Dataset of chemical concentration levels in food from the second French total diet study

<strong>Information about the TDS2 data set </strong> <em>1. Description of the study</em> The TDSs are based on a standardized method and have been recommended by the World Health Organization (WHO) and the European Food Safety Authority (EFSA) for many years. They consist in collecting food products representative of the population's consumption, preparing these foods "as consumed" taking into account population’s common practices, combining them as composite/pooled samples, analyzing these samples, assessing the population's exposure to the targeted substances and finally assessing the risk for substances for which reference values exist. Within the framework of the 2nd French TDS (TDS2), nearly 20,000 food products were collected in about 30 cities throughout the French metropolitan territory and prepared to form 1,319 samples. These products corresponded to 212 types of food representing nearly 90% of the diet of adults and children in France. Each sample was composed of 15 sub-samples of the same food (same label) and mass, allowing for the representation of different brands and taking into account consumer food preferences (product origin, varieties, brands, preparation methods, places of purchase, etc.). The sub-samples prepared "as consumed" were thus representative of the consumption of the food. With a few exceptions, all samples were replicated twice during the study to cover potential seasonal variability in composition or contamination. Different foods were also collected in different regions of France, to take account of potential regional differences in contamination. The More details about the study methodology are provided in the TDS2 reports: https://www.anses.fr/en/system/files/PASER2006sa0361Ra1EN.pdf https://www.anses.fr/en/system/files/PASER2006sa0361Ra2EN.pdf <em>2. Description of the data set</em> The dataset presents data at the food item level, i.e. seasonal sample values were averaged at the regional level, then regional sample values were averaged to represent the national level. The original regional dataset is available in French at the following URL https://www.data.gouv.fr/fr/datasets/r/06797551-c1ad-40c5-8527-b3515d28fecd. The TDS2 data set includes 173 substances: trace elements, minerals, environmental pollutants, mycotoxins, neoformed compounds, additives and phytoestrogens. The analytical methods used for each family of substances and the analytical limits (LOD and LOQ ) are described in the study reports. 2.1. Pre-processing of concentration data Three hypotheses for the management of left censored data (concentration value below the analytical limits) were applied for all substances: The lowerbound (LB) hypothesis: Concentrations below the LOD (undetected substances) were replaced by 0, and concentrations below the LOQ but above the LOD (known as "traces") were replaced by the LOD value; The middlebound (MB) hypothesis: Concentrations below the LOD were replaced by ½ LOD, and concentrations below the LOQ but above the LOD were replaced by ½ (LOD+LOQ) ; Upperbound (UB) hypothesis: Concentrations below the LOD have been replaced by the LOD value, and concentrations below the LOQ but above the LOD have been replaced by the LOQ. In order to use the concentration value to assess exposures in