食品中毒死蜱残留的暴露评估
|
摘要
随着世界人口的不断增长,如何正确解决食品及其安全性问题已成为各国政府共同关注的问题,其中农药在确保农作物稳产、高产的问题上发挥着重要的作用,但同时农药在食品中的残留和毒性也为世界所关注。常用农药品种百余种,主要有有机磷类、有机氯类农药、氨基甲酸酯类农药、杀蚕毒素类农药以及拟除虫菊酯类农药等等。我国作为农产品和食品出口大国,在国际贸易中经常面临各进口国利用农药残留来设置技术性壁垒,因此开展建立在科学数据基础之上的,符合国际食品安全暴露评估通用规则的农药残留膳食暴露评估,评估食品中农药残留给我国人口健康带来的风险,将为管理部门提出更有效的管理措施以应对进口国设置的技术性贸易壁垒,保护我国消费者利益提供有力的技术支持,此项研究具有重要的学术价值和现实意义。
本研究运用2000-2006年全国污染物监测网数据以及2005和2006年的出口食品农药残留监控数据,我国山东省2005年出口日本的速冻菠菜农药残留监控数据,2002年中国居民营养与健康状况调查数据,构建了食品中农药残留的暴露评估模型,包括点评估模型及概率评估模型等,这些模型的使用很大程度上依赖于所获得的评估数据以及对于评估参数变异性和不确定性的分析,本论文以毒死蜱为例,基于我国毒死蜱污染现状以及国际食品中农药残留暴露评估研究的现状,旨在研究：第一,我国农药残留检测确证方法,农药残留的总体情况分析；第二,毒死蜱在我国膳食结构中的暴露风险；第三,农药残留膳食暴露点评估与概率评估的区别；第四,出口蔬菜与国内消费蔬菜毒死蜱残留暴露情况的区别以及概率评估量化变异性与不确定性的方法,主要研究内容及成果如下：
一、通过对食品中农药残留分析检测技术研究以及农药残留数据的统计分析处理研究得出如下结论：
1.为了表明农药残留检测中多残留检测方法与确证方法同样重要,而且检测方法是暴露评估的基础,检测结果的准确程度直接决定评估的结果,因此本论文增加建立了我国尚没有建立检测标准方法的敌菌灵GC-MS确证检测方法,该方法在测量范围内具有良好的线性关系,测定的回收率范围及检出限均符合我国进出口粮谷中敌菌灵的检测限量要求。
2.现有数据的农药残留统计结果表明,动物源食品中的农药残留检出率高于植物源食品,因此动物源食品中的农药残留也值得我们去关注,即在动物体内有蓄积作用的农药残留也应该给予重视。经过浓缩加工的食品农药残留检出率高,经过深加工的食品农药残留检出率没有初级产品农药残留检出率高；毒死蜱检出率排名1-6位的食品组别分别为“茎类蔬菜”、“浆果类水果”、“豆类”、“叶类蔬菜”、“芸苔类蔬菜”、“茶类”,应该重点监控。2003-2006年毒死蜱超标的食品组别均有“茎类蔬菜”、“鳞球茎类蔬菜”、“叶类蔬菜”、“根块茎蔬菜类”,说明这几类蔬菜应当是我国蔬菜监管的重点。
3.2003-2006年全国毒死蜱含量地区分布情况表明,我国整体南部地区比北部地区毒死蜱的残留高。从各地区的毒死蜱残留P97.5数据来看分别为珠海、福建、浙江、云南位居前列；从毒死蜱残留最大值来看,重庆、湖北、广西、广东位居前列,这可能与我国南北地区气候差异,农药施用剂量与施用频率有关。
4.本次研究采用的全国各类食品中的农药残留监控数据,由于对各类食品监控目的不同,采用的检测方法不同,各种污染物数据的样本量差异很大,在农药残留监控数据中98%以上均为未检出数据的情况下,未检出数据的最低检出限就显得非常的重要,尤其在样本量少的情况下,会出现异常数据结果,因此也更体现出数据质量对于评估结果的重要性,在以后的工作中我们应当有目的的去采集各类数据,使得数据的代表性和数据质量更高,这样的评估结果会更可靠。
二、通过对输日菠菜和我国全人群消费菠菜毒死蜱残留量的膳食暴露点评估,得出如下结论：
1.毒死蜱的急性暴露评估：第一,%aPAD尤其以1-6岁幼儿偏高,其中1-3岁农村的幼儿%aPAD为115.74,4-6岁农村的幼儿%aPAD为110.43>100属于高危人群,这可能是由于儿童的单位体重消耗的食品量大造成毒死蜱的摄入量偏高；第二,各年龄组的男性人群毒死蜱的%aPAD略比女性人群低,可能原因是女性饮食结构中摄入的蔬菜相对比例较男性大；第三,从地域组的评估结果来看,18岁以下年龄组毒死蜱%aPAD农村人口显著高于城市人口,18岁以上成年人农村人口的毒死蜱%aPAD略高于城市人口,这与城乡居民在蔬菜总消费量、饮食结构差异和农村人群体重略低有关；第四,从全人群的评估情况来看菠菜中毒死蜱的%aPAD为43.09<100属于安全范围。
2.毒死蜱的慢性暴露评估：第一,当毒死蜱的残留检测未检出数据以LOQ代替时,城市1-3岁幼儿%cPAD最大值为6.59<<100,因此菠菜中毒死蜱残留的慢性暴露可以不作为评估和关注的重点；第二,菠菜中毒死蜱的%cPAD值女性略高于男性,这与急性膳食暴露的评估结果一致；第三,1-3岁和7-10岁的儿童,菠菜中毒死蜱的%cPAD值城市略高于农村；其余年龄段人群中,菠菜中毒死蜱的%cPAD值农村均高于城市；第四,在不同年龄段范围内,毒死蜱残留检测数据中未检出数据分别以LOQ和0代替,点评估结果%cPAD值相差5-9倍。
3.毒死蜱的急性与慢性膳食暴露评估结果的比较：从菠菜中毒死蜱急性和慢性暴露点评估结果可以看出,任意组人群的%aPAD值均远远高于%cPAD值,全人群的%aPAD为43.05,%cPAD为2.94,菠菜中毒死蜱的急性暴露风险约为慢性暴露的15倍,因此毒死蜱的急性毒性是我们今后研究和管理工作的重点。
4.出口日本与我国国内消费菠菜的暴露情况：评估结果的差异主要体现在加工因子上,由于我国饮食结构的差异性本次菠菜全国膳食摄入的点评估中,加工因子设定为1,若我国人群膳食也采用严格的水洗、漂烫等家庭烹调或加工步骤,国内消费菠菜的%aPAD为17.22,%cPAD为0.69；出口菠菜%aPAD为16.2,%cPAD为0.4,均在安全范围内。
三、通过对食品中毒死蜱残留膳食暴露量的概率评估得出如下结论：
1.菠菜中毒死蜱未检出数据分别以0、1/2LOQ、LOQ代替,概率评估结果均表明年龄越小毒死蜱暴露量越高,1-6岁低龄幼儿单位体重毒死蜱暴露量最高;P99.99百分位数出现4-6岁儿童的菠菜毒死蜱暴露量的最大值；菠菜中毒死蜱暴露也存在地区差异,农村人群暴露量要高于城市人群,可见农村食品中毒死蜱的残留可能比城市要严重。各年龄地域组人群每日暴露量均未超标且远远小于ARfD。
2.菠菜中毒死蜱暴露量分布为左偏态分布。从四分位数间距P75-P25看,中间50%人群变异较小,较为稳定。菠菜中毒死蜱暴露地区差异,同样表现出P99,P99.9,P99.99农村人群暴露量高于城市人群,暴露量高端百分位数(如P99)是低端百分位数(如P25)的近18倍,暴露量分布为左偏态分布。农村食品中毒死蜱的残留比城市要严重。
3.菠菜以及全食品中毒死蜱残留概率评估的不确定性分析结果显示,在95%可信区间都包含了变异性结果各个百分位数。这说明按100万次抽样是足够的,暴露量百分位数能够达到稳定,Monte Carlo模拟较好地量化了变异性；如果暂不考虑不确定性,只要模拟次数足够,仅采用Monte Carlo方法算得的百分位数已经稳定,能够满足实际需求。同时不确定性分析结果也表明百分位数越高,95%可信区间越宽,不确定性越大。
4.全食品毒死蜱概率评估中未检出数据以LOQ代替,结果表明全人群的暴露量变异性P99.99为11.139小于ARfD值(ARfD=100)。各年龄组的亚人群处于暴露量最大的1-3岁幼儿和4-6岁幼儿,其暴露量的最大值为17.544小于ARfD值。从城乡的毒死蜱暴露量变异性可以看出,农村的毒死蜱暴露量要大于城市人群,暴露量的最大值30.02小于ARfD值。
5.急性概率评估的食品贡献度分析结果显示,按食品大类(type)进行分析,全人群、城市人群、农村人群食品中毒死蜱摄入量居于前三位的分别是：蔬菜类及制品、水果类及制品、薯类淀粉及制品。为更精确地描述人群摄入毒死蜱的具体食品的贡献度,按较细化的食品组别(group)进行分析,全人群食品中毒死蜱摄入量居于前三位的分别是嫩茎、叶、花菜类,茄果、瓜菜类,鲜豆类,贡献度分别为81.75%、5.94%、2.75%,累计达90.44%；全人群及各亚群的高暴露人群(P95)食品贡献度与其总人群食品贡献度相同,这与毒死蜱膳食暴露量呈左偏态分布的规律是一致的。
6.由于慢性膳食暴露评估没有考虑食品加工因子以及毒死蜱自身的半衰期等等因素,因此评估数据虽然显示毒死蜱各百分位数的变异性结果都小于ADI值,但如果考虑FQPA系数就会处于警戒值了,这与毒死蜱的慢性毒性表现与毒死蜱慢性中毒的发生情况是完全不符合的,因此本部分研究也可以看出农药残留的慢性膳食暴露概率评估还需要有各方面大量的数据支持才可以进行评估。
7.当全人群的毒死蜱残留未检出数据分别以0、1/2LOQ、LOQ代替时,对于均数的影响还是比较大的,但对于高百分位数如P99.9甚至P99.99的差异就越来越小了。
四、点评估结果与概率评估结果的比较
1.毒死蜱在菠菜和全食品中的风险估计说明,我国全人口摄入菠菜中毒死蜱点评估风险结果为处于高风险的农村1-3岁和4-6岁幼儿,其%aPAD分别为115.74和110.43,在概率评估中99.9%位数的%aPAD结果分别为24.3和22.2；99%位数的%aPAD结果急剧下降,分别为1.818和1.936；95%位数的%aPAD结果与99%位数结果相近,分别为1.449和1.320。可以看出概率评估结果比较切合实际,而且从概率评估中也可以看出菠菜中毒死蜱急性毒性风险比较高的人群也是安全的。
2.点评估与概率评估结果相同,即各年龄、地域亚人群组中农村1-3岁和4-6岁幼儿为风险最高人群,其次是城市1-3岁和4-6岁幼儿,点评估的结果要比概率评估结果高2-160倍不等,因此点评估是非常保守的。对于保护大多数人来说,其结果可以作为初步参考或确定是否需要进行概率评估的筛选措施。在整个评估过程中也可以看出概率评估所需要的数据量以及软件和硬件、时间的投入都要比点评估要高的多。点评估和概率评估都是膳食暴露评估的两种重要手段,而且不可以相互代替。当点评估结果显示风险较小时就没有必要进行概率评估。
3.从我国人口全食品中毒死蜱急性膳食暴露量概率评估的风险估计结果来看,处于风险最高的亚人群组依次为城市1-3岁幼儿、农村1-3岁和4-6岁幼儿%aPAD分别为91.197、82.895、和83.255各数值接近但均小于100,这是在没有考虑安全因子的情况下评估的结果,可以认为我国各人群组食品中毒死蜱的暴露处于安全剂量,同时在亚人群中城市和农村1-6岁幼儿仍然是我们毒死蜱农药残留关注的重点,考虑到本研究第一章中毒死蜱在菠菜中的加工因子研究,建议广大消费者食用经过清洗和漂烫过的蔬菜,毒死蜱的含量会大大下降。
With the increasing population of the world, all government are facing the problem with how to solve food supply and food safety properly, and therein pesticide has been playing an important role in protecting crop and enhancing yield, however, all people concern pesticides residue in food and their toxicity. There are hundreds of pesticides, such as phosphate pesticide, chlorinated pesticide, etc. As a major food exporting country, China is frequently facing technical trade barrier with pesticide residues set up by some importing countries during international trade, therefore, based on scientific data, developing dietary exposure assessment for pesticides residues under general guideline to total exposure assessment and evaluating health risk from pesticides residue will provide powerful scientific basis for management department to take more effective measures in order to break down technical trade barrier from importing countries and protect domestic consumer benefit. This study does have important academic value and practical benefits.
