某校結核病群聚事件調查：成人接觸者發病風險預測模式建構

背景及目標：某校園結核病群聚事件為我國近年來規模較大之校園結核疫情，公衛單位投入極多人力及時間於該事件處置，並首次實施成人潛伏結核感染治療，但直到校園環境通風改善介入措施完成後疫情始獲得控制。本研究希望能釐清環境通風不良與結核疫情之間的因果關係，作為防範類似事件的實證依據。亦希望能發展出一套適用於成人接觸者的發病風險評估系統，讓衛生單位能夠在第一時間辨認高風險接觸者，並提供其潛伏結核感染治療。方法：採用回溯性世代追蹤研究，納入2010年11月29日源頭個案通報至2013年12月31日止，經疫情監測與本事件個案結核菌株經基因型別比對相同、或培養陰性但疫調具流病相關性，共24名結核病個案之所有接觸者，以羅吉斯迴歸及考克斯比例風險迴歸對接觸者發病之危險因子進行單變項、多變項分析。我們進一步將結核病危險因子及其對累積發病風險之影響，以Cox Proportional Hazards Model計算其風險計分，再以ROC曲線驗證成人接觸者發病風險計分模式的整體預測準確度。結果：1,665名接觸者中19名發病，經由羅吉斯迴歸之多變項分析結果，可知指標個案初痰塗片結果陽性（OR=19.959, P=0.0041）、同住接觸者（OR=29.805, P=0.0026）及環境通風不良（OR=40.333, P=0.0004）等項符合統計顯著意義（p<0.05）。經由考克斯比例風險迴歸之多變項分析模式，計算出指標個案初痰塗片結果陽性之風險計分為3分、同住接觸者之風險計分為4分，環境通風不良之風險計分為4分。以ROC的AUC判斷整合上述三項危險因子之成人接觸者發病風險計分模式預測準確度達0.848。成人接觸者若具有危險因子三、二、一、零項，三年內累積發病風險分別達100%、4.31%、0.21%或0％。結論：本研究結果可知環境通風不良確實與結核病接觸者發病有正相關，對高危險環境（如醫院、人口密集機構、校園等）應加強其環境通風品質，以利減少結核病群聚事件發生。此外，本研究亦發展出成人接觸者發病風險計分模式，讓衛生單位能夠在第一時間辨認高風險接觸者，及時提供預防性治療。

關鍵字

結核病；群聚事件；接觸者；環境通風；風險計分

並列摘要

Background and Objectives: A large tuberculosis outbreak with 24 cases occurred in a university during 2010 to 2013. Poor ventilation in the classroom had been proposed as a risk factor for this outbreak, but the causation has not yet been established. This study aimed to investigate whether poor ventilation is a risk factor for tuberculosis among adult TB contacts, and developed a risk predictive score for contact investigation. Methods: We conducted a retrospective cohort study for all contacts of the 24 cases involved in this outbreak from November 29, 2010 to December 31, 2013. The 24 cases are either of the same genotype of Mycobacterium tuberculosis, or are culture-negative but with epidemiological link. We used logistic regression and Cox proportional hazard regression to analyze risk factors for TB among contacts. We further used Cox proportional hazard regression model to develop a risk predictive score. Results: Among the 1,665 contacts, 19 developed TB. The results of logistic regression analysis indicate that sputum smear-positivity (OR=19.959, P=0.0041), roommate contact (OR=29.805, P=0.0026), and poor ventilation (OR=40.333, P=0.0004), are statistically significant risk factors for TB in contacts. An 11-point scoring system was developed, including smear-positivity, roommate contact, and poor ventilation. Area under the receiver operating characteristic curve was 0.848 (95% confidence interval, 0.783–0.913). The risk of developing active TB within 3 years is 100%, 4.31%, 0.21%, and 0% for contacts with risk scores of 11, 7~8, 3~4, and 0, respectively. Conclusions: Our results show poor ventilation in classroom is a risk factor for TB among adult contacts. Our risk predictive score system can assist public health agencies to identify high risk contacts at initial investigation so that preventive treatment can be provided in time.