【作者】 蒋定国；

【导师】 吴永宁； 刘沛； 王竹天；

【作者基本信息】 中国疾病预防控制中心 ， 营养与食品卫生学， 2008， 博士

【摘要】 暴露评估是食品安全风险评估的重要阶段，对于发展中国家尤为重要。暴露评估模型包括点评估和概率评估等，这些模型的使用依赖于所获得的评估数据及其对于评估参数变异性和不确定度分析。本论文以铅和镉作为典型的重金属污染物，在构建中国食品污染监测和食物消费量2个数据库与暴露评估模型的基础上，分别采用点评估和概率评估模型进行铅、镉的暴露评估实证研究，通过不同模型的比较研究，不仅获得了平均暴露水平和高暴露水平的参数，而且获得了参数变异性和不确定度等资料。本研究包括食品污染物监测数据、食物消费量数据和暴露评估模型研究等三个方面，其主要研究内容和结果如下：1利用污染物监测数据库开展铅、镉污染的点评估参数探索性分析本研究对全国污染物监测数据进行了严格和规范的数据整理，进行了统一的食品分类和食品编码，在此基础上建立了我国目前最大的食品中铅和镉含量数据库，其中包括了我国主要食品中31883条铅含量基础数据和29580条镉含量基础数据。使用SAS软件对我国36类食品中铅和镉的污染状况进行了统计分析，并建立了点评估的污染物浓度系列参数。铅：从平均值来看，污染水平排在前三位的食品是皮蛋、茶叶、干食用菌，其铅含量分别为1．782mg/kg、1．762mg/kg和0．392mg/kg。污染水平较高的食品有10类，其平均值都高于0．1mg/kg，其余23类食品的污染水平较低。从超标率来看，粮食类超标率大于7．9%，其中大米及制品超标率为11．5%；奶类、软饮料类、水果和蔬菜类中除了叶类蔬菜、鳞茎蔬菜和浆果及其他小水果的超标率低于10%外，其余都在11%～23%之间；皮蛋、禽畜肉类的超标率都大于11%；其余食品的超标率低于10%。从高百分位数P90、P95和P97．5来看，除了调味品、酒类、罐头食品、茶叶、鲜奶及其液态制品、奶粉、干食用菌和鲜食用菌等8类消费量较少的食品的高百分位数值低于国家限量值外，其它28食品的高百分位数值均大于国家限量值。镉：从平均值来看，污染水平最高的食品是动物肾脏，高达2．691mg/kg。软体贝类、干食用菌、甲壳类、畜禽内脏(除肾脏)和鲜食用菌中污染水平亦较高，分别为0．534mg/kg、0．362 mg/kg、0．350mg/kg、0．21 3 mg/kg和0．114 mg/kg。污染水平处于中等水平的食品有4类，其平均值都高于0．05mg/kg。其余26类食品的污染很低。从超标率来看，肾脏高达30．49%，食用菌为19．22%，甲壳类为16．07%，软体贝类12．1%，其余食品低于10%。从高百分位数P90、P95和P97．5来看，有9类食品的高百分位数值低于国家限量值，20类食品的高百分位数值大于国家限量值。目前尚未制定饮料类、蜜饯类、调味品、茶叶和奶类等食品的限量标准，需要予以关注。2利用食物消费量数据库开展点评估参数的探索性分析采用与食品污染物监测数据库相一致的食品编码和分类规则，对2002年中国居民营养与健康状况调查所得的1390种食品的消费量资料进行数据清理和数据合并，建立了我国最大的食物消费量数据库，其中包括1643859条膳食消费量有效数据。使用SAS软件，对城乡不同人群的膳食模式和食物消费量进行统计分析，获得了点评估所需的各种参数如人群平均体重、各类食品的平均消费量和高百分位数等。从平均值来看，农村人口的谷类、薯芋类和蔬菜类食物的日均消费量较高，而城市人口的水果、肉类、水产类、蛋类、奶类、豆类和饮料类等食物的消费量较大。从全国总体看，粮食类如米、面、薯和豆类的消费量最高，蔬菜消费量居第二，肉蛋奶消费量第三，水果和水产类第四，其它食物的消费较少。从消费量的各百分位数来看，对于P50而言，人群消费食品的类别非常单一，如1～6岁的儿童只消费粮食类食物，18岁以上年龄段的男性还增加少量叶类蔬菜和肉类的摄入，而其他7岁以上人群也仅仅摄入粮食类和少量叶类蔬菜。在P90、P95、P97．5处，食品消费多样性和数量逐渐增加。对于主要食品而言，高百分位数的数值大约是均值的2～5倍，而对于平均消费量少的食品来说，高百分位数的数值可能是均值的10倍以上，差异非常显著。3铅和镉的膳食摄入量的点评估在上两部分工作的基础上，使用点评估模型的计算公式，分别计算出全国、城市和农村各9个年龄—性别组铅、镉的各类食品摄入量和膳食总摄入量，并经分析发现以下规律：(1)年龄越小，暴露量越高。即儿童摄入量高于成人，属于高危人群。农村人群的摄入量高于城市人群；男性的摄入量高于女性。按平均值计算出的摄入量通常是按中位数计算出的摄入量的2～3倍；从高百分位数得到的摄入量显著增高，以致大多数人群组别的计算结果大大超过PTWI值。(2)按均值—均值计算出的暴露量：10岁以下儿童的铅摄入量已在PTWI的70%～98%之间，11岁以上各人群的铅摄入量在PTWI的40%～60%之间；10岁以下儿童的镉摄入量已经大大超过PTWI，11～14岁男性的镉摄入量也已超过PTWI(101%)，其它年龄—性别组的镉摄入量在PTWI的70%～90%之间。(3)按不同食品类别对各年龄性别组暴露量的贡献大小分析：①铅：粮食类的贡献最大，占总摄入量的约50%，蔬菜占25%左右，肉蛋奶占12%左右，水果和水产类占6%左右，以上几类食物的贡献总共占总摄入量的96%以上，其它食品的贡献不足4%；②镉：也是粮食类的贡献最大，占总摄入量的约65%，蔬菜占14%左右，肉蛋奶和水产类占总摄入量都小于10%，以上几类食物的贡献总共占总摄入量的95%以上，其它食品的贡献不足5%。4铅和镉的膳食摄入量的概率评估(1)构建了概率评估模型所需三部分数据(即膳食调查数据、污染物监测数据和用于在膳食数据和污染物监测数据间建立联系的桥梁数据)的总数据库。(2)运用Monte carlo模拟方法获得了城市、农村和全国不同年龄—性别组的铅、镉的膳食总摄入量的平均值和各百分位数的概率分布，以及这些摄入量分布的中位数等基本统计量。结果表明，我国儿童的铅和镉膳食暴露的平均水平已不安全，高分位数的暴露量处于更高水平。应尽快采取必要措施加以控制食品的铅、镉等重金属污染，以保护绝大部分人群的健康。以高危人群，即10岁以下儿童为对象重点分析了铅、镉摄入量的变异性。1～10岁各年龄组儿童铅、镉摄入量的分布统计量描述显示，均数、各百分位数的概率分布的中位数和均数基本一致，标准差也较小。从四分位间距和p95：p5比值来看，其变异性变化也不大，均值和各百分位数比较稳定，说明Monte carlo方法能很好地量化摄入量的变异性。以4～6岁年龄组儿童为例，重点分析摄入量的概率分布与PTWI的关系。从均数分布看，4～6岁儿童每日铅暴露均值的概率分布属于偏态分布，对城市而言，在分布的高端处已超过PTWI，而对农村而言，在分布的中间就已超过PTWI，表明4～6岁儿童铅暴露的平均水平已不安全。从p75分布看，城市在分布的高端处超过PTWI，而农村则在分布的低端处即已超过PTWI。至于p90、p95和p99的概率分布，所有暴露量都超过PTWI。从均数分布看，4～6岁城乡儿童每日平均镉暴露的概率分布为近似正态分布，其概率分布的所有暴露量均超过PTWI；从高百分位数的分布看，p75、p90、p95的城乡共6个概率分布的所有暴露量都已超过PTWI。(3)运用Bootstrap方法和Monte carlo方法获得了1～10岁儿童的铅暴露量的平均值和各百分位数的可信区间和四分位间距，显示均数和各百分位数比较稳定，数据分布集中，表明Monte carlo模拟较好地量化了变异性。同时发现百分位数越高，95%可信区间越宽，不确定性越大。与p50比较，均数的不确定性较大，农村儿童暴露量均数的95%可信区间甚至已与p97．