This study builds exposure assessment model for pesticides residue in food based on data from national pollutant monitoring network in 2000-2006, monitoring data for pesticides residue in export food in 2005-2006, monitoring data for pesticides residue in frozen spinach exported from Shandong Province to Japan in 2005, the data of national dietary survey in 2002. This model includes point estimation and probabilistic assessment model, etc. and this model is built on data available and analysis on variability and uncertainty. This paper takes chlorpyriofos as example to study exposure risk of chlorpyriofos in Chinese dietary structure, difference between point estimation and probabilistic assessment of pesticides residue in diet, difference of exposure assessment for cholorpyriofos residue in export vegetable and domestic market and analysis methods for variability and uncertainty by probabilistic assessment. Most important achievement as below:
Ⅰ. Based on research on analysis method about pesticide residue in food and statistical analysis on pesticide residue data in food, conclusions are drawn as the following:
1. The detection process for anilazine GC-MS has been established, which proves that multiple residual detection method of high flux for pesticide is as important as its validation method. The process has good performance in linear relationship, the range of recovery rate and detection limit can meet standard from anilazine detection method for import and export food in China
2. The positive rate of pesticide is higher in animal food than in vegetable food. Therefore the pesticide residues in animal food should be paid more attention, namely the pesticide that is accumulated in animal body needs special attention. Meanwhile, concentrated pesticide has a relatively higher positive rate, whereas the positive rate of pesticide in food product that undergone deep processing is lower than that of the primary farm product. "Stem vegetable", "Berries fruit", "Beans", "Leafy vegetable", "Brassica vegetable", and "Tea" are rank 1 to 6 respectively in the list of positive rate of chlorpyrifos,. The fact that "Bulb vegetable", "Stem vegetable", "Leafy vegetable" and "Tuber vegetable" are all found on the list of non-conformity chlorpyrifos content in food category in 2003-2006. This shows that these vegetable categories should be highlight of vegetable safety supervision in China.
3. The distribution of chlorpyrifos by region in 2003-2006 in China indicates that chlorpyrifos residue is higher in the southern area than that in the northern area in China. The top 4 at P97.5 by region are Zhuhai, Fujian, Zhejiang, Yunnan, respectively, which are all located in the South of China. The top list of the maximum chlorpyrifos residue are Chongqing, Hubei, Guangxi, Guangdong respectively. This result could be related to the climate difference between southern and northern part of China.
4. As per monitoring data we collected in all groups of food, due to different objectives and different methods and different sample size may lead to different lowest detection limit. Therefore given the case of 98% residue is non-detectable, the lowest detection limit of the non-detected will be extremely important, especially when the sample size is relatively limited and this may lead to outliers. This shows that quality of data will have direct impact on assessment result and therefore assessment results will be much more reliable if data are representative and collected with clear objective.
Ⅱ. Through point estimation of dietary exposure assessment for chlorpyrifos residue in spinach for export and for domestic consumption, conclusions are drawn as below:
1. As per acute exposure assessment for chlorpyrifos:Ⅰ. the %PAD of chlorpyrifos is relatively higher in the group aged 1-6, amongst the %aPAD of the rural group aged 1-3 is 115.74 and the %PAD of the rural group aged 4-6 is 110.43 which is above 100, it shows that this group is high risk population and probably due to higher dietary intake per Kg body weight by children;Ⅱ. The %aPAD in male groups is relatively lower than that in female groups, and this could be that vegetable consumption in female groups is higher than in male group;Ⅲ. Based on assessment by region, the results show that the %aPAD of rural population in the children and adolescent group under 18 is apparently higher than urban group, and the figures in adult aged above 18 in rural group is also a bit higher than in urban group. The conclusion is related to the total vegetable consumption, differences in diet structure and body weight;Ⅳ. Based on assessment by total population, the %aPAD of chlorpyrifos is 43.09<100 which is safe..
