以空間分析探討台灣高雄市之堪薩斯分枝桿菌感染人群之熱點分佈

Spatial cluster analysis of Mycobacterium kansasii infection in Kaohsiung, Taiwan

劉柏辰(Bo-Chen Liu)

指導教授 ： 林先和

國立臺灣大學/公共衛生學院/流行病學與預防醫學研究所/碩士(2020年)

https://doi.org/10.6342/NTU202003946

摘要

背景：過去即有研究指出非結核分枝桿菌感染在地理空間上的分布應非均質的。然而過去的研究方法不僅受限於資料的地理層級，所使用的資料卻也將所有非結核分枝桿菌的感染彙整在一起分析。由於高雄確診的堪薩斯分枝桿菌感染人數急劇地上升，因此我們將針對堪薩斯分枝桿菌，以空間分析了解有關此細菌感染在高雄的熱點分佈。
方法：此研究使用2015年至2017年間，高雄四家大型醫院之堪薩斯分枝桿菌資料。所有觀察到之病人住址皆轉換成1997台灣大地基準之座標。在資料的地理層級為最小統計區下，我們應用全局莫蘭指數估計該細菌感染發生率的空間分佈，並利用局部莫蘭指數找出高盛行率群聚的核心位置。再利用空間相對風險函數探查顯著高風險感染熱區的範圍與位置，用以驗證局部莫蘭指數所找出的群聚核心位置與其是否一致。其中，空間相對風險函數所使用之對照組位置資訊是隨機從門牌系統中抽樣出來的，用以代表隨機感染的發生位置。
結果：在2015至2017年間，高雄市共有537位堪薩斯分枝桿菌感染者。根據全局莫蘭指數分析的結果，最小統計區的盛行率在空間上呈現顯著正相關，表明高盛行率的統計區彼此群聚在一起。（空間自相關等於7.4 ×〖10〗^(-3)，p值＝0.013）而局部莫蘭指數的顯著地圖指出小港區存在兩個高盛行率的群聚核心。針對高雄市區進行的空間相對風險函數分析，結果找出兩個顯著的感染熱區分別位於前金與鹽埕區的邊界以及小港區的內部。關於此三年間觀察到的感染者位置，在前金與鹽埕區的熱區中，共有十六位，而小港區的熱區內則有三十五位。高雄市顯著的空間相對風險範圍落在1.54至2.27之間，其中最高的顯著相對風險位於小港區的熱區裡。從用以檢測抽樣位置之不確定性的敏感度分析發現，所有分析的結果皆指出小港區的西側存在顯著熱區。雖然前金與鹽埕區的顯著熱區並非出現於每次的敏感度分析結果中，然而此區域的空間相對風險仍一致地相對高於除小港區外的其他區。
結論：此研究檢測出的兩個顯著高風險熱區說明高雄市堪薩斯分枝桿菌感染者的分布並非均質。感染來源與易感染人群的空間分布均可能有其特定的群聚趨勢。依此我們找出的結果，應當解讀為可能影響此感染症的所有因素在空間上的綜合表現。我們推論存在於小港區的感染熱區成因，可能是因為此區的居民更容易透過重工業工廠的灑水系統暴露於堪薩斯分枝桿菌而致。未來尚須更多生物與環境上的證據以佐證本研究之結論。

關鍵字

非結核分枝桿菌 ； 堪薩斯分枝桿菌 ； 莫蘭指數 ； 空間相對風險函數 ； 核密度估計

並列摘要

Background: Previous studies suggested that nontuberculous mycobacterium (NTM) infection may not be homogeneously distributed in a geographic region. The detection of NTM spatial cluster in previous studies was limited by the administrative boundary and the pooled data of different NTM species infection. We aimed to analyze the spatial pattern of mycobacterium kansasii (M. kansasii) infection in Kaohsiung, where the number of laboratory-confirmed mycobacterium kansasii infection has been increasing.
Methods: Data of M. kansasii infection were collected from four hospitals in Kaohsiung City between 2015 and 2017. The residence addresses of all the observed cases were transformed into coordinates under the Taiwan geodetic datum 1997 (TWD97). Under the spatial level of the smallest geographic unit, we applied the global Moran’s I to estimate the spatial pattern of incidence risk and used the local Moran’s I to identify the core of the high-risk clusters. We further used the spatial relative risk function to detect the high-risk area for the confirmation of the results from the local Moran’s I. To represent the randomly infected location, we sampled the location from the address database for the control group of spatial relative risk function.
Results: From 2015 to 2017, there were 537 observed cases in Kaohsiung City. The spatial autocorrelation was 7.4×〖10〗^(-3) (p-value = 0.013). The significance map of local Moran’s I revealed that there were two cores of the high-risk cluster located in the Xiaogang district. The spatial relative risk function also detected two significant areas in urban areas. One located across the Qianjin district and Yancheng district, the other mainly sat in the Xiaogang district. For Qianjin-Yancheng high-risk area, 16 observed cases included in the significant contour. In the Xiaogang district, there were 35 cases in the high-risk area. The significant spatial relative risk ranged from 1.54 to 2.27 in this city. The highest spatial relative risk was in the Xiaogang district. Through the sensitivity analysis of randomly sampled addresses, all heatplots indicated the existence of the significant high-risk area in the Xiaogang district. Our results showed a highly suspected hotspot for infection in the western Xiaogang district. Although significant high-risk areas of Qianjin and Yancheng districts did not present in all the sensitivity results, their relative risks in these areas were also higher than the other places.
Conclusions: The distinct significant high-risk clusters indicated that the M. kansasii infection was not randomly distributed. Infected sources may have specific spatial patterns, yet the geospatial distribution of the susceptible host probably exists clustered spatial patterns too. Therefore, our results should be interpreted as a combination of all factors that may relate to this infection. People who lived in the Xiaogang district may be more likely to be exposed to M. kansasii. The transmission may happen through the splashing sanitation water in the environment of heavy industries. Further researches for biological and environmental evidence are needed.

並列關鍵字

Nontuberculous mycobacterium ； Mycobacterium kansasii ； Moran’s I ； Spatial relative risk function ； Adaptive kernel density estimation

參考文獻

1. Prevots DR, Marras TK. Epidemiology of human pulmonary infection with nontuberculous mycobacteria: a review. Clin Chest Med. 2015;36: 13-34.

2. Park SC, Kang MJ, Han CH, et al. Prevalence, incidence, and mortality of nontuberculous mycobacterial infection in Korea: a nationwide population-based study. BMC Pulm Med. 2019;19: 140.

3. Huang HL, Cheng MH, Lu PL, et al. Epidemiology and Predictors of NTM Pulmonary Infection in Taiwan - a Retrospective, Five-Year Multicenter Study. Sci Rep. 2017;7: 16300.

4. Shiau MY, Lee MS, Huang TL, Tsai JN, Chang YH. Mycobacterial Prevalence and Antibiotic Resistance Frequency Trends in Taiwan of Mycobacterial Clinical Isolates From 2002 to 2014. Medicine (Baltimore). 2016;95: e2942.

5. Hoefsloot W, van Ingen J, Andrejak C, et al. The geographic diversity of nontuberculous mycobacteria isolated from pulmonary samples: an NTM-NET collaborative study. Eur Respir J. 2013;42: 1604-1613.