5的95%可信区间相当，说明采用均数进行相关计算，所得结果的不确定性可能较大。(4)点评估结果与概率评估结果的比较两种评估方法所得到的不同人群组暴露量的基本趋势是一致的。从均值看，点评估结果和概率评估的结果基本一致，但点评估结果略低于概率评估结果。点评估在低百分位数得到的暴露量较低而高百分位数得到的暴露量则明显偏高，不确定性较大，而概率评估的结果比较切合实际。对于保护大多数人来说，点估计结果可以作为初步参考，在制定标准和政策应更需要考虑概率性评估结果。5小结本课题对我国各类食品中铅和镉的污染水平，我国不同人群铅和镉的膳食暴露量，以及膳食暴露的点评估模型和随机概率评估模型等进行了较系统的研究。在建立我国食品铅和镉污染监测数据库、食物消费量数据库和暴露评估模型的基础上，对我国居民膳食的铅、镉暴露分别进行了点评估和概率评估。结果表明，我国部分人群，尤其是儿童的膳食铅、镉暴露已经处于较不安全的状况，应予以足够的重视。点评估和概率评估所得到的结果基本一致，但点评估结果的变异和不确定性较大。故点估计结果可作为初步参考，而在制定标准和政策时应重点考虑概率评估的结果。本研究结果可为全面、正确地评价铅和镉对我国居民的健康风险并提出针对性的预防措施提供重要的依据。更多还原

【Abstract】 Exposure assessment is one of the key parts of the risk assessment process in foodsafety,which is especially important to developing countries.Exposure assessmentmodels include such as point estimate and probabilistic assessment,etc.Applicationsof models rely on the assessment data obtained,variability and uncertainty analysis ofassessment parameters.In this study,the dietary exposure assessments for lead andcadmium,as representative heavy metal contaminations,were carried out by the pointestimate and probabilistic assessment models,on the basis of establishing of the Chinafood contamination monitoring database,Chinese food consumption database andexposure assessment models,and the contrastive studies of different models werefinished to obtain the parameters of mean exposure level and high exposure level,andalso to obtain the results of variability and uncertainty of the parameters.Threeresearch programs of lead and cadmium were carried out in details which include inthe analysis of contamination monitoring data of them in food and food consumptiondata,and the dietary exposure assessments for lead and cadmium.Ⅰ.Study on point estimate parameters of lead and cadmium by establishingfood contamination monitoring databaseIn this part,national food contamination data were collected strictly and normatively,and checked up.Uniform food classification and food coding of the data werecoordinated with Chinese food consumption data.As a result,the largest foodcontamination database of lead and cadmium in China was built,including 31883 dataon content of lead and 29580 data on content of cadmium in main foods.Lead and cadmium contamination were explored in 36 groups of foods in China andanalyzed statistically by SAS software to establish a series of parameters for pointestimate.Lead:Average level of lead contamination in preserved egg was the highest (1.782mg/kg),followed by tea (1.762 mg/kg) and dried edible fungus (0.392mg/kg).Average levels of lead in ten groups of foods exceeded 0.1mg/kg,but the contamination in the other 23 groups of foods were lower.It showed that the violatedrate of foodstuff exceeded 7.9%,among them the rice and its product was 11.5%.Except that the violated rate of leaf vegetables,bulb vegetables and berries wereunder 10%,the violated rate of milk and its products,beverages,fruits andvegetables were 11%～23%.The violated rate of preserved egg,meat of poultry andlivestock exceeded 11%,while the violated rate of the rest foods were all under 10%.The result of analyzing for high percentile showed that there were 28 groups offoods whose lead contamination exceeded the national limit,while the rest wereunder the standard.Cadmium:Average levels of cadmium in animal kidney was the highest (2.691mg/kg),followed