2. As per chronic exposure assessment of chlorpyrifos:Ⅰ. The max %cPAD in urban children is 6.59 which is far below than 100 when non-detected chlorpyrifos residue replaced by LOQ, therefore it is not necessary to put the chronic assessment of chlorpyrifos under highlight;Ⅱ. The %cPAD of chlorpyrifos in spinach is relatively higher in female group than in male group and this is aligned with the result from acute assessment;Ⅲ. For children aged 1-3 and 7-10, the %cPAD in urban group is higher than in rural group, however, for people in other age group, the %cPAD in rural group has higher figure than in urban group;Ⅳ. For different age group, if non-detected chlorpyrifos residue is replaced by LOQ and 0, the %cPAD from point estimation varies from 5-9 times.
3. Comparison of assessment results from the acute and chronic exposure assessment of chlorpyrifos:acute and chronic exposure point estimation shows that the %aPAD of any group is much higher than the %cPAD. The %aPAD of the total population is 43.05 and while the %cPAD is 2.94. The risk from acute exposure assessment is 15 times higher than chronic exposure assessment. Therefore, acute toxicology of chlorpyrifos should be highlight of research and management in upcoming works.
4. Difference between spinach exported to Japan and spinach for domestic consumption:results show that main difference is from process factors due to variability in China dietary structure. If Chinese consumer can follow the procedure to wash and bleach vegetable carefully, the %aPAD is 17.22 and the %cPAD is 0.69 for domestic consumer and %aPAD is 16.2 and %cPAD is 0.4 for export spinach for Japan, all data is within safe range.
Ⅲ. Conclusions are drawn based on the probabilistic assessment for exposure of chlorpyrifos in diet:
1. When non-detected data for chlorpyrifos in spinach are replaced with 0, 1/2LOQ or LOQ, probabilistic assessment shows the younger the age is, the higher for exposure of chlorpyrifos as per Kg body weight, group aged 1-6 has the highest exposure, and group aged 4-6 has the maximum exposure at percentile of 99.99. The exposure of chlorpyrifos in spinach also varies from region to region, as a result, the rural group has a relatively higher exposure than urban group and this indicates that a higher chlorpyrifos residue in food in rural area. However, the daily exposure of chlorpyrifos by age and region is below the limit and far less than ARfD.
2. The distribution of exposure of chlorpyrifos in spinach is a left skew distribution. The quartile range P75-P25 shows that 50% of the population in the middle has less variability. Different exposure of chlorpyrifos by region indicates that rural group has higher level of exposure than urban group at percentile of P99, P99.9, P99.99. Where exposure level of rural group higher than urban group, there is high level of percentile (e.g. P99), there is nearly 18 times of exposure level than where percentile is low (e.g. P25). Exposure is a left skew distribution and chlorpyfrifos residue in rural area is much more serious than in urban area.
3. The uncertainty analysis from probabilistic assessment on chlorpyrifos residue in spinach and in food shows that all percentiles of variability are within the 95% confidence interval. It shows 1 million sample size is sufficient to stablize the exposure percentiles and the Monte Carlo simulation has quantified the variability. If we put aside the uncertainty issue, the Monet Carlo percentiles are sufficiently stable for actual application. Meanwhile, the uncertainty analysis also shows that the higher the percentile is, the broader 95% confidence interval is and the higher the uncertainty is.
4. If non-detected data is replaced with LOQ in the probabilistic assessment of chlorpyrifos in all food, the result shows the variability of the exposure in total population groups at P99.99 is 11.139 which is lower than ARfD (ARfD=100). Subgroup aged 1-3 and 4-6 has the highest exposure as 17.544 which is lower than ARfD. From the variability of exposure of chlorpyrifos in urban and rural area, it concludes that exposure in rural area is higher than in urban area and maximum exposure in rural is 30.02 which is less than ARfD.
5. The food contribution analysis in the acute probabilistic assessment shows that, based on food types, the top 3 of chlorpyrifos intake in total population, urban group and rural group are:vegetable and its product, fruit and its product, potato starch and its product. In order to describe the specific contribution of chlorpyrifos intake by subgroup of food, the top 3 contributors of chlorpyrifos intake in the total population are:"young stem, leaves and flower vegetables"; "solanaceous fruit vegetable, and cucurbit"; "fresh bean". And their contribution is 81.75%,5.94% and 2.75% respectively with a sum of 90.44%. Food contribution from all group and subgroup with high exposure is same than that in total population, and this is aligned with the left skew distribution of exposure of chlorpyrifos in diet.
6. Due to food processing factors and the half life of chlorpyrifos are not considered during chronic dietary exposure assessment, results show that variability at each percentile in chlorpyrifos are all less than the ADI value, however, if the FQPA coefficient is taken into account, variablity will be close to alerting limit, and this is not in accord with chronic toxicology and symptoms of chronic toxicity of chlorpyrifos. Thus we concluded that chronic toxicology probabilistic assessment needs more data support.
7. When non-detected data of chlorpyrifos residue is replaced with 0,1/2LOQ or LOQ respectively, we found that it has significant effect on the mean value, however difference becomes smaller and smaller at high percentiles such as P99.9 or even P99.99.
Ⅵ. Comparison between the results of point estimation and probabilistic assessment
1. Risk assessment of chlorpyrifos in spinach and in all food shows that point estimation result from total population is that rural group aged 1-3 and 4-6, which the %aPAD is 115.74 and 110.43 respectively. The %aPAD at 99.9% percentile is 24.3 and 22.2 respectively in probabilistic assessment, and the %aPAD at 99% percentile is as small as 1.449 and 1.320 respectively. The figure shows that results from probabilistic assessment is reasonable and high risk population from acute toxicology assessment is still safe.
2. The point estimation has the same result than probabilistic assessment, namely the rural group aged 1-3 and 4-6 are high risk population, and the urban group aged 1-3 and 4-6 next to it. Result of point estimation is 2-160 times higher than that of probabilistic assessment, so point estimation is quite conservative. From the point of protecting as much people as we can, result could be used as primary reference or as criteria for judging whether probabilistic assessment is necessary. Across the assessment process, we can clearly see that probabilistic assessment requires more data, software and hardware, time and resource than the point estimation. Those two methods are very important for dietary exposure assessment and can not replace each other. The probabilistic assessment is not necessary when the point estimation shows relatively little risk.
3. According to risk assessment from the probabilistic assessment of acute dietary exposure assessment of chlorpyrifos in all food in China, the sub-groups with highest risk are the urban group aged 1-3, rural group aged 1-3 and 4-6, and the %aPAD is 91.197,82.895 and 83.255 respectively and all are close but less than 100. Safety factors are not taken into account for this case, therefore we concluded that exposure of chlorpyrifos for each group is within safety range. Meanwhile children aged 1-6 either in rural or urban area should be highlight in research of chlorpyrifos residue. Given research on processing factors provided in the chapter 1, we suggest consumers wash and bleach before have it and this enables great reduction of chlorpyrifos residue in food.
