平地鄉結核病新近感染與再復發之探討-以彰化縣為例

Abstract

背景 雖然在西方國家已普遍使用DNA fingerprinting來預測結核病是否為新近感染結核病，但在高盛行率的國家，則受限於價格及潛在病灶之再活化與新近之再感染間錯綜複雜的關係而無法廣泛使用。 目的 因此本研究嘗試以另一種方法僅仰賴例行性的流行病學調查資料來研討： (1)評估結核病發生病例的性別差異、時間趨勢和地理分布的差異性 (2)發展指標病例分析法來估計不同地區在指標病例發生後時間間隔的分佈 (3)以(1)及(2)得到的實證結果來評估結核病的新近感染與潛在病灶的再活化 (4)根據(1)~(3)所得到的結果提出結核病在不同鄉鎮間的可能傳染模式 材料與方法 資料來源為彰化縣自1991年至2002年間的結核病登記資料，共有9673例結核病發生病例。首先計算年齡性別標準化罹病率(SMR)，並使用SMR將彰化縣26個鄉鎮分為三個區域：高危險區域、中危險區域、低危險區域。並利用地理區域及Poisson回歸模式來分析年齡別的曲線及時間趨勢。 此外本研究發展指標病例分析法依各種指標病例的可能組合來產生模擬資料，並與實際指標病例發生後時間間隔的分佈來進行比較，以估計第二個病例的平均有效再生數。並合併使用Cox回歸模式來區分新近感染與潛在病灶之再活化，並檢測與新近感染相關的傳染模式。

結果 依地理區域分析來看，我們發現有二處高發生率的聚集分佈，ㄧ呈西南方向分佈，由二林鎮到大林鄉，另ㄧ處則集中在彰化縣的南部三鄉鎮。除了一些鄉鎮如員林鎮、大林鄉、大城鄉，其餘鄉鎮男女性發生率成互補的關係，即若男性發生率高，則女性發生率低，反之亦然。 在時間趨勢的分析方面，在調整年齡、性別、鄉鎮之後，結核病的年發生率有6%的增加。此外，並發現年齡、性別及年代三者間對結核病的發生率有顯著的交互作用。 在指標病例的分析方面，產生第二個病例的有效再生數在所有鄉鎮均大於1。我們並發現位於二林的彰化監獄的病例，有最短的時間間隔(指標病例至產生第二例的時間間隔)，平均僅6.82個月。平均年有效再生數在所有年齡層為1.68，在15-24歲、25-34歲、35-44歲、45-44歲、55-64歲及65歲以上族群分別為1.78、1.38、1.90、1.63、2.15及1.74。在高危險地區，在埔鹽鄉的年有效再生數為1.29，其次為二林鎮的1.27。若以指標病例的年齡分層，則發現平均有效再生數在25-34歲為最大，為1.52。若結合Cox回歸分析，可發現回歸係數在各年齡層均不同。使用相同的研究方法，則可以發現在任二個鄰近的鄉鎮有單向或雙向的新近傳染發生。 結論 本研究提出ㄧ個系統性的方法來評估潛在病灶之再活化與新近感染相關的時間趨勢、性別差異及年齡差異。我們發現新近感染似乎發生在男性或所有年齡層而非西方國家所觀察到的主要發生在年輕族群。由方法學的觀點來看，指標病例分析法可以和DNA fingerprinting互補，以利偵測新近感染的發生。

Abstract Background Despite the widespread use of DNA fingerprinting in detecting recent infection in Western countries, the feasibility of applying this technique to the country with high prevalence of TB may be limited to cost and intricate relationship between re-activation and recent infection. Aims We provided an alternative approach merely based on routine surveillance of epidemiological data (1) to assess sexual difference, time trend, geographic variation in TB incident cases; (2) to develop an index-case approach to estimate the distribution of time interval since occurrence of index cases by geographic area; (3) to assess recent infection or re-activation using the empirical evidence from (1) and (2); (4) to propose possible transmission routes across townships based on (1)-(3). Methods The epidemiological data on a total of 9,673 TB incident cases in ChangHua, one of middle counties in Taiwan, between 1991 and 2002 were collected. Age-sex adjusted Standard Morbidity Ratio (SMR) was first applied to categorizing 26 townships into three classes: high-risk area, moderate-risk area, and low-risk area. Age-specific and time-trend by three risk areas were analyzed by graphics or Poisson regression model. An index-case approach by generating data with every possible combination of index case was developed to fit the distribution of time-interval since occurrence of index cases and to estimate average number of reproductive number of secondary cases. This approach together with Cox regression model was applied to distinct recent infection form re-activation and to identify possible transmission related to recent infection. Results In geographic analysis, we identified two high rate areas, one extending southwestward from “二林鎮“ to “大村鄉“ and the other clustering in three townships of southern ChangHua. Except some townships such as “員林鎮”,”大村鄉”,”大城鄉“, there was a reciprocal relationship between male and female, high incidence rate in male but low in female and vice versa. In time trend analysis, there was a 6% (6%-7%) increase in TB incidence rate from 1991 to 2002 after controlling for age, gender, and townships. Three-way interactions with respect to age, gender, and calendar year were demonstrated. In index-case analysis, average reproductive number of secondary TB cases in all townships was larger than 1. We found TB cases from prison located in “二林鎮” had the shortest average interval time for reproducing secondary cases, approximately 6.82 month. Annual reproductive number was approximately 1.68 for all age groups, 1.78 for age 15-24, 1.38 for age 25-34, 1.90 for age 35-44, 1.63 for age 45-54, 2.15 for age 55-64, and 1.74 for age 65+ . Of high-risk township, annual reproductive number was approximately 1.29 in “埔鹽鄉“ followed by 1.27 in “二林鎮” and so on. By the stratification of age of index cases, it was found that average reproductive number, 1.52, was the largest in age 25-34 years. By the application of index-case approach coupled with Cox regression, we found the regression coefficients vary with age groups. Using the similar approach, the possible one-way or two-way recent infections in any two adjacent areas were proposed after taking back and forth between two areas into account. Conclusions The present study developed a systematic method to assess time trend, sexual difference, and age variation with respect to re-activation or re-infection. Our findings suggest newly infection may be involved in male and all age groups unlike only seen in the young subjects in Western countries. For the methodological viewpoint, our index-case approach method may complement DNA fingerprinting to detect newly infection.