方啟泰FANG, CHI-TAI臺灣大學:流行病學研究所吳彥珍NG, IN-CHANIN-CHANNG2010-05-052018-06-292010-05-052018-06-292009U0001-2907200902310300http://ntur.lib.ntu.edu.tw//handle/246246/180867背景核病在全球是一個很重要的健康問題。雖然全球的結核病發生率在2003年已到達頂峰，但因著人口成長的關係，總死亡個案及新發個案仍持續在上升。在台灣結核病是眾多傳染病中通報個案最多的一個疾病。過去的觀念一直認為大部份的結核病是由於體內的結核菌再活化而導致疾病，因此疫情調查及接觸者調查均被視為較不重要的防治方法。事實上，在感染結核菌的起初兩年其發病危險性是最高的；另外，從基因定型結果也發現有一部份的結核病人檢體屬於同一型菌株。這些都意味著結核病的發生很有可能是由於近期傳播而非再活化。因此，早期發現結核病人對於阻斷傳染便扮演著很重要的角色；而對於特定地區民眾的篩檢會是一個達到早期發現並具有較高成本效益的方法。究目的用空間統計的方法找出在台灣是否有近期傳播的存在以及近期傳播在哪些地方發生。法用空間延遲模型(spatial lag model)探討結核病累積發生率在鄉鎮層級是否有空間相依性(spatial dependency)。若控制了其他變項後，空間延遲變項(spatial lag variable)仍然顯著則代表在鄉鎮層級近期傳播是存在的。另外，使用最近鄰近層級分析法(Nearest Neighbor Hierarchical Clustering) 及空間掃瞄統計(Spatial Scan Statistics)分析點資料以找出台北地區是否有結核病的空間群聚。在校正總人口及老年人口後仍發現有空間群聚的地方則很有可能是近期傳播發生的地方。之後再對於空間群聚中的病人檢體進行間距寡核酸分型(Spoligotyping)及MIRU-VNTR 基因定型。果空間延遲模型(spatial lag model)中，控制了其他變項後空間延遲變項(spatial lag variable)仍然顯著，這代表鄰近地方的結核病累積發生率對當地的發生率有很重要的影響。在一個全域的範圍來看，近期傳播在台灣是存在的。在校正總人口及老年人口後仍發現有空間群聚，則拒絕了結核病是由再活化而導致的虛無假設。有空間群聚的地方很有可能有近期傳播發生，基因定型的結果也發現在同一群聚中有相同基因型的結核病人。論論用鄉鎮層級及點資料的分析都指出在台灣社區中有近期傳播的發生。結合空間分析及基因定型的方法能夠有效找出發生近期傳播的特定區域。Backgrounduberculosis (TB) remains a major global health problem. Although the global TB incidence peaked in 2003, total number of deaths and cases are still rising due to population growth. In Taiwan, the number of TB reported case is highest among all infectious diseases. Previously it was believed that reactivation was the major mechanism for tuberculosis development, and so epidemiological research and contact investigation were regarded as less important control measures. However, from the fact that the risk of becoming active disease is the highest in the first two years after infection, and from genotyping results that there is a substantial proportion of clustered isolates; these imply that incidence of TB could be due to recent transmission rather than reactivation. In this way, early detection would be important to block transmission and targeted local area screening would be a more cost-effective method to achieve it.tudy Aimy means of spatial statistic methods, we would like to find out if recent transmission exists in Taiwan as well as where recent transmission takes place. ethodse explored if there was spatial dependency of TB cumulative incidence in the township level by spatial lag regression model. Significance of the spatial lag variable after controlling other variables inferred recent transmission in the township level. Using point data, we also detected TB spatial clusters in Taipei by means of Nearest Neighbor Hierarchical Clustering (NNH) and Spatial Scan Statistics. Recent transmission might take place where clusters could still be detected after adjusting for total population and elderly population. Spoligotyping and MIRU-VNTR assay were used to differentiate isolates of M. tuberculosis for the TB cases in the clusters.esultsn the spatial lag model, spatial lag parameters were significant after adjusting for other variables. TB cumulative incidence of neighboring townships had substantial effect on that of itself which implied that recent transmission existed on a global scale. By means of cluster detection method, after adjusting for total population and elderly population, clusters could still be found, thus rejecting the hypothesis that active TB cases were originated from reactivation of latent TB. Recent transmission took place where the clusters were found. Genotyping result reinforced this interpretation.onclusionsoth township-level analysis and point-data analysis strongly indicated that recent TB transmission occurred at community in Taiwan. Using spatial analysis and genotyping techniques can be an effective method for identifying discrete geographic areas in which on-going TB transmission is likely to occur.