本研究运用2000-2006年全国污染物监测网数据以及2005和2006年的出口食品农药残留监控数据,我国山东省2005年出口日本的速冻菠菜农药残留监控数据,2002年中国居民营养与健康状况调查数据,构建了食品中农药残留的暴露评估模型,包括点评估模型及概率评估模型等,这些模型的使用很大程度上依赖于所获得的评估数据以及对于评估参数变异性和不确定性的分析,本论文以毒死蜱为例,基于我国毒死蜱污染现状以及国际食品中农药残留暴露评估研究的现状,旨在研究：第一,我国农药残留检测确证方法,农药残留的总体情况分析；第二,毒死蜱在我国膳食结构中的暴露风险；第三,农药残留膳食暴露点评估与概率评估的区别；第四,出口蔬菜与国内消费蔬菜毒死蜱残留暴露情况的区别以及概率评估量化变异性与不确定性的方法,主要研究内容及成果如下：
一、通过对食品中农药残留分析检测技术研究以及农药残留数据的统计分析处理研究得出如下结论：
1.为了表明农药残留检测中多残留检测方法与确证方法同样重要,而且检测方法是暴露评估的基础,检测结果的准确程度直接决定评估的结果,因此本论文增加建立了我国尚没有建立检测标准方法的敌菌灵GC-MS确证检测方法,该方法在测量范围内具有良好的线性关系,测定的回收率范围及检出限均符合我国进出口粮谷中敌菌灵的检测限量要求。
2.现有数据的农药残留统计结果表明,动物源食品中的农药残留检出率高于植物源食品,因此动物源食品中的农药残留也值得我们去关注,即在动物体内有蓄积作用的农药残留也应该给予重视。经过浓缩加工的食品农药残留检出率高,经过深加工的食品农药残留检出率没有初级产品农药残留检出率高；毒死蜱检出率排名1-6位的食品组别分别为“茎类蔬菜”、“浆果类水果”、“豆类”、“叶类蔬菜”、“芸苔类蔬菜”、“茶类”,应该重点监控。2003-2006年毒死蜱超标的食品组别均有“茎类蔬菜”、“鳞球茎类蔬菜”、“叶类蔬菜”、“根块茎蔬菜类”,说明这几类蔬菜应当是我国蔬菜监管的重点。
3.2003-2006年全国毒死蜱含量地区分布情况表明,我国整体南部地区比北部地区毒死蜱的残留高。从各地区的毒死蜱残留P97.5数据来看分别为珠海、福建、浙江、云南位居前列；从毒死蜱残留最大值来看,重庆、湖北、广西、广东位居前列,这可能与我国南北地区气候差异,农药施用剂量与施用频率有关。
4.本次研究采用的全国各类食品中的农药残留监控数据,由于对各类食品监控目的不同,采用的检测方法不同,各种污染物数据的样本量差异很大,在农药残留监控数据中98%以上均为未检出数据的情况下,未检出数据的最低检出限就显得非常的重要,尤其在样本量少的情况下,会出现异常数据结果,因此也更体现出数据质量对于评估结果的重要性,在以后的工作中我们应当有目的的去采集各类数据,使得数据的代表性和数据质量更高,这样的评估结果会更可靠。
二、通过对输日菠菜和我国全人群消费菠菜毒死蜱残留量的膳食暴露点评估,得出如下结论：
1.毒死蜱的急性暴露评估：第一,%aPAD尤其以1-6岁幼儿偏高,其中1-3岁农村的幼儿%aPAD为115.74,4-6岁农村的幼儿%aPAD为110.43>100属于高危人群,这可能是由于儿童的单位体重消耗的食品量大造成毒死蜱的摄入量偏高；第二,各年龄组的男性人群毒死蜱的%aPAD略比女性人群低,可能原因是女性饮食结构中摄入的蔬菜相对比例较男性大；第三,从地域组的评估结果来看,18岁以下年龄组毒死蜱%aPAD农村人口显著高于城市人口,18岁以上成年人农村人口的毒死蜱%aPAD略高于城市人口,这与城乡居民在蔬菜总消费量、饮食结构差异和农村人群体重略低有关；第四,从全人群的评估情况来看菠菜中毒死蜱的%aPAD为43.09<100属于安全范围。
2.毒死蜱的慢性暴露评估：第一,当毒死蜱的残留检测未检出数据以LOQ代替时,城市1-3岁幼儿%cPAD最大值为6.59<<100,因此菠菜中毒死蜱残留的慢性暴露可以不作为评估和关注的重点；第二,菠菜中毒死蜱的%cPAD值女性略高于男性,这与急性膳食暴露的评估结果一致；第三,1-3岁和7-10岁的儿童,菠菜中毒死蜱的%cPAD值城市略高于农村；其余年龄段人群中,菠菜中毒死蜱的%cPAD值农村均高于城市；第四,在不同年龄段范围内,毒死蜱残留检测数据中未检出数据分别以LOQ和0代替,点评估结果%cPAD值相差5-9倍。
3.毒死蜱的急性与慢性膳食暴露评估结果的比较：从菠菜中毒死蜱急性和慢性暴露点评估结果可以看出,任意组人群的%aPAD值均远远高于%cPAD值,全人群的%aPAD为43.05,%cPAD为2.94,菠菜中毒死蜱的急性暴露风险约为慢性暴露的15倍,因此毒死蜱的急性毒性是我们今后研究和管理工作的重点。
4.出口日本与我国国内消费菠菜的暴露情况：评估结果的差异主要体现在加工因子上,由于我国饮食结构的差异性本次菠菜全国膳食摄入的点评估中,加工因子设定为1,若我国人群膳食也采用严格的水洗、漂烫等家庭烹调或加工步骤,国内消费菠菜的%aPAD为17.22,%cPAD为0.69；出口菠菜%aPAD为16.2,%cPAD为0.4,均在安全范围内。
三、通过对食品中毒死蜱残留膳食暴露量的概率评估得出如下结论：
1.菠菜中毒死蜱未检出数据分别以0、1/2LOQ、LOQ代替,概率评估结果均表明年龄越小毒死蜱暴露量越高,1-6岁低龄幼儿单位体重毒死蜱暴露量最高;P99.99百分位数出现4-6岁儿童的菠菜毒死蜱暴露量的最大值；菠菜中毒死蜱暴露也存在地区差异,农村人群暴露量要高于城市人群,可见农村食品中毒死蜱的残留可能比城市要严重。各年龄地域组人群每日暴露量均未超标且远远小于ARfD。
2.菠菜中毒死蜱暴露量分布为左偏态分布。从四分位数间距P75-P25看,中间50%人群变异较小,较为稳定。菠菜中毒死蜱暴露地区差异,同样表现出P99,P99.9,P99.99农村人群暴露量高于城市人群,暴露量高端百分位数(如P99)是低端百分位数(如P25)的近18倍,暴露量分布为左偏态分布。农村食品中毒死蜱的残留比城市要严重。
3.菠菜以及全食品中毒死蜱残留概率评估的不确定性分析结果显示,在95%可信区间都包含了变异性结果各个百分位数。这说明按100万次抽样是足够的,暴露量百分位数能够达到稳定,Monte Carlo模拟较好地量化了变异性；如果暂不考虑不确定性,只要模拟次数足够,仅采用Monte Carlo方法算得的百分位数已经稳定,能够满足实际需求。同时不确定性分析结果也表明百分位数越高,95%可信区间越宽,不确定性越大。
4.全食品毒死蜱概率评估中未检出数据以LOQ代替,结果表明全人群的暴露量变异性P99.99为11.139小于ARfD值(ARfD=100)。各年龄组的亚人群处于暴露量最大的1-3岁幼儿和4-6岁幼儿,其暴露量的最大值为17.544小于ARfD值。从城乡的毒死蜱暴露量变异性可以看出,农村的毒死蜱暴露量要大于城市人群,暴露量的最大值30.02小于ARfD值。
5.急性概率评估的食品贡献度分析结果显示,按食品大类(type)进行分析,全人群、城市人群、农村人群食品中毒死蜱摄入量居于前三位的分别是：蔬菜类及制品、水果类及制品、薯类淀粉及制品。为更精确地描述人群摄入毒死蜱的具体食品的贡献度,按较细化的食品组别(group)进行分析,全人群食品中毒死蜱摄入量居于前三位的分别是嫩茎、叶、花菜类,茄果、瓜菜类,鲜豆类,贡献度分别为81.75%、5.94%、2.75%,累计达90.44%；全人群及各亚群的高暴露人群(P95)食品贡献度与其总人群食品贡献度相同,这与毒死蜱膳食暴露量呈左偏态分布的规律是一致的。
6.由于慢性膳食暴露评估没有考虑食品加工因子以及毒死蜱自身的半衰期等等因素,因此评估数据虽然显示毒死蜱各百分位数的变异性结果都小于ADI值,但如果考虑FQPA系数就会处于警戒值了,这与毒死蜱的慢性毒性表现与毒死蜱慢性中毒的发生情况是完全不符合的,因此本部分研究也可以看出农药残留的慢性膳食暴露概率评估还需要有各方面大量的数据支持才可以进行评估。
7.当全人群的毒死蜱残留未检出数据分别以0、1/2LOQ、LOQ代替时,对于均数的影响还是比较大的,但对于高百分位数如P99.9甚至P99.99的差异就越来越小了。
四、点评估结果与概率评估结果的比较
1.毒死蜱在菠菜和全食品中的风险估计说明,我国全人口摄入菠菜中毒死蜱点评估风险结果为处于高风险的农村1-3岁和4-6岁幼儿,其%aPAD分别为115.74和110.43,在概率评估中99.9%位数的%aPAD结果分别为24.3和22.2；99%位数的%aPAD结果急剧下降,分别为1.818和1.936；95%位数的%aPAD结果与99%位数结果相近,分别为1.449和1.320。可以看出概率评估结果比较切合实际,而且从概率评估中也可以看出菠菜中毒死蜱急性毒性风险比较高的人群也是安全的。
2.点评估与概率评估结果相同,即各年龄、地域亚人群组中农村1-3岁和4-6岁幼儿为风险最高人群,其次是城市1-3岁和4-6岁幼儿,点评估的结果要比概率评估结果高2-160倍不等,因此点评估是非常保守的。对于保护大多数人来说,其结果可以作为初步参考或确定是否需要进行概率评估的筛选措施。在整个评估过程中也可以看出概率评估所需要的数据量以及软件和硬件、时间的投入都要比点评估要高的多。点评估和概率评估都是膳食暴露评估的两种重要手段,而且不可以相互代替。当点评估结果显示风险较小时就没有必要进行概率评估。
3.从我国人口全食品中毒死蜱急性膳食暴露量概率评估的风险估计结果来看,处于风险最高的亚人群组依次为城市1-3岁幼儿、农村1-3岁和4-6岁幼儿%aPAD分别为91.197、82.895、和83.255各数值接近但均小于100,这是在没有考虑安全因子的情况下评估的结果,可以认为我国各人群组食品中毒死蜱的暴露处于安全剂量,同时在亚人群中城市和农村1-6岁幼儿仍然是我们毒死蜱农药残留关注的重点,考虑到本研究第一章中毒死蜱在菠菜中的加工因子研究,建议广大消费者食用经过清洗和漂烫过的蔬菜,毒死蜱的含量会大大下降。
With the increasing population of the world, all government are facing the problem with how to solve food supply and food safety properly, and therein pesticide has been playing an important role in protecting crop and enhancing yield, however, all people concern pesticides residue in food and their toxicity. There are hundreds of pesticides, such as phosphate pesticide, chlorinated pesticide, etc. As a major food exporting country, China is frequently facing technical trade barrier with pesticide residues set up by some importing countries during international trade, therefore, based on scientific data, developing dietary exposure assessment for pesticides residues under general guideline to total exposure assessment and evaluating health risk from pesticides residue will provide powerful scientific basis for management department to take more effective measures in order to break down technical trade barrier from importing countries and protect domestic consumer benefit. This study does have important academic value and practical benefits.
