食品中毒群聚事件問卷資料的終點分析與存活分析

Endpoint Analysis and Survival Analysis of Food Poisoning Cluster Questionnaire Data

食品中毒群聚事件調查目的主要在確定與事件有關的病因物質和原因食品。其中，判定原因食品常無法由例行性檢驗來判定，必須採用流行病學調查來探討相關的原因食品。分析流行病學調查問卷資料的方法有終點分析法和存活分析法。兩者都可以提供統計檢定的結果，後者還能以存活分析圖展現發病過程的趨勢。本文以終點分析法和存活分析法分別探討和比較新北市某中學學生食用團膳午餐後引起產氣夾膜桿菌食品中毒群聚事件問卷資料的原因食品。終點分析結果顯示宮保雞腿與食品中毒事件有統計顯著相關（調整年級別相對危險比3.69，95%信賴區間1.41–9.65）。存活分析結果也顯示宮保雞腿與食品中毒事件有統計顯著相關（調整年級別的Cox風險比值為4.09，95%信賴區間1.51–11.12）。後者考量發病過程的分析法更能精確地估算食用原因食品宮保雞腿的風險比值。

The objective of investigating foodborne outbreaks is mainly to identify the causative etiologies and foods related to the incident. Causative foods often cannot be found by routine testing, instead, we need to use epidemiological methods to explore possible causative foods. Methods for analyzing epidemiological questionnaire data include endpoint analysis and survival analysis. Both methods provide statistical test results, but the survival analysis curves can also show the trend of the disease process. We used both endpoint analysis and survival analysis to explore and compare the results for identifying causative foods that caused a Clostridium perfringens foodborne outbreak after eating a catered lunch among students of a middle school in New Taipei City in 2019. Endpoint analysis of the results showed that eating Kung Pao drumsticks was statistically significantly associated with the outbreak (grade-adjusted relative risk: 3.69, 95% confidence interval: 1.41–9.65). Survival analysis of the results also showed that eating Kung Pao drumsticks was statistically significantly associated with the outbreak (grade-adjusted Cox hazard ratio: 4.09, 95% confidence interval: 1.51–11.12). The latter method, which considered the trend of the disease process, can more accurately estimate the risk of eating Kung Pao drumsticks.