目 錄試委員會審定書………………………………………………………………… i言…………………………………………………………………………………ii文摘要……………………………………………………………………………iii文摘要……………………………………………………………………………vhapter 1 Introduction.1 Background……………………………………………………………………1.2 Study Aim………………………………………………………………………4hapter 2 Literature Review .1 Tuberculosis (TB) .1.1 Global Burden of Tuberculosis……………………………………………6.1.2 Tuberculosis in Taiwan……………………………………………………7.1.3 Natural History of Tuberculosis …………………………………………9.1.4 Recent Transmission and Reactivation……………………………………10.1.5 Factors related to Tuberculosis infection…………………………………11.2 Tuberculosis Control.2.1 BCG vaccination……………………………………………………………12.2.2 Treatment of Latent TB infection (LTBI) ………………………………13.2.3 DOTS Strategy……………………………………………………………14.2.4 Contribution and Limitation of Genotyping and Contact Investigation …………………………………………………………………………15.3 Spatial Pattern Analysis and Spatial Regression Analysis .3.1 Brief account on Geographic Information System (GIS) and Spatial Analysis…………………………………………………………18.3.2 Application of Spatial Analysis in tuberculosis……………………………19hapter 3. Materials and Methods.1 Data Source………………………………………………………………24.2 Spatial Regression analyses.2.1 Spatial Autocorrelation…………………………………………………27.2.2 Spatial Lag Model………………………………………………………30.3 Cluster Detection Methods .3.1 Data format…………………………………………………………………32.3.2 Nearest Neighbor Hierarchical Clustering (NNH) ………………………33.3.3 Spatial Scan Statistic ………………………………………………………34.4 Genotyping techniques…………………………………………………………35hapter 4 Results.1 Descriptive statistics of regression variables and stepwise regression………36.2 Important variables found in Lag model………………………………………37.3 Spatial Tuberculosis Patterns……………………………………………………40.4 Cluster Detection and Genotyping result………………………………………40hapter 5 Discussions.1 Existence of Recent Transmission………………………………………………43.2 Factors associated with Tuberculosis incidence…………………………………44.3 Usage of different parameters in cluster detection methods……………………45.4 Strengths and Weaknesses of different cluster detection methods………………50.5 Limitations of this study…………………………………………………………52.6 Conclusions………………………………………………………………………53cknowledgements…………………………………………………………………55eferences…………………………………………………………………………56　目　錄igure 1. Moran’s I statistic for ln(TB_INCI) and regression residuals …………70igure 2. Moran’s I statistic for ln(TB_INCI_6) and regression residuals………70igure 3. Spatial distribution of dependent and independent variables……………71igure 4. Chloropleth map and histogram of Spatial Multiplier…………………72igure 5. Spatial Distribution of TB cases within NTUH 10km buffer area………72igure 6. Mean center and standard deviation of TB cases………………………73igure 7. Kernel density map for TB cases (warm hue disclosing higher density)73igure 8. Clusters detected from NNH, RNNH and Spatial Scan Statistics (10km buffer from NTUH)………………………………………………74igure 9. The clusters of which genotyping results of cases were obtained ………75igure 10. Clusters detected from NNH, RNNH and Spatial Scan Statistics (15km buffer from NTUH)……………………………………………76igure 11. Clusters detected from NNH & RNNH for different number of minimum points…………………………………………………………77igure 12. Scatter plot for population and no. of control in township level………77　目　錄able 1. Source of information used in regression models………………………78able 2. Descriptive statistics of variables and univariate regression results……79able 3. Correlation matrix showing Pearson’s correlation coefficient……………80able 4. Multiple regression results………………………………………………81able 5. Result of spoligotyping and MIRU-VNTR of TB patients………………82application/pdf1506045 bytesapplication/pdfen-US結核病近期傳播空間分析地理資訊系統空間群聚偵測tuberculosisrecent transmissionspatial analysisgeographic information system (GIS)spatial cluster detection[SDGs]SDG3thesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/180867/1/ntu-98-R96842013-1.pdf