This study builds exposure assessment model for pesticides residue in food based on data from national pollutant monitoring network in 2000-2006, monitoring data for pesticides residue in export food in 2005-2006, monitoring data for pesticides residue in frozen spinach exported from Shandong Province to Japan in 2005, the data of national dietary survey in 2002. This model includes point estimation and probabilistic assessment model, etc. and this model is built on data available and analysis on variability and uncertainty. This paper takes chlorpyriofos as example to study exposure risk of chlorpyriofos in Chinese dietary structure, difference between point estimation and probabilistic assessment of pesticides residue in diet, difference of exposure assessment for cholorpyriofos residue in export vegetable and domestic market and analysis methods for variability and uncertainty by probabilistic assessment. Most important achievement as below:
Ⅰ. Based on research on analysis method about pesticide residue in food and statistical analysis on pesticide residue data in food, conclusions are drawn as the following:
1. The detection process for anilazine GC-MS has been established, which proves that multiple residual detection method of high flux for pesticide is as important as its validation method. The process has good performance in linear relationship, the range of recovery rate and detection limit can meet standard from anilazine detection method for import and export food in China
2. The positive rate of pesticide is higher in animal food than in vegetable food. Therefore the pesticide residues in animal food should be paid more attention, namely the pesticide that is accumulated in animal body needs special attention. Meanwhile, concentrated pesticide has a relatively higher positive rate, whereas the positive rate of pesticide in food product that undergone deep processing is lower than that of the primary farm product. "Stem vegetable", "Berries fruit", "Beans", "Leafy vegetable", "Brassica vegetable", and "Tea" are rank 1 to 6 respectively in the list of positive rate of chlorpyrifos,. The fact that "Bulb vegetable", "Stem vegetable", "Leafy vegetable" and "Tuber vegetable" are all found on the list of non-conformity chlorpyrifos content in food category in 2003-2006. This shows that these vegetable categories should be highlight of vegetable safety supervision in China.
3. The distribution of chlorpyrifos by region in 2003-2006 in China indicates that chlorpyrifos residue is higher in the southern area than that in the northern area in China. The top 4 at P97.5 by region are Zhuhai, Fujian, Zhejiang, Yunnan, respectively, which are all located in the South of China. The top list of the maximum chlorpyrifos residue are Chongqing, Hubei, Guangxi, Guangdong respectively. This result could be related to the climate difference between southern and northern part of China.
4. As per monitoring data we collected in all groups of food, due to different objectives and different methods and different sample size may lead to different lowest detection limit. Therefore given the case of 98% residue is non-detectable, the lowest detection limit of the non-detected will be extremely important, especially when the sample size is relatively limited and this may lead to outliers. This shows that quality of data will have direct impact on assessment result and therefore assessment results will be much more reliable if data are representative and collected with clear objective.
Ⅱ. Through point estimation of dietary exposure assessment for chlorpyrifos residue in spinach for export and for domestic consumption, conclusions are drawn as below:
1. As per acute exposure assessment for chlorpyrifos:Ⅰ. the %PAD of chlorpyrifos is relatively higher in the group aged 1-6, amongst the %aPAD of the rural group aged 1-3 is 115.74 and the %PAD of the rural group aged 4-6 is 110.43 which is above 100, it shows that this group is high risk population and probably due to higher dietary intake per Kg body weight by children;Ⅱ. The %aPAD in male groups is relatively lower than that in female groups, and this could be that vegetable consumption in female groups is higher than in male group;Ⅲ. Based on assessment by region, the results show that the %aPAD of rural population in the children and adolescent group under 18 is apparently higher than urban group, and the figures in adult aged above 18 in rural group is also a bit higher than in urban group. The conclusion is related to the total vegetable consumption, differences in diet structure and body weight;Ⅳ. Based on assessment by total population, the %aPAD of chlorpyrifos is 43.09<100 which is safe..
2. As per chronic exposure assessment of chlorpyrifos:Ⅰ. The max %cPAD in urban children is 6.59 which is far below than 100 when non-detected chlorpyrifos residue replaced by LOQ, therefore it is not necessary to put the chronic assessment of chlorpyrifos under highlight;Ⅱ. The %cPAD of chlorpyrifos in spinach is relatively higher in female group than in male group and this is aligned with the result from acute assessment;Ⅲ. For children aged 1-3 and 7-10, the %cPAD in urban group is higher than in rural group, however, for people in other age group, the %cPAD in rural group has higher figure than in urban group;Ⅳ. For different age group, if non-detected chlorpyrifos residue is replaced by LOQ and 0, the %cPAD from point estimation varies from 5-9 times.
3. Comparison of assessment results from the acute and chronic exposure assessment of chlorpyrifos:acute and chronic exposure point estimation shows that the %aPAD of any group is much higher than the %cPAD. The %aPAD of the total population is 43.05 and while the %cPAD is 2.94. The risk from acute exposure assessment is 15 times higher than chronic exposure assessment. Therefore, acute toxicology of chlorpyrifos should be highlight of research and management in upcoming works.
4. Difference between spinach exported to Japan and spinach for domestic consumption:results show that main difference is from process factors due to variability in China dietary structure. If Chinese consumer can follow the procedure to wash and bleach vegetable carefully, the %aPAD is 17.22 and the %cPAD is 0.69 for domestic consumer and %aPAD is 16.2 and %cPAD is 0.4 for export spinach for Japan, all data is within safe range.
Ⅲ. Conclusions are drawn based on the probabilistic assessment for exposure of chlorpyrifos in diet:
1. When non-detected data for chlorpyrifos in spinach are replaced with 0, 1/2LOQ or LOQ, probabilistic assessment shows the younger the age is, the higher for exposure of chlorpyrifos as per Kg body weight, group aged 1-6 has the highest exposure, and group aged 4-6 has the maximum exposure at percentile of 99.99. The exposure of chlorpyrifos in spinach also varies from region to region, as a result, the rural group has a relatively higher exposure than urban group and this indicates that a higher chlorpyrifos residue in food in rural area. However, the daily exposure of chlorpyrifos by age and region is below the limit and far less than ARfD.
