Slob Wout
National Institute of Public Health and the Environment (RIVM), 3720 BA Bilthoven, The Netherlands.
Food Chem Toxicol. 2006 Jul;44(7):933-51. doi: 10.1016/j.fct.2005.11.001. Epub 2006 Feb 3.
Probabilistic dietary exposure assessments that are fully based on Monte Carlo sampling from the raw intake data may not be appropriate. This paper shows that the data should first be analysed by using a statistical model that is able to take the various dimensions of food consumption patterns into account. A (parametric) model is discussed that takes into account the interindividual variation in (daily) consumption frequencies, as well as in amounts consumed. Further, the model can be used to include covariates, such as age, sex, or other individual attributes. Some illustrative examples show how this model may be used to estimate the probability of exceeding an (acute or chronic) exposure limit. These results are compared with the results based on directly counting the fraction of observed intakes exceeding the limit value. This comparison shows that the latter method is not adequate, in particular for the acute exposure situation. A two-step approach for probabilistic (acute) exposure assessment is proposed: first analyse the consumption data by a (parametric) statistical model as discussed in this paper, and then use Monte Carlo techniques for combining the variation in concentrations with the variation in consumption (by sampling from the statistical model). This approach results in an estimate of the fraction of the population as a function of the fraction of days at which the exposure limit is exceeded by the individual.
完全基于从原始摄入量数据进行蒙特卡洛抽样的概率性膳食暴露评估可能并不合适。本文表明，首先应使用能够考虑食物消费模式各个维度的统计模型来分析数据。讨论了一种（参数化）模型，该模型考虑了（每日）消费频率以及消费量方面的个体间差异。此外，该模型可用于纳入协变量，如年龄、性别或其他个体属性。一些示例说明了如何使用此模型来估计超过（急性或慢性）暴露限值的概率。将这些结果与基于直接计算观察到的摄入量超过限值的比例的结果进行比较。这种比较表明，后一种方法并不充分，特别是在急性暴露情况下。提出了一种用于概率性（急性）暴露评估的两步法：首先通过本文讨论的（参数化）统计模型分析消费数据，然后使用蒙特卡洛技术将浓度变化与消费变化相结合（通过从统计模型中抽样）。这种方法得出了作为个体超过暴露限值的天数比例的函数的人群比例估计值。