by bivalve mollusks and dried edible fungus,crustaceans,offal oflivestock and poultry and fresh edible fungus separately with concentration of0.534mg/kg,0.362 mg/kg,0.350 mg/kg,0.213 mg/kg and 0.114 mg/kg.Averagelevels of cadmium in four groups of foods exceeded 0.05 mg/kg,but thecontamination in the other 26 groups of foods were lower.Cadmium contamination inanimal kidney exceeded the national limit standard,and the violated rate of sampleswas 30.49%.The violated rates of the edible fungus,crustaceans,and bivalvemollusks were separately 19.22%,16.07% and 12.1%,and those of the rest foodswere all under 10%.The result of analyzing for high percentile showed that therewere 20 groups of foods whose contamination exceeded the national limit standard,while 9 groups of the rest were under the standard.At present,there are not limitstandards about beverages,confects,spices,tea and milk and its products in China,and it should be paid more attention.Ⅱ.Study on point estimate parameters of food consumption by establishingfood consumption databaseAfter collection and verification of food consumption data from the national survey ofthe nutrition and health status of the Chinese people in 2002,food classification andfood coding for 1390 foods were carried out in accord with national foodcontamination database,and the largest Chinese food consumption database was built,including 1643859 data of food consumption.Analysis for dietary pattern and food consumption of different population from urban and rural areas were carried out by SAS software,and a series of pointestimate parameters of food consumption were established such as mean bodyweights of population,mean values and high percentiles of food group’sconsumptions.The result of analyzing for mean value showed that daily consumptions of cereals,potatoes and vegetables of rural population were higher than urban population,andby contraries,daily consumptions of fruit,meat,aquatic product,eggs,dried beanproduct,and beverage of urban population were higher.For all groups of the Chinesepopulation,the consumptions of cereals,such as rice,flour,as well as potato andpulse was the highest,and followed by vegetables,and then meat egg and milk,fruitand aquatic product,while the consumptions of other foods were very low.The result of analyzing for food consumptions percentiles showed that there wereonly the consumptions of foodstuff in P50 of children from 1-6 year old,and formale population above 18 year old,only a little leaf vegetable and a little meatconsumption was increased,and for other age group population,only a little leafvegetable consumption was increased.In P90,P95,and P97.5,the consumptionvarieties and amounts of foods gradually increased.High percentiles in main foodswere about 2-5 times of mean values,and those in low consumption foods wereprobably over 10 times of mean values.Ⅲ.Point estimate for dietary exposure of lead and cadmiumOn the basis of above two parts of work,dietary intakes of lead and cadmium fromsingle food group and all food groups about 9 age-sex groups in national,urban andrural populations were calculated by the formula of point estimate model.The resultsshowed as following:1.The smaller was the age,the higher was the dietary intake of Pb and Cd.Childrenwere more sensitive and their dietary intakes were very high.The dietary intake ofrural populations was usually higher than that of urban populations,and the dietaryintake of male was usually higher than female.The mean value of dietary intake wasusually 2-3 times higher than median,and the high percentiles of dietary intakeswere very high,which for most of populations greatly exceeded PTWI. 2.The result of analyzing for dietary intakes from mean contamination-meanconsumption combination showed that the dietary intakes of lead of children under10 years old were between 70% and 98% of PTWI,and those of children over 11years old were between 40% and 60% of PTWI.The dietary intakes of cadmium ofchildren under 10 years old greatly exceeded PTWI,those for male over 11-14 yearsold was 101% of PTWI and those of other age-sex populations were between 70%and 90% of PTWI.3.The result of analyzing for contribution of food group to dietary intakes showedthat the dietary intake of lead in staple foodstuff such as rice,flour,as well as potatoand pulse was the highest,which was about 50% of total intake,and those ofvegetable was about 25% of total intake.The dietary intake of lead in meat,egg andmilk was about 12%,and that in fruit and aquatic product was about 6%,and thosein all of other food groups were under 4%;The dietary intake of cadmium infoodstuff was also the highest,which was about 65% of total intake,and that invegetable was about 14% of total intake,and that in meat,egg,milk and aquaticproduct was all under 10%,and those in all of other food groups were under 5%.Ⅳ.Study on probabilistic assessment for dietary exposure of lead andcadmium1.The total database of probabilistic assessment model,including three parts of data,i.e.food consumption data,chemical concentration data and bridge data forconnecting food consurnption data with chemical concentration data was established.2.The distributions of the rnean and percentiles of dietary intakes about lead andcadmium of different age-sex groups in national,urban and rural populations werecalculated by Monte Carlo simulation,and obtained the medians of thosedistributions.The results showed that mean dietary intakes of lead and cadmium ofChinese children were high and unsafe,and moreover,the percentiles of dietaryintakes were higher.To protect most of populations’ safety,the essential measures tocontrol the food contaminated by heavy metals should be carried out as quickly aspossible.Variability of dietary intakes of sensitive children aged 1-10 were quantified by Monte Carlo methods.The basic statistics of the dietary intakes distributions of themean and the percentiles about lead and cadmium of children aged 1-10 showed thatthe medians of those distributions were basically accordant with their mean values,and standard deviation were small.The quartile spaces and the rate of P95:P5suggested that the variability of dietary intakes were also small.These resultsshowed that the variability of dietary intakes was quantified well by Monte Carlomethods,and the mean and the percentiles were stable.The relation of the dietary intakes distributions and PTWI were analyzed for childrenaged 4-6.For lead,the dietary intakes distributions of the mean weren’t normal,whose high percentile exceeded PTWI in urban children,but whose middlepercentile exceeded PTWI in rural children.The dietary intakes distributions of P75exceeded PTWI in urban children at high percentile,but at low percentile exceededPTWI in rural children.For p90,p95 and p99,their whole distributions all exceededPTWI.For cadmium,the dietary intakes distributions of the mean were normal forurban and rural children,which entirely exceeded PTWI.The dietary intakesdistributions of P75,p90 and p95 all exceeded PTWI too.3.The reliable spaces and quartile spaces of the mean and the percentiles of the leaddietary intakes of children aged 1-10 were obtain by Monte Carlo method andBootstrap method,which showed that the mean and the percentiles were stable andtheir distributions were concentrative,which suggested that the variability wasquantified well by Monte Carlo simulation.Simultaneity,the higher was thepercentile,the broader were the reliable spaces of 95%,and the larger was theuncertainty too.The uncertainty of the mean was larger than P50,and the 95%reliable space of the mean for rural children was almost equivalent to that of P97.5,which suggested that the uncertainty of result calculated from the mean might belager.4.Comparing with the results of point estimateThe trend of the dietary intakes from two assessment models was basically accordantfor different group’s populations.The mean values from two models were basicallyaccordant,but the results of point estimate model were a little lower than probabilistic assessment model.The dietary intakes from low percentiles were lower,but that from high percentiles were higher in point estimate model,which indicatedthat the results of probabilistic assessment model were relatively suitable to reality.To protect most of populations’ safety,the results of point estimate might be as initialreference,but the results of probabilistic assessment should be more taken intoaccount when establishing standards and policies.ⅤSummaryIn this study,the systematic analysis on the lead and cadmium contaminationsituations in various groups of foods in China,the dietary intakes of different group’spopulations in China,point estimate model and probabilistic assessment model ofdietary exposure were carried out.On the basis of establishing of the China nationalfood contamination monitoring database,and Chinese food consumption database aswell as exposure assessment models,point estimate and probabilistic assessment fordietary exposure of lead and cadmium in Chinese people were finished.The resultsfrom this study showed the dietary exposures of lead and cadmium in parts ofpopulations in China,especially children,were high and unsafe,to which should bepaid enough attention.The results of the dietary intakes from point estimate andprobabilistic assessment was basically accordant,but the variability and uncertainty ofpoint estimate were larger.Therefore,the results of point estimate might be as initialreference,but the results of probabilistic assessment should be more taken intoaccount when establishing maximum limits and control policies.The results of thisstudy can provide the important science base for comprehensively and accuratelyassessing the health risk of lead and cadmium in Chinese people and putting forwardeffective and precautionary measures.更多还原