The European Flavours, Additives, and food Contact materials Exposure Task (FACET) project developed a freely downloadable  to probabilistically estimate dietary exposure to chemical substances from flavors, food additives and food contact materials (FCMs) in Europe (Oldring et al 2013b). The 4-year project lasted from 2008 to 2012 and was co-funded by the Directorate General of Research and Innovation (DG Research) of the European Commission (EC) (Oldring et al 2013b). The project consisted of 10 work packages (WPs) of which WP4 focused exclusively on primary packaging materials. The tool models exposure based on national dietary consumption data, the market share of different packaging types and sizes used for each food, the materials and structures which form these pack types and the usage and concentration of different substances in the materials. The model estimates the amount of substances transferring into the foods either by migration modeling or by use of actual measurements of migration into simulants or extraction solvents (Oldring et al 2013a).

The national consumption data were obtained from 8 European Member States and covered varying years (1992 to 2007). Whilst some of the 15 surveys included data on more than 1000 individuals over 7-8 days, including the weighing of consumed food, other surveys only covered a 24-hour period and were carried out by telephone.* The FACET program classified consumed foods into 174 different food groups at the most detailed tier of a three tier coding system. Higher tiers provide significantly more detail in terms of packaging type, materials, and substances used (Oldring et al 2013b). Information on the market share of different packaging types was obtained from  (2005 data on retail packaging type from 21 countries) and in addition from European can makers (EMPAC) and European coating suppliers (CEPE) for metal packaging materials. Information on substance application and relative use was obtained from the FACET Industry Group (FIG), made up of representatives from European FCM trade associations. The FIG also provided information on the substances usage, packaging construction, substance concentrations and physico-chemical parameters (Oldring et al 2013b). Currently, more than 6000 substances are listed in the tool, covering plastics, coatings, metals, paper and board, and inks and adhesives with detailed statistics on around 600 (Oldring et al 2013b). Some substances included are also in use as food additives or flavorings. Companies further provided around 400 material descriptions, as well as use conditions and pack sizes (Oldring et al 2013b). All these data were uniformly coded in order to link substance use to materials, and those in turn to packaging composition, application, pack size, market shares and eventually consumption (Oldring et al 2013b). Finally, for can coatings, the FIG supplied in-house migration data on the materials used, which were either based on extraction or migration tests into simulants (Oldring et al. 2013a). In order to estimate migration for the large variety of different FCMs including multi-layer materials, FCMs were clustered according to similar according to similar chemical and physical properties (polarity, diffusion behavior). Migration was then modeled employing physico-chemical parameters (Seiler et al. 2013). Around half of the total 6475 substances listed have substance-specific partition coefficients and a molecular weight; for substances lacking the necessary thermodynamic information and chemical property parameