【作者】 何梅；

【导师】 杨月欣；

【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】 卫生部在2007年1 2月份颁布了《食品营养标签管理规范》（以下简称《规范》）,是我国食品工业发展的里程碑,是我国食品营养标签由混乱无序的状况向有序的法制化管理迈出的关键性一步,是营养科学的一次实践,涉及了营养学基本定义、检测方法和数据评价等。其中食品营养成分的检测和合理标示是食品营养标签顺利实施的保障和前提。由于我国法规标准中有关食物营养的标准还很薄弱,国内实验室对营养成分检测的开展也十分有限,因此对实验室检测营养成分的准确度和误差的研究以及对营养成分在标签上的标示评价尚属空白。研究目的通过营养成分的检测标准体系比对、实验室检测误差的研究以及标签标示值的风险研究,了解我国营养成分检测体系的现况和实验室的检测误差,并评估在此基础上标示的风险,为我国《食品营养标签管理规范》的顺利实施提供可行性建议。研究内容和方法1.关键营养成分检测标准体系的比对研究以营养成分检测标准体系中检测方法、食物采集方法和有证标准物质三个重要组成部分作为比较对象,通过网络查询、书籍、会议交流和讨论以及专家咨询和访谈相结合的方式获得国内外相关检测体系的比较信息。对CODEX、AOAC以及我国国标中的8种营养成分（蛋白质、脂肪、碳水化合物、水分、灰分、钠、钙、胆固醇）的检测方法在数量、适用的食品种类、检测原理以及质量控制等方面规定的差异。比较CODEX、AOAC以及我国国标中食物采集方法在数量以及食品种类等方面的差异;同时也对美国、加拿大以及我国的营养标签法规要求的样本采集量进行了比较。此外,还比较了8个国家的有证标准物质在研发数量、食物基质分布以及营养成分特性值分布方面的差异。以观察我国在营养成分检测标准体系的框架建设、检测方法的特异性、灵敏性、准确性和检测条件要求等方面的现况、优势和不足。2.关键营养成分的实验室检测误差研究本研究按照消费者摄入频率高、食物基质多样化、营养成分含量范围宽的标准,确定7类14种包装食品作为实验室的分析样品。选择我国东北、西部、东南地区的7个通过省市级以上计量认证的实验室,用国标检测方法对食物样品中的脂肪、蛋白质、钠、钙、胆固醇的含量进行三次平行样品的测定。同时各实验室平行测定水分和灰分1次,由中国CDC营养与食品安全所统一测定总膳食纤维的含量。利用各实验室的结果计算14种食品中的碳水化合物含量和能量值,并计算能量、脂肪、蛋白质、碳水化合物、钠、钙、胆固醇的实验室检测误差和95%相对扩展不确定度(U_0.95)。在检测食物样品之前,课题组统一发放2个CRM作为盲样,测定其总氮、钠和钙的含量,以部分了解实验室的检测水平和现状。3.能量及核心营养成分标签标示值的风险评估通过实验室联合分析、数据库查询和计算、食品企业的调研和网络/科学文献检索相结合的方式,确定实验室检测结果与标签标示值间的误差来源和等级。并按照《规范》中对标示值变化范围的要求,以样本量为3、6或12,对能量及核心营养成分在添加或强化以及自然存在形式下的标签标示值的风险概率进行评估。并结合误差数据,探讨标示值的确定方法及降低风险的途径。研究结果1.关键营养成分检测标准体系的比对研究我国的检测标准体系的建设已经有一定的基础,表现在:1)国家标准中关于8个营养成分（蛋白质、脂肪、水分、灰分、碳水化合物、钙、钠、胆固醇）的检测方法有52个,食物采集方法有6个;2)检测方法和食物采集方法在食物原料上占优势,尤其是食用菌系列;3)检测方法的原理与CODEX和AOAC方法的基本一致,说明国标方法的灵敏度和检测条件基本与国外一致;4)我国食品分析用CRM的食物基质达到有21种。但我国的检测标准体系的建设还不健全,表现在:1)我国国标中8个营养成分的检测方法和食物采集方法的数量分别比CODEX和AOAC少37%和61%;2)检测标准体系的特异性不高:国标中对于复杂加工的食品种类还没有独立的检测方法和食物采集方法;针对普通食品检测方法的适用范围规定模糊;食品分析用CRM在食物基质种类和特性值指标上存在一定的重复和单一;3)体系的准确性不高:我国营养标签要求的样本量略少于美国和加拿大的相应要求;国标检测方法基本没有针对检测的室内重复性精密度和实验室再现性精密度范围进行规定;食品分析用CRM有特性值的营养成分集中在矿物质和总氮,相对单一。2.关键营养成分实验室的检测误差在检测关键营养成分（蛋白质、脂肪、钠、钙、胆固醇）时,7个实验室中有1个不能完全按照适用的食品种类选择相应的检测方法;所有实验室在检测14种不同食品的成分时,仅使用1个测定方法,而没有做到“如有多个平行方法时,可根据适用范围选择适宜的方法”进行检测。实验室对2个盲样（均是有证标准物质）的6次总氮、钠和钙的检测合格率分别为55.9%、61.1%和52.6%,且市级实验室的检测合格率基本低于55%。反映出实验室对营养成分检测的准确度差。7个实验室检测关键营养成分的误差高于AOAC相应方法的要求,尤其实验室间分析误差占总误差的80%以上。实验室检测的精密度误差与成分的含量呈现负相关。检测结果的95%相对扩展不确定平均为:能量4.55%,脂肪20.08%,蛋白质7.47%,碳水化合物7.80%,钠29.95%,钙39.10%,胆固醇45.83%。与《食品营养标签管理规范》要求的20%标示值允许变化范围相比,仅蛋白质的检测误差在范围之内。3.能量及核心营养成分标示值的风险评估标签标示值与检测值之间的误差来自于实验室内检测误差、实验室间检测误差、食品生产误差和食物原料营养成分含量的波动,其中来自于实验室间检测误差或食物原料成分含量波动的比例最大可达到78.82%和93.89%,认为后二者是误差的主要来源。对营养标签中能量和核心营养成分标示值的两类风险评估表明,企业风险（P_α）和消费者风险（P_β）随着样本量n、相对偏差δ的增加、以及总误差的降低而迅速降低。认为增加样本量至12,和/或降低误差范围,尤其是降低实验室间检测误差和食物原料成分的波动,是降低标示值风险的最直接及有效的二个途径。当n=12,δ=10%时,在同样的误差条件下,自然存在的非限制性摄入的营养成分标示值的企业风险概率最低,同时限制性摄入的营养成分标示值的企业风险概率最高。而消费者风险概率略低于相应的企业风险概率约1%。根据本研究获得的误差数据,结合《规范》中对标示值变化范围的要求,如果以检测值的平均值进行标示,当n=12时,自然存在的能量及核心营养成分标示值的P_α分别为:能量25.0%、脂肪和钠24.3%、蛋白质5.4%、碳水化合物19.3%。相应的消费者风险概率略低1%左右。如果以检测值的95%可信限的下限（蛋白质、碳水化合物）或上限（能量、脂肪、钠）进行标示,则能量及核心营养成分的P_α大为降低,分别为:能量9.4%、脂肪和钠2.6%、蛋白质0.2%、碳水化合物4.9%。研究结论1.我国对关键营养成分的检测标准体系的建设已经有了一定的基础,但还不健全。国标检测方法的灵敏度和检测条件基本与CODEX和AOAC方法的要求一致,但在食品适用范围的特异性方面还有待完善和提高,且检测方法的准确性和质量控制要求还有待进一步提高。2.我国实验室对营养成分检测的开展现状不容乐观,在检测方法的选择上没有完全按照食品种类进行,对营养成分检测的合格率低,实验室间检测误差大,见测准确度有待进一步提高。3.营养成分的实验室间分析误差和食物原料成分含量的波动是标签标示值误差的主要来源。能量和核心营养成分标示值的风险概率随着样本量n、相对偏差δ的增加、以及总误差的降低而迅速降低。在目前误差条件下,当以检测结果的平均值进行标示,在n=12时,能量及核心营养成分标示值的最大风险分别为:能量25.0%、脂肪和钠24.3%、蛋白质5.4%、碳水化合物19.3%。当以检测值的95%可信限进行标示时,则标示值的风险概率大为降低。本研究有以下的建议:1.在我国营养标签实施过程中,应大力推广使用营养成分国标分析方法,着力建设各类深加工食品的特定检测方法和食物采集方法,完善方法的质量控制要求,发展食品分析用有证标准物质的研制和生产,以健全和发展我国营养成分的检测标准体系。2.鉴于食品营养标签在我国暂时属于自愿执行,而目前核心营养成分的实验室分析误差和食物原料含量的波动都比较大,建议监管部门在执行《规范》过程中,应以指导食品企业统一规范的标示营养标签为主要执法手段。本研究的创新之处:1.本研究比较了8个关键营养成分的检测标准体系与国外的差距,研究结果将为《规范》的实施和发展提供可行性建议。2.本研究获得了实验室对关键营养成分检测误差的一手资料,将对我国普及实验室营养成分检测工作起积极的推动作用。3.本研究首次开展了我国《营养标签管理规范》执行的相关评估研究,其结果将为《规范》在监督检测领域的实施方向提供科学、客观的参考数据,同时也为食品企业进行标签的标示提供可行性建议。4.本次评估推进了我国科学建设法规标准的改革之路,是我国科学发展观的体现。更多还原

【Abstract】 “The regulation of food nutrition labeling management”(Thereinafter briefly named“the regulation”)has been established and announced by Chinese Ministry of Health in December 2007.It is the milestone for food industry in China,and the key step that make the chaos situation for our country’s food nutrition labeling to become orderly managed,it is the practice of nutritional science.It involves basic definition for nutrition terms,analysis method,data evaluation,etc.Nutrients analysis and reasonable declaration are the guarantee and prerequisite for nutrition labeling implementation.Because regulations and standards about nutrition in our country are still very weak and nutrients analysis in labs is also very limited,therefore the accuracy for nutrients measurement and nutrition labeling research in our country is still far from enough.ObjectivesTo assess the current status of nutrients analysis standard system in respects of analysis method,sampling method,and certificated reference materials in China.To assess the current status of detection level and error for key nutrients in labs in China.To assess the risk of declared value of energy and core nutrients on food nutrition labeling.Methods1.Comparing research on key nutrients analysis standard system The status,advantage,and disadvantage of Chinese national analysis standard system for nutrients were learned from comparing analysis standard methods,sampling methods,and certificated reference material(CRM).The differences were observed by analyzing methods for 8 nutrients(protein,fat, carbohydrate,moisture,ash,sodium,calcium,and cholesterol)among CODEX,AOAC and Chinese National Analysis Standards(Thereinafter briefly named GB methods)in respects of methods number,involved food groups, detection theory,and analysis error requirements.The differences were compared for food sampling methods among CODEX,AOAC and Chinese National Standards in respects of number,including food groups,and sampling amount.And meanwhile the difference was compared with CRMs in 8 countries in respects of number,food raw material,and nutrients values.2.The analytic variation for key nutrients in labs in China Seven kinds and fourteen prepackaged foods,which including cereals,beans, meats,dairies,beverages,nuts,and snacks were selected as food samples for analysis,according to the condition of high intake frequency,wide nutrient content range,and multi diversity of food basic material.Seven labs were selected to participate the collaborative analysis for 7 key nutrients(protein,fat, moisture,ash,sodium,calcium and cholesterol),according to analysis ability certification,regional distribution,and administration rank.Protein,fat,sodium, calcium and cholesterol of 14 food samples were required to three paralleled measures by using GB methods,while moisture and ash one paralleled measure.Total dietary fiber was detected by lab of Institute of Nutrition and Food Safety,China CDC.The concentration of carbohydrate and energy,and 95%relative uncertainty of analysis of all the analyzed and calculated nutrients were calculated by researching group.3.The risk assessment for declared values of energy and core nutrients The source and classification of variety between analysis result and declared value on labels were affirmed by combination of lab analysis,food composition database searches and calculation,food industry survey,and website or science literature searches.The risk of energy and core nutrients declared values was assessed under the added or fortified status,or naturally contained status.Then the declared methods and decreasing risk approaches were discussed combined with the error data.Results1.Comparison of analysis standard system The number of 8 nutrients analysis in GB methods is less than the ones of CODEX and AOAC.The involved food groups of GB methods were more integrity for raw material especially fungus than CODEX and AOAC methods, but with much less complicated manufactured food.There were 21 CRMs in China,but the raw food materials of them were repeatable and nutrients were singularly.This proved that the structure of GB methods has been basically established,but need to be more improved.The analyzed principal of GB methods was almost the same as CODEX and AOAC.This indicates that the sensitivity and detection condition for GB methods in China is almost the same as CODEX and AOAC.The involved detail food kinds were ambiguously described for the methods about general food.It indicated that the food involved specification for GB methods need to be improved.The sample number of Chinese NL demand is less than the ones of America and Canada NL.The analysis RSD of inner and inter labs for methods were almost absent in China.The raw food materials were repeatable and the special nutrients values were too singular for Chinese CRMs.These status showed the specificity and accuracy of analysis standard system in China need to be developed.2.The analysis variety in labs All labs only use one method to analyze a nutrient for all food samples.The E lab could not select methods according to the food group.This indicated the labs did not select all the methods according to the food group.The up to grade rates of total nitrogen,sodium and calcium analysis results for 2 single blind samples,which were all CRMs),were respectively 55.9%,61.1% and 52.6%,and less than 55%for city class labs.This indicates that the current status for labs in our country to conduct nutrient analysis is far from optimistic.The analysis precision of Chinese labs,representative by RSD,was less than AOAC methods prescription,especially RSD inter-labs contributing 80%in total analysis RSD of labs.The RSD is in negative correlation to concentration. The 95%relative accuracies of nutrients analysis results were energy 4.55%, fat 20.08%,total nitrogen 7.47%,carbohydrate 7.80%,sodium 29.96%,and calcium 39.10%.Only nitrogen analysis accuracy was in accordance with NL regulation.3.Risk assessment for declared values of energy and core nutrients The variety between declared value and detected value was called error,which consisted of analysis variety inner and inters labs,producing error,and raw material error.The most proportion of error from analysis inters labs and raw material was respectively 78.82%and 93.89%,and they were regarded as the main source for error.The risk rate decreased as sample number and/or relative difference between detective and declared values increase,and/or error decrease.It was regarded that increasing the sample number to 12,and decreasing the error range were the most two effective approaches. When sample number equal to 12 andδ=10%,the industry risk for non restrictively intake nutrients were lower than restrictively ones naturally contained in food,and risk for consumer is moderately 1%lower than for industry.If the declared value was the average analysis result,under the current error data from this study and the intolerance range of the NL Regulation,the industry risk rate for naturally contained energy and core nutrients were respectively energy 25.0%,fat and sodium 24.3%,protein 5.4%,carbohydrate 19.3%.And the consumer risk is 1%less than the industry risk.If the declared value was the upper limit(energy,fat,sodium)or the lower limit(protein, carbohydrate)of 95%credible range of analysis result,the industry risk rates were most decreased,and were respectively energy 9.4%,fat and sodium 2.6%,protein 0.2%,carbohydrate 4.9%.Conclusions1.The structure of GB method has been basically established in China,but far from enough.The sensitivity and detection condition is almost same as CODEX and AOAC,but the measurement method and quality control still need to be improved.2.The analysis status in China was not optimistic.The labs were not all selected methods according to the food group.The up to grade rates of nutrients analysis were low.The analysis variety inter labs were high.Apart from protein,the analysis accuracy of energy and core nutrients is out of intolerance range of NL Regulation.3.Error from analysis inters labs and raw material was regarded as the main source for total error.The risk rate was decreased as sample number and/or relative difference between detective and declared values increase,and/or error decrease.If the declared value was the average analysis result,under the condition of current error data from this study and the intolerance of the NL Regulation,the industry risk rate for naturally contained energy and core nutrients were respectively energy 25.0%,fat and sodium 24.3%,protein 5.4%, carbohydrate 19.3%.When using the upper limit(enegy,fat,sodium)or the lower limit(protein,carbohydrate)of 95%credible range of analysis result,the risk rates were most decreased.Advises1.We should greatly improve the GB analysis method for Nutrients and establish the methods for complicated manufactured foods,strictly demand the error range,and develop the CRMs in respects of food group and organic nutrients values when carrying out the NL regulation. 2.Since NL regulation is voluntary in China and error is large in labs analysis and food raw materials,the governmental supervising departments should currently emphasize on guiding the unification and legal nutrition labeling,but not punishment.更多还原