relation to the available toxicological reference values, some data were processed. Thus, for mercury and arsenic, substances analysed in their total form, speciation assumptions were applied to the concentrations in order to estimate the organic and inorganic part in each sample (cf. study report, volume 1, p35). Similarly, for some compounds, sums were calculated. These are for example the sums of dioxins and furans, weighted by the toxic equivalency factors of 2005, or the sums of the 3 congeners of polybrominated biphenyls (PBBs). For these groups, data are available both for the substance details and for the sums of the three PBB congeners. 2.2. Thesaurus he list below presents the variables in the data set and their definitions: family: name of the chemical group of the substance; subst: abbreviated name of the substance; subst_sn: short name of the substance; subst_cn: complete name of the substance; unit: unit of the mean concentration value of the substance; mean_CONTA_LB: mean of the concentration values of sub-samples of the analyzed food calculated under the lower bound hypothesis; mean_CONTA_MB: mean of the concentration values of sub-samples of the analyzed food calculated under the middle bound hypothesis; mean_CONTA_UB: mean of the concentration values of sub-samples of the analyzed food calculated under the upper bound hypothesis; Food_code: identification number of the analyzed food; Food_group: food category of the analyzed food; Type_foods: analyzed food.<strong>TDS2数据集说明</strong> <em>1. 研究概况</em> TDS系列方法基于标准化流程制定，多年来被世界卫生组织（World Health Organization, WHO）与欧洲食品安全局（European Food Safety Authority, EFSA）推荐采用。其核心流程为：采集具有人群消费代表性的食品样品，按照人群日常食用方式进行“即食制备”，将样品组合为复合/混合样本，对样本开展实验室分析，评估人群对目标物质的膳食暴露量，最终针对存在毒理学参考值的物质开展风险评估。 在法国第二届TDS项目（TDS2）中，研究人员在法国本土约30座城市采集了近20000份食品样品，经标准化制备后形成1319份分析样本。这些样品涵盖212类食品，覆盖法国成人与儿童近90%的膳食摄入结构。每份分析样本由15份同品类、同标签、同质量的子样本组成，可覆盖不同品牌的同类产品，并兼顾消费者的食品选择偏好（如产品产地、品种、品牌、制备方式、购买渠道等）。经“即食制备”的子样本可准确代表该食品的实际消费形态。除少数例外情况外，所有样本均进行了两次重复检测，以覆盖成分或污染水平的潜在季节性差异。同时，研究人员在法国不同地区采集不同品类食品，以考量污染水平的潜在区域差异。更多研究方法细节可参阅TDS2项目官方报告：https://www.anses.fr/en/system/files/PASER2006sa0361Ra1EN.pdf https://www.anses.fr/en/system/files/PASER2006sa0361Ra2EN.pdf <em>2. 数据集说明</em> 本数据集提供食品单品层面的统计数据：首先将季节性样本的检测值按区域维度进行平均，再将各区域的平均样本值进一步整合以代表全国整体水平。原始区域数据集可通过以下法语链接获取：https://www.data.gouv.fr/fr/datasets/r/06797551-c1ad-40c5-8527-b3515d28fecd。 TDS2数据集共包含173种被测物质，涵盖微量元素、矿物质、环境污染物、真菌毒素、新形成化合物、食品添加剂与植物雌激素。针对每一类物质所使用的分析方法及分析限值（检测限（Limit of Detection, LOD）与定量限（Limit of Quantitation, LOQ））已在研究报告中详细说明。 2.1 浓度数据预处理 针对所有被测物质，研究人员采用三种假设方案处理左截尾数据（即浓度值低于分析限值的检测结果）： - 下界（Lowerbound, LB）假设：将低于检测限（未检出）的浓度值替换为0，将低于定量限但高于检测限的浓度值（即“痕量样本”）替换为检测限值； - 中界（Middlebound, MB）假设：将低于检测限的浓度值替换为½倍检测限，将低于定量限但高于检测限的浓度值替换为½(检测限+定量限)； - 上界（Upperbound, UB）假设：将低于检测限的浓度值替换为检测限值，将低于定量限但高于检测限的浓度值替换为定量限值。 为基于浓度值评估与现有毒理学参考值相关的膳食暴露量，研究人员对部分数据进行了额外处理：例如，对于总形态分析的汞与砷，研究人员采用形态分配假设对原始浓度值进行换算，以估算每份样本中的有机态与无机态占比（详见研究报告第一卷第35页）。类似地，部分复合污染物需计算总和指标：例如以2005年毒性当量因子加权的二噁英与呋喃总和，或多溴联苯（Polybrominated Biphenyls, PBBs）的3种同系物总和。针对这类物质组，数据集同时提供单物质详情数据与3种PBB同系物的总和统计数据。 2.2 变量与术语说明 以下列表展示了数据集中的变量及其定义： - family：物质所属化学类群的名称； - subst：物质的缩写名称； - subst_sn：物质的短名称； - subst_cn：物质的完整名称； - unit：物质平均浓度值的计量单位； - mean_CONTA_LB：基于下界假设计算得到的分析食品子样本浓度平均值； - mean_CONTA_MB：基于中界假设计算得到的分析食品子样本浓度平均值； - mean_CONTA_UB：基于上界假设计算得到的分析食品子样本浓度平均值； - Food_code：分析食品的唯一识别编号； - Food_group：分析食品所属的食品类别； - Type_foods：分析食品的具体名称。