2. The distribution of exposure of chlorpyrifos in spinach is a left skew distribution. The quartile range P75-P25 shows that 50% of the population in the middle has less variability. Different exposure of chlorpyrifos by region indicates that rural group has higher level of exposure than urban group at percentile of P99, P99.9, P99.99. Where exposure level of rural group higher than urban group, there is high level of percentile (e.g. P99), there is nearly 18 times of exposure level than where percentile is low (e.g. P25). Exposure is a left skew distribution and chlorpyfrifos residue in rural area is much more serious than in urban area.
3. The uncertainty analysis from probabilistic assessment on chlorpyrifos residue in spinach and in food shows that all percentiles of variability are within the 95% confidence interval. It shows 1 million sample size is sufficient to stablize the exposure percentiles and the Monte Carlo simulation has quantified the variability. If we put aside the uncertainty issue, the Monet Carlo percentiles are sufficiently stable for actual application. Meanwhile, the uncertainty analysis also shows that the higher the percentile is, the broader 95% confidence interval is and the higher the uncertainty is.
4. If non-detected data is replaced with LOQ in the probabilistic assessment of chlorpyrifos in all food, the result shows the variability of the exposure in total population groups at P99.99 is 11.139 which is lower than ARfD (ARfD=100). Subgroup aged 1-3 and 4-6 has the highest exposure as 17.544 which is lower than ARfD. From the variability of exposure of chlorpyrifos in urban and rural area, it concludes that exposure in rural area is higher than in urban area and maximum exposure in rural is 30.02 which is less than ARfD.
5. The food contribution analysis in the acute probabilistic assessment shows that, based on food types, the top 3 of chlorpyrifos intake in total population, urban group and rural group are:vegetable and its product, fruit and its product, potato starch and its product. In order to describe the specific contribution of chlorpyrifos intake by subgroup of food, the top 3 contributors of chlorpyrifos intake in the total population are:"young stem, leaves and flower vegetables"; "solanaceous fruit vegetable, and cucurbit"; "fresh bean". And their contribution is 81.75%,5.94% and 2.75% respectively with a sum of 90.44%. Food contribution from all group and subgroup with high exposure is same than that in total population, and this is aligned with the left skew distribution of exposure of chlorpyrifos in diet.
6. Due to food processing factors and the half life of chlorpyrifos are not considered during chronic dietary exposure assessment, results show that variability at each percentile in chlorpyrifos are all less than the ADI value, however, if the FQPA coefficient is taken into account, variablity will be close to alerting limit, and this is not in accord with chronic toxicology and symptoms of chronic toxicity of chlorpyrifos. Thus we concluded that chronic toxicology probabilistic assessment needs more data support.
7. When non-detected data of chlorpyrifos residue is replaced with 0,1/2LOQ or LOQ respectively, we found that it has significant effect on the mean value, however difference becomes smaller and smaller at high percentiles such as P99.9 or even P99.99.
Ⅵ. Comparison between the results of point estimation and probabilistic assessment
1. Risk assessment of chlorpyrifos in spinach and in all food shows that point estimation result from total population is that rural group aged 1-3 and 4-6, which the %aPAD is 115.74 and 110.43 respectively. The %aPAD at 99.9% percentile is 24.3 and 22.2 respectively in probabilistic assessment, and the %aPAD at 99% percentile is as small as 1.449 and 1.320 respectively. The figure shows that results from probabilistic assessment is reasonable and high risk population from acute toxicology assessment is still safe.
2. The point estimation has the same result than probabilistic assessment, namely the rural group aged 1-3 and 4-6 are high risk population, and the urban group aged 1-3 and 4-6 next to it. Result of point estimation is 2-160 times higher than that of probabilistic assessment, so point estimation is quite conservative. From the point of protecting as much people as we can, result could be used as primary reference or as criteria for judging whether probabilistic assessment is necessary. Across the assessment process, we can clearly see that probabilistic assessment requires more data, software and hardware, time and resource than the point estimation. Those two methods are very important for dietary exposure assessment and can not replace each other. The probabilistic assessment is not necessary when the point estimation shows relatively little risk.
3. According to risk assessment from the probabilistic assessment of acute dietary exposure assessment of chlorpyrifos in all food in China, the sub-groups with highest risk are the urban group aged 1-3, rural group aged 1-3 and 4-6, and the %aPAD is 91.197,82.895 and 83.255 respectively and all are close but less than 100. Safety factors are not taken into account for this case, therefore we concluded that exposure of chlorpyrifos for each group is within safety range. Meanwhile children aged 1-6 either in rural or urban area should be highlight in research of chlorpyrifos residue. Given research on processing factors provided in the chapter 1, we suggest consumers wash and bleach before have it and this enables great reduction of chlorpyrifos residue in food.
引文
1.陈君石.危险性评估与食品安全.中国食品卫生杂志,2003,15(1)：3-6.
2.陈振德,陈雪辉,冯明祥等,毒死蜱在菠菜中的残留动态研究[J].农业环境科学学报2005,24(4)：728-731
3.蔡雪亚,金丕焕,等.用Bootstrap方法计算中位数的可信区间[J].中国卫生统计.2002,19(3)：185-186.
4.陈红.Bootstrap方法及其应用[J].青岛大学学报：工程技术版.1997,12(3)：78-83.
5.洪锡熙译[美]詹姆斯.R.埃文斯戴维.L.奥儿森著《模拟与风险分析.Simulation and Risk Analysis》
6.韩晓梅,闵捷等中国膳食暴露评估模型桥梁数据库构建初探[J]卫生研究Vol.37 No.6 Nov.2008 725-727.
7.金同铭.菠菜的营养与药用[J].蔬菜,2000,(1)：32.
8.刘沛,刘元宝,王灿楠.膳食暴露评估模型及构建方法[J].中华预防医学杂志,2007,41(6)：502-504.
9.刘勤,金丕焕.Bootstrap方法及其在医学统计中的应用[J].中华预防医学杂志.1998：52-53.
10.李靖欣,陈炳为,王灿楠等.SAS/AF及其在开发中国膳食暴露评估模型软件中的应用[J].中国卫生统计,2008,25(6)：642-647.
11.潘守奇,朱莉萍,孙军,等.果蔬及其制品中甲胺磷、乙酰甲胺磷、氧化乐果、久效磷残留量检测[J].现代仪器.2008,14(4)：64-67.
12.裴鹿成.蒙特卡罗方法及其在粒子输运问题中的应用[M].北京：科学出版社,1980.
13.吴永宁主编现代食品安全科学[M]化学工业出版社2003.72-109.
14.王大宁主编,罗祎执行主编等食品安全风险分析指南[M]中国标准出版社,2004.59-121.
15.王绪卿,吴永宁.食品污染监测低水平数据处理问题[J].中华预防医学杂志.2002, 36(4)：278-279.
16.徐钟济.蒙特卡罗方法[M].上海：上海科学技术出版社,1985.
17.徐秦.Monte-Carlo方法的应用[J].中国卫生统计.1996：57-58.
18.杨月欣王光亚等主编中国疾病预防控制中心营养与食品安全所编著中国食物成分表2002北京大学医学出版社
19.杨月欣主编中国疾病预防控制中心营养与食品安全所编著中国食物成分表2004北京大学医学出版社
20.张琼,李凡群.Bootstrap方法在无失效数据分析中的应用[J].安庆师范学院学报：自然科学版.2006,12(3)：26-28.
21.现代数学手册.随机数学卷[M].武汉：华中科技大学出版社,2000.
22.Acceptable Daily Intake:Inception, evolution[J]. Regul Toxicol Pharmacol.1998,8: 45-60.
23.A FRAMEWORK FOR CUMULATIVE RISK ASSESSMENT http:// rsi.ilsi.org/file/rsiframrpt.pdf.
24.Bosgra, S., Van Eijkeren, J.C.H., Van der Schans, M.J., Langenberg, J.P., Slob, W., 2009.Toxicodynamic analysis of the combined cholinesterase inhibition by paraoxon and methamidophos in human whole blood. Toxicol. Appl. Pharmacol.236,9-15.
25.Clower M.Pesticide- residues in food, United- States Food and Drug Administrations program for immunoassay[J].Acs Symposium Series,1991,451:49-58
26.Chicharro M, Zapardiel A, Bermejo E, et al.Determination of 3-amino- 1,2,4-triazole(amitrole) in environmental waters by capillary electrophoresis[J].Talanta,2003(59):37-45
27.Burmaster, D.E., Anderson, P.D,1994 Principles of good practice for the use of Monte Carlo techniques in human health and ecolo-gical risk assessments. Risk Analysis 14,477-481.
