系統性評估當前食品中戴奧辛最大限量:機率累積性健康風險評估

Abstract

戴奧辛是一種持久性有機污染物，具有高度的親脂性與生物累積性。戴奧辛與生殖功能異常、發育問題、神經毒性、免疫毒性、甲狀腺激素異常及癌症等健康風險有關。研究指出，飲食攝取是人體暴露於戴奧辛的主要途徑。因此，歐洲食品安全局已訂定戴奧辛在食品中的最大限量，以預防戴奧辛對民眾健康產生不良健康效應。目前，台灣的最大限量主要是參考歐洲食品安全局的標準，由於各地區飲食習慣的不同，該標準是否適用於台灣仍待系統性評估。 為了評估台灣食品藥物管理署制定的戴奧辛最大限量是否能夠維護民眾健康，我們進行了系統性機率累積性健康風險評估。本研究假設食品中戴奧辛的殘留濃度屬於截斷常態分布，並分別假設符合最大限值的機率，也就是最大限值為食品中戴奧辛的殘留濃度分布的95%、97%、與99%的信賴區間上限。此分布的最大值為平均值加上5個標準差，最小值等於檢測極限，而此分布仍受變異係數的影響。為此參考國內發表的食品中戴奧辛檢測數據，標準差與平均數的比值(變異係數)為0.1至1.7，取其平均值1.2作為評估的基礎。國人每日食物攝取量和體重皆引用自台灣國家攝食資料庫，並假設為截斷常態分布。使用 Crystal Ball 軟體來執行蒙地卡羅模擬，進行10000次運算。參考世界衛生組織訂定的戴奧辛及多氯聯苯每日耐受量1-4 pg TEQ/kg bw 以及歐洲食品安全局每周耐受量2 pg TEQ/kg bw進行估算民眾的危害指數，並假設為連續型均勻統計分布。在本研究中，因食物攝取量資料庫分為一般民眾和消費者兩個族群，每個族群的年齡層分別為0-3歲、3-6歲、6-12歲、12-16歲、16-18歲、19-65歲和大於65歲，因此我們就每個族群與年齡層分別執行機率累積性健康風險評估。 將最大限值假設為整個戴奧辛殘留濃度分布的95%上限，變異係數固定為1.2時，結果顯示參考世界衛生組織的耐受量下，一般民眾的平均危害指數分別為4.73、2.55、2.18、1.53、1.39、1.45、1.28；在消費者族群中為11.69、6.64、5.89、4.08、3.64、4.36、3.27。所有組別的危害指數都大於1，顯示可能對民眾健康的保護程度不足。英國、匈牙利、法國、奧地利一般民眾組 HI 值的平均值分別為0.78、0.89、1.23、0.97，消費者組 HI 值的平均值分別為1.18、1.76、1.65、1.72。進一步分析對各年齡層危害指數的貢獻度，最高的為魚類及其製品，次要為其他水產類及其製品。危害指數會隨著食品中戴奧辛符合最大限量的機率提高而下降，另外也會隨著檢測數據的變異係數增加而下降。 因戴奧辛檢測成本高，食物種類繁多，要針對每種食物採樣分析至具代表性的樣本數不易。因此本論文建立的方法可以供政府作參考，執行系統性的風險評估，評估當前的戴奧辛最大限量是否有效維護民眾健康，並且探討國人最主要的戴奧辛暴露貢獻源，做為政府市政優先順序之參考。

Dioxins are a group of highly lipophilic and persistent contaminants and ubiquitously accumulated in environment. Exposures to dioxins have been associated with risks of reproductive, developmental, neurodevelopmental immunotoxin, thyroid hormone adverse effects and cancers. Dietary intakes account for more than 95% of dioxin exposures. Therefore, the maximum levels (ML) of dioxins in various foods were set to protect the general public from adverse effects by many international organizations and countries. To evaluate the “safety” of the current ML set by the Taiwan Food and Drug Administration (TFDA), an probabilistic cumulative health risk assessment was developed. The residue distribution of dioxins in a food was assumed to belong to a truncated normal distribution, and their compliance rate is 95%, 97%, and 99% so that its ML equals the upper bound of 95%, 97%, and 99% confidence interval, respectively. Further assumption was made that the maximum (Max) equals the mean + 5 standard deviations (SD) of its distribution, and the minimum (Min) equals the limit of detection. Daily food intake rates and body weight were cited from the National Food Consumption Database of Taiwan and assumed to belong to truncated normal distributions. The Monte Carlo simulation was run using the Crystal ball software for 10000 trials. According to dioxin data published from Taiwan, the coefficient of variation (CV) in the dioxin residue distributions in food varied from 0.1 to 1.7. In all following simulations, the coefficient of variation was fixed to its average value of 1.2. The tolerable daily intake (TDI) for dioxins and dioxin-like PCBs of 1-4 pg TEQ/kg-day is recommended by the World Health Organization (WHO), and the tolerable weekly intake (TWI) of 2 pg TEQ/kg-week recommended by the European Food Safety Authority (EFSA) are used for assessment. The food consumption rate data are classified into the general public and consumer only, and each category is further stratified into seven age groups: 0-3, 3-6, 6-12, 12-16, 16-18, 19-65, and >65 years old. Based on the TDI by WHO, the mean hazard index(HI) in the general population were 4.73, 2.55, 2.18, 1.53, 1.39, 1.45, and 1.28, and in the consumer were 11.69, 6.64, 5.89, 4.08, 3.64, 4.36, and 3.27. The major contributions to the mean HI were fishes and seafoods to the Taiwan populations. The mean HI in the general population groups in the UK, Hungary, France, and Austria were 0.78, 0.89, 1.23, and 0.97, respectively, and in the consumer groups were 1.18, 1.76, 1.65, and 1.72. With the cost of sampling and analysis of dioxins in food, and numerous categories of food, representative dioxin residue data in food are difficult to obtain for systematical risk assessment. In this study, a simple statistical method is developed for cumulative probabilistic risk assessment to evaluate the protection of the current ML of food. Our results demonstrate that our method is very cost-efficient for systematically evaluate the ML of food and to identify the major exposure sources of dioxins through food consumption to set priority for ML revisions to better protect the general public.