0.358179for2005-2012. The Moran’s/of eight years all had the significance.3. General G was significant in2009-2012, while not significant in2005-2008.4. According to Local Moran’s I analysis, the high-high regions were Guizhou and Guangxi in2005. The high-high regions were Guizhou, Guangxi and Hainan in2006～2007.The high-high regions were Guizhou, Guangxi and Hainan in2008.The high-high regions were Guizhou, Xinjiang and Xizang in2009～2011.The high-high regions were Xinjiang and Xizang in2012. The low-high region was Yunnan in2006～2008.The high-low region was Xinjiang in2008.5. According to Local General G analysis, the high value clustered regions were Xinjiang and Guangdong in2006. The clustered regions were Xinjiang and Xizang in2009～2012. The clustered regions were Guizhou, Hunan, Guangdong, Guangxi and Xinjiang in2005.The clustered regions were Guizhou, Yunnan, and Guangxi in2006. The clustered regions were Guizhou, Yunnan, Guangdong, Guangxi,Hainan and Xinjiang in2007. The clustered regions were Guizhou, Yunnan, Guangdong, Guangxi and Hainan in2008.6. The study obtained5high incidence regions using spatial scan analysis in2005～2012, including one most likely cluster and4secondary likely clusters. One most likely cluster had9with RR as1.39and P as less than0.01,mainly located in Guangdong, Hainan, Guangxi, Hunan, Jiangxi, Fujian, Guizhou, Hubei, and Chongqing. In2005,the study obtained4high incidence regions with statistical significance. One most likely cluster had9regions with RR as1.37and P as less than0.01,located in Jiangxi, Hunan, Hubei, Hainan, Guizhou, Guangxi, Fujian, Chongqing and Guangdong. In2005,2006,2007,2008and2012, the study all obtained2high incidence regions with statistical significance. One most likely cluster had16regions, RR are1.39,1.45,1.44,1.45and P as less than0.01, located in Yunnan, Xinjiang, Xizang, Sichuan, Shaanxi, Shanxi, Qinghai, Ningxia, Hunan, Hubei, Henan, Hainan, Guizhou, Guangxi, Gansu and Chongqing. In2009, the study obtained5high incidence regions with statistical significance. One most likely cluster had4regions with RR as1.54and P as less than0.01,located in Hainan, Guizhou, Guangxi and Guangdong. In2010, the study obtained4high incidence regions with statistical significance. One most likely cluster had4regions with RR as1.46and P as less than0.01,located in Hunan, Hainan, Guizhou, Guangxi and Guangdong, In2011, the study obtained3high incidence regions with statistical significance. One most likely cluster had14regions with RR as1.42and P as less than0.01,located in Yunnan, Xizang, Sichuan, Shaanxi, Qinghai, Jiangxi, Hunan, Hubei, Hainan, Guizhou, Guangxi, Gansu, Chongqing and Guangdong.7. The study obtained7spatial-temporal clusters with statistical significance using space-time scan analysis in2005～2012.One most likely cluster had1region with RR as1.16and P as less than0.01, located in Guangdong for the year2009～2010. Six statistically significant secondary clusters were also detected for high incidence of TB.8. Through spatial Lag Model and spatial Error Model, we find that human mortality and per thousand health technical personnel number have positive influence on the TB incidence. But the population density, per capita GDP, total health expenditure per capita and beds per thousand medical institutions have negative influence on the TB incidence. The TB incidences are higher in the regions with more human mortality and per thousand health technical personnel number. The more TB incidences, the lower the population density, per capita GDP, total health expenditure per capita and beds per thousand medical institutions are.Conclusion1. Global Moran’s I are all above0in2005～2012.It shows that in these eight years the incidences of TB are positively correlated, not random.2. General G was significant in2009～2012, while not significant in2005～2008.3. The Local Moran’s I analysis showed that the main high-high regions were Guizhou, Xinjiang, Guangxi and Hainan. More attention should be paid to these statistically significant high-high, high-low and low-high areas.4. The Local General G analysis showed that the high value clustered regions were Xinjiang, Guangdong and Xizang. The clustered regions were Guizhou, Yunnan, Guangxi, Hainan and Hunan. So these regions should be focused mainly when we control and prevent TB.5. According to the spatial scan statistics, the most likely clusters were Jiangxi, Hunan, Hubei, Hainan, Guizhou, Guangxi, Fujian, Chongqing and Guangdong in2005,Yunnan, Xinjiang, Xizang, Sichuan, Shaanxi, Shanxi, Qinghai, Ningxia, Hunan, Hubei, Henan, Hainan, Guizhou, Guangxi, Gansu and Chongqing in2005,2006,2007,2008and2012, Hainan, Guizhou, Guangxi and Guangdong in2009, Hunan, Hainan, Guizhou, Guangxi and Guangdong in2010, Yunnan, Xizang, Sichuan, Shaanxi, Qinghai, Jiangxi, Hunan, Hubei, Hainan, Guizhou, Guangxi, Gansu, Chongqing and Guangdong in2011.When compared the clusters of the spatial autocorrelation analysis with those of the space-time scan statistic, both methods detected similar and significant high-risk clustering. However, differences also existed for these two methods basing on different criteria and indicators. Space-time scan analysis showed that the most likely statistically significant cluster was Guangdong, for the year2009～2010(RR=1.16, p<0.001).Therefore, space-time scan statistic methods considering time dimension should be proposed as a dynamic supplement to spatial statistical methods.6. Through spatial Lag Model and spatial Error Model, we find that human mortality and per thousand health technical personnel number have positive influence on the TB incidence. But the population density, per capita GDP, total health expenditure per capita and beds per thousand medical institutions have negative influence on the TB incidence.