28.Copeland, T.L., Holbrow, A.M., Otani, J.M., Connor, K.T., Paus-tenbach, D.J.,1994. Use of probabilistic methods to understand the conservatism in California's approach to assessing health risks posed by air contaminants.Air Waste 44,1399-1413.
29.Cassin, M.H., Paoli, G.M., Lammerding, A.M.,1998. Simulation modeing for microbial risk assessment. Journal of Food Protection 61,1560-1566.
30.Cullen A C, Frey H C. Probabilistic Techniques in Exposure Assessment:A Handbook for Dealing with Variability and Uncertainty in Models and Inputs[R].,1999.
31.Finley, B., Proctor, D., Scott, P., Harrington, N., Paustenbach, D., Price, P.,1994. Recommended distributions for exposure factors frequently used in health risk assessment. Risk Analysis 14,533-553.
32.Ferrier H, Nieuwenhuijsen M, Boobis A,et al.Current knowldege and recent developments in consumer wxposure assessment of pesticides:a UK perspective [J].Food Additives and Contaminants,2002,19(9):837-852.
33.FELSOT A.Pesticide risk assessment-dietary, aggregate, cumulative[R/OL] http://feql.wsu.edu/entom558/fall2005entom/110705Pt1ADOC.pdf,2005-11-7.
34.Ferrier H, Nieuwenhuijsen M, Boobis A. Current knowledge and recent developments in consumer exposure assessment of pesticides:a UK perspective[J]. Food additives and contaminants.2002:837-852.
35.Gibney M J, Van Der Voet H. Introduction to the Monte Carlo project and the approach to the validation of probabilistic models of dietary exposure to selected food chemicals[J]. Food additives and contaminants.2003:1-7.
36.Hoover, S.M.,1999. Exposure to persistent organochlorines in Cana-dian breast milk: a probabilistic assessment. Risk Analysis 19,527-545.
37.Hamey,P.Y.,2000 A practical application of probabilistic modeling in assessment of dietary exposure of fruit consumers to pesticide residues. Food Additives and Contaminants 17,601-610.
38.JENSEN A F, PETERSEN A, GRANBY K. Cumulative risk assessment of the intake of organophosphorus and carbamate pesticides in the Danish diet[J]. Food Additives and Contaminants,2003,20(8):776-785.
39.Lambe.J.,Cadby,P.,Gibney,M.,2001.Comparison of stochastic stances with theoretical added maximum daily intakes (TAMDI)and maximized survey-derived intakes(MSDI). Food Additives and Contaminants 19,3-15
40.Mckone, T.E., Bigen, K.T.,1992 Uncertainties in health-risk assessment:an integrated case study based on tetrachloroethylene in California groundwater. Regulatory Toxicology and Pharmacology 15,86-103.
41.Mcnamara C, Naddy B, Rohan D. Design, development and validation of software for modelling dietary exposure to food chemicals and nutrients[J]. Food additives and contaminants.2003:8-26.
42.Miyamoto K-, Kotake M. Estimation of daily bisphenol a intake of Japanese individuals with emphasis on uncertainty and variability[J]. Environmental sciences:an international journal of environmental physiology and toxicology.2006:15-29.
43.Paustenbach, D.J., Meyer, D.M., Sheehan, P.J., Lau, V.,1991. An assessment and quantitative uncertainty analysis of the health risks to workers exposed to chromium contaminated soils. Toxicology and Industrial Health 7,159-196.
44.Price,P.S.,Curry,C.L.Goodrum,P.E.,Gray,M.N.,McCrodden, J.I., Harrington,N.W., Carlson-Lynch,H.,Keenan, R.E.,1996 Monte Carlo modeling of time-dependent exposure using a micro-exposure event approach. Risk Analysis 16,339-348.
45.Price, P., Chaisson, C.F., Young, J.S., Christensen, C., Doyle, E., Suhre, F.B.,1999. Background document for the Session:Review of an Aggregate Exposure Assessment Tool FIFRA Scientific Advi-sory Panel. The LifeLineTM Project to model Aggregate Exposure to Pesticides, Arlington,WA.
46.Petersen, B.J.,2000. Probabilistic modelling:theory and practice. Food Additives and Contaminants 17,591-599.
47.Price, S.P., Young, J.S., Chaisson, C.F.,2000. Assessing Aggregate and Cumulative Pesticide Risks Using LifeLineTM Version 1.0. Areport submitted to EPA Science Advisory Panel August 31 2000.
48.Report of the Joint FAO/WHO Expert Consultation, Application of risk analysis to food standards issues; Geneva, Switzerland,13-17,March,1995;
49.Rees, N.M., Day, M.J.,2000. UK consumption databases relevant to acute exposure assessment. Food Additives and Contaminants 17,575-581.
50.R.McKinlay,J.A.Plant,J.N.B.Bell,N.Voulvoulis calculating human exposure to endocrine disrupting pesticides via agricultural and non-agricultural exposure routes Science of the total environment 398(2008)1-12.
51.Richardson, J.R., Chambers, H.W., Chambers, J.E.,2001. Analysis of the additivity of in vitro inhibition of cholinesterase by mixtures of chlorpyrifos-oxon andazinphos-methyl-oxon. Toxicol. Appl. Pharmacol.172,128-139.
52.Slob, W., Krajnc, E.,1994. Interindividual variability in modeling exposure and toxicokinetics:a case study on cadmium. Environ-mental Health Perspectives 102,78-81.
53.SUHRE F B. Pesticide residues and acute risk assessment-the US EPA approach[J]. Food Additives and Contaminants,2000,17(7):569-573.
54.http://cares.ilsi.org/file/background_miller.pdf.
55.The EFSA Journal(2007)538,1-88 opinion of the scientific panel on plant protection products and their residues on a request from the commission on acute dietary intake assessment of pesticide in fruit and vegetables adopted on 19 April 2007 http://www.efsa.europa.eu.
56.http://library.wur.nl/wasp/bestanden/luwpubrd_00319014_A502_001.pdf.
57.http://www.maff.go.jp/j/zyukyu/fbs/index.html
58.Zartarian V G, Xue J, Ozkaynak H A, et al. A Probabilistic Arsenic Exposure Assessment for Children Who Contact Chromated Copper Arsenate (CCA)-Treated Playsets and Decks, Part 2:Sensitivity and Uncertainty Analyses[J]. Risk Analysis.2006: 533-541.
59.Zheng J, Frey H C. Quantitative analysis of variability and uncertainty with known measurement error:methodology and case study.[J]. Risk analysis:an official publication of the Society for Risk Analysis.2005:663-675.
2.陈振德,陈雪辉,冯明祥等,毒死蜱在菠菜中的残留动态研究[J].农业环境科学学报2005,24(4)：728-731
3.蔡雪亚,金丕焕,等.用Bootstrap方法计算中位数的可信区间[J].中国卫生统计.2002,19(3)：185-186.
4.陈红.Bootstrap方法及其应用[J].青岛大学学报：工程技术版.1997,12(3)：78-83.
5.洪锡熙译[美]詹姆斯.R.埃文斯戴维.L.奥儿森著《模拟与风险分析.Simulation and Risk Analysis》
6.韩晓梅,闵捷等中国膳食暴露评估模型桥梁数据库构建初探[J]卫生研究Vol.37 No.6 Nov.2008 725-727.
7.金同铭.菠菜的营养与药用[J].蔬菜,2000,(1)：32.
8.刘沛,刘元宝,王灿楠.膳食暴露评估模型及构建方法[J].中华预防医学杂志,2007,41(6)：502-504.
9.刘勤,金丕焕.Bootstrap方法及其在医学统计中的应用[J].中华预防医学杂志.1998：52-53.
10.李靖欣,陈炳为,王灿楠等.SAS/AF及其在开发中国膳食暴露评估模型软件中的应用[J].中国卫生统计,2008,25(6)：642-647.
11.潘守奇,朱莉萍,孙军,等.果蔬及其制品中甲胺磷、乙酰甲胺磷、氧化乐果、久效磷残留量检测[J].现代仪器.2008,14(4)：64-67.
12.裴鹿成.蒙特卡罗方法及其在粒子输运问题中的应用[M].北京：科学出版社,1980.
13.吴永宁主编现代食品安全科学[M]化学工业出版社2003.72-109.
14.王大宁主编,罗祎执行主编等食品安全风险分析指南[M]中国标准出版社,2004.59-121.
15.王绪卿,吴永宁.食品污染监测低水平数据处理问题[J].中华预防医学杂志.2002, 36(4)：278-279.
16.徐钟济.蒙特卡罗方法[M].上海：上海科学技术出版社,1985.
17.徐秦.Monte-Carlo方法的应用[J].中国卫生统计.1996：57-58.
18.杨月欣王光亚等主编中国疾病预防控制中心营养与食品安全所编著中国食物成分表2002北京大学医学出版社
19.杨月欣主编中国疾病预防控制中心营养与食品安全所编著中国食物成分表2004北京大学医学出版社
20.张琼,李凡群.Bootstrap方法在无失效数据分析中的应用[J].安庆师范学院学报：自然科学版.2006,12(3)：26-28.
21.现代数学手册.随机数学卷[M].武汉：华中科技大学出版社,2000.
22.Acceptable Daily Intake:Inception, evolution[J]. Regul Toxicol Pharmacol.1998,8: 45-60.
23.A FRAMEWORK FOR CUMULATIVE RISK ASSESSMENT http:// rsi.ilsi.org/file/rsiframrpt.pdf.
24.Bosgra, S., Van Eijkeren, J.C.H., Van der Schans, M.J., Langenberg, J.P., Slob, W., 2009.Toxicodynamic analysis of the combined cholinesterase inhibition by paraoxon and methamidophos in human whole blood. Toxicol. Appl. Pharmacol.236,9-15.
25.Clower M.Pesticide- residues in food, United- States Food and Drug Administrations program for immunoassay[J].Acs Symposium Series,1991,451:49-58
26.Chicharro M, Zapardiel A, Bermejo E, et al.Determination of 3-amino- 1,2,4-triazole(amitrole) in environmental waters by capillary electrophoresis[J].Talanta,2003(59):37-45
27.Burmaster, D.E., Anderson, P.D,1994 Principles of good practice for the use of Monte Carlo techniques in human health and ecolo-gical risk assessments. Risk Analysis 14,477-481.
28.Copeland, T.L., Holbrow, A.M., Otani, J.M., Connor, K.T., Paus-tenbach, D.J.,1994. Use of probabilistic methods to understand the conservatism in California's approach to assessing health risks posed by air contaminants.Air Waste 44,1399-1413.
29.Cassin, M.H., Paoli, G.M., Lammerding, A.M.,1998. Simulation modeing for microbial risk assessment. Journal of Food Protection 61,1560-1566.
30.Cullen A C, Frey H C. Probabilistic Techniques in Exposure Assessment:A Handbook for Dealing with Variability and Uncertainty in Models and Inputs[R].,1999.
31.Finley, B., Proctor, D., Scott, P., Harrington, N., Paustenbach, D., Price, P.,1994. Recommended distributions for exposure factors frequently used in health risk assessment. Risk Analysis 14,533-553.
32.Ferrier H, Nieuwenhuijsen M, Boobis A,et al.Current knowldege and recent developments in consumer wxposure assessment of pesticides:a UK perspective [J].Food Additives and Contaminants,2002,19(9):837-852.
33.FELSOT A.Pesticide risk assessment-dietary, aggregate, cumulative[R/OL] http://feql.wsu.edu/entom558/fall2005entom/110705Pt1ADOC.pdf,2005-11-7.
34.Ferrier H, Nieuwenhuijsen M, Boobis A. Current knowledge and recent developments in consumer exposure assessment of pesticides:a UK perspective[J]. Food additives and contaminants.2002:837-852.
35.Gibney M J, Van Der Voet H. Introduction to the Monte Carlo project and the approach to the validation of probabilistic models of dietary exposure to selected food chemicals[J]. Food additives and contaminants.2003:1-7.
36.Hoover, S.M.,1999. Exposure to persistent organochlorines in Cana-dian breast milk: a probabilistic assessment. Risk Analysis 19,527-545.
37.Hamey,P.Y.,2000 A practical application of probabilistic modeling in assessment of dietary exposure of fruit consumers to pesticide residues. Food Additives and Contaminants 17,601-610.
38.JENSEN A F, PETERSEN A, GRANBY K. Cumulative risk assessment of the intake of organophosphorus and carbamate pesticides in the Danish diet[J]. Food Additives and Contaminants,2003,20(8):776-785.
39.Lambe.J.,Cadby,P.,Gibney,M.,2001.Comparison of stochastic stances with theoretical added maximum daily intakes (TAMDI)and maximized survey-derived intakes(MSDI). Food Additives and Contaminants 19,3-15
40.Mckone, T.E., Bigen, K.T.,1992 Uncertainties in health-risk assessment:an integrated case study based on tetrachloroethylene in California groundwater. Regulatory Toxicology and Pharmacology 15,86-103.
41.Mcnamara C, Naddy B, Rohan D. Design, development and validation of software for modelling dietary exposure to food chemicals and nutrients[J]. Food additives and contaminants.2003:8-26.
42.Miyamoto K-, Kotake M. Estimation of daily bisphenol a intake of Japanese individuals with emphasis on uncertainty and variability[J]. Environmental sciences:an international journal of environmental physiology and toxicology.2006:15-29.
43.Paustenbach, D.J., Meyer, D.M., Sheehan, P.J., Lau, V.,1991. An assessment and quantitative uncertainty analysis of the health risks to workers exposed to chromium contaminated soils. Toxicology and Industrial Health 7,159-196.
44.Price,P.S.,Curry,C.L.Goodrum,P.E.,Gray,M.N.,McCrodden, J.I., Harrington,N.W., Carlson-Lynch,H.,Keenan, R.E.,1996 Monte Carlo modeling of time-dependent exposure using a micro-exposure event approach. Risk Analysis 16,339-348.
45.Price, P., Chaisson, C.F., Young, J.S., Christensen, C., Doyle, E., Suhre, F.B.,1999. Background document for the Session:Review of an Aggregate Exposure Assessment Tool FIFRA Scientific Advi-sory Panel. The LifeLineTM Project to model Aggregate Exposure to Pesticides, Arlington,WA.
46.Petersen, B.J.,2000. Probabilistic modelling:theory and practice. Food Additives and Contaminants 17,591-599.
47.Price, S.P., Young, J.S., Chaisson, C.F.,2000. Assessing Aggregate and Cumulative Pesticide Risks Using LifeLineTM Version 1.0. Areport submitted to EPA Science Advisory Panel August 31 2000.
48.Report of the Joint FAO/WHO Expert Consultation, Application of risk analysis to food standards issues; Geneva, Switzerland,13-17,March,1995;
49.Rees, N.M., Day, M.J.,2000. UK consumption databases relevant to acute exposure assessment. Food Additives and Contaminants 17,575-581.
50.R.McKinlay,J.A.Plant,J.N.B.Bell,N.Voulvoulis calculating human exposure to endocrine disrupting pesticides via agricultural and non-agricultural exposure routes Science of the total environment 398(2008)1-12.
51.Richardson, J.R., Chambers, H.W., Chambers, J.E.,2001. Analysis of the additivity of in vitro inhibition of cholinesterase by mixtures of chlorpyrifos-oxon andazinphos-methyl-oxon. Toxicol. Appl. Pharmacol.172,128-139.
52.Slob, W., Krajnc, E.,1994. Interindividual variability in modeling exposure and toxicokinetics:a case study on cadmium. Environ-mental Health Perspectives 102,78-81.
53.SUHRE F B. Pesticide residues and acute risk assessment-the US EPA approach[J]. Food Additives and Contaminants,2000,17(7):569-573.
54.http://cares.ilsi.org/file/background_miller.pdf.
55.The EFSA Journal(2007)538,1-88 opinion of the scientific panel on plant protection products and their residues on a request from the commission on acute dietary intake assessment of pesticide in fruit and vegetables adopted on 19 April 2007 http://www.efsa.europa.eu.
56.http://library.wur.nl/wasp/bestanden/luwpubrd_00319014_A502_001.pdf.
57.http://www.maff.go.jp/j/zyukyu/fbs/index.html
58.Zartarian V G, Xue J, Ozkaynak H A, et al. A Probabilistic Arsenic Exposure Assessment for Children Who Contact Chromated Copper Arsenate (CCA)-Treated Playsets and Decks, Part 2:Sensitivity and Uncertainty Analyses[J]. Risk Analysis.2006: 533-541.
59.Zheng J, Frey H C. Quantitative analysis of variability and uncertainty with known measurement error:methodology and case study.[J]. Risk analysis:an official publication of the Society for Risk Analysis.2005:663-675.