人口旅運動態與登革熱傳播的時空分析

Spatial-temporal Analysis of Dengue Transmission through Daily Movement of Population

黃宜庭(Yi-Tin Huang)

指導教授 ： 溫在弘

國立臺灣大學/理學院/地理環境資源學研究所/碩士(2011年)

https://doi.org/10.6342/NTU.2011.00109

摘要

目地：本研究之目的在於探討人口旅運動態對於登革熱傳播的影響，分析人口移動及地理鄰近結構的傳播特徵及評估其區域間之差異。
方法：本研究以以時間掃描將2001~2003年在高雄地區之登革熱疫情，區分成群聚初期和群聚期，進而以群聚初期為初始疫情，透過旅運矩陣與地理鄰近矩陣模擬群聚期的病例分佈，分析旅運傳播與鄰近擴散的特徵。其次，本研究利用雙空間延遲地理加權迴歸模型(Dual Spatial Lag with Geographically Weighted Regression, DSLGWR)，找出旅運傳播與鄰近擴散因子在登革熱傳播上的區域差異，這些差異形成各區域的傳播特徵。
結果：旅運矩陣的模擬結果，能夠解釋大部分群聚期病例的分佈，亦發現高雄地區的旅運結構可能具有重疊子群，且子群內部的旅運連結強度不同，而造成有些地區較快受到波及，有些則較慢，旅運連結強度較強的子群(如小港區及鳳山市)且與前期病例分佈子群重疊大的(前鎮區)會較快受到波及，旅運連結強度較低的子群(如鼓山、鹽埕與旗津區)以及與前期病例分佈子群幾乎沒有重疊(如左營與楠梓區)，則較慢波及。鄰近矩陣之模擬結果，則只能解釋環繞在群聚期前期交通區及其周邊的群集現象。雙空間延遲地理加權迴歸模型所找到的區域傳播特徵，可以概分為三類: (1) 鄰近擴散傳播為主的地區為高雄較落後或較早開發的地區，(2) 旅運傳播為主的地區為工業區與新發展的地區，(3) 兩者影響力差不多的地區，則處在高雄中央的帶狀地帶，此區人口密集且交流頻繁，新舊社區交雜。
結論：群聚期前期的地區進行接觸式之鄰近擴散傳播及利用移動的人口將登革熱逐漸傳播至較遠的社區。以區域傳播特徵搭配病例前期分佈之分析，發現高雄2002年的登革熱疫情，主要由人的移動傳播所主導。且人口越多的地區移動，傳播能力越強。由於區域傳播特徵不同，故防疫策略應要有所差異。

關鍵字

登革熱 ； 人口旅運 ； 時間掃瞄統計 ； 空間自迴歸 ； 地理加權迴歸

並列摘要

Objective: The purpose of the study is to analyze the role of population movement in the dengue transmission and to differentiate the diffusion characteristics of the epidemic due to population movement and geographic proximity.
Methods: Temporal scan statistics was used to divide into the initial and clustering periods of the dengue epidemic. We then used the disease cases during the initial period and the matrices of population travel and geographic proximity to simulate the spatial patterns of the clustering period. The dual spatial lag with geographically weighted regression was used to identify the spatial heterogeneity of diffusion characteristics due to population travel and geographic proximity.
Results: The results indicated that most of cases during the clustering period can be explained by the simulation with the population travel matrix. The areas with stronger travel connectivity (ex. Hdisoksng Precinct and Fengshan city) and with large overlap of initial period (ex. Chienchen Precinct) were diffused earlier. On the other hand, the areas with weaker travel connectivity (ex. Kushan, Yenchen, Chiching) and with almost none overlap of initial period (ex. Tsoying, Nantzu Precinct) were diffused later. The simulation with the geographic proximity matrix can only explain the clustering patterns of the areas and the surroundings where cases occurred during the initial period. The regression results indicated that the effect of geographic proximity is significant on disease diffusion in the areas with early development; effect of population travel is significant in the industrial areas and new development areas. Both effects are significant in the central belt areas with densely populated of Kaohsiung.
Conclusion: The patterns of dengue epidemic in 2001-2003 mixed the contagious and relocation diffusion due to population travel and was spread by the population movement mainly. Therefore, implementation of prevention and control strategies should take different diffusion characteristics into further consideration.

並列關鍵字

dengue ； population movement ； temporal scan statistics ； spatial autocorrelation ； geographically weighted regression

參考文獻

行政院衛生署疾病管制局 (2009). 登革熱防治工作指引. 防疫學苑系列, 行政院衛生署疾病管制局.

行政院衛生署疾病管制局 (2009). 登革熱防治工作指引.

行政院衛生署疾病管制局 (2010). 傳染病病例定義暨防疫檢體採檢送驗事項--登革熱. 衛生署疾病管制局.

行政院衛生署疾病管制局 (2010). 傳染病統計資料查詢系統.

林政宏 (2007). 台灣地區登革熱擴散之空間分析, 臺灣大學地理環境資源學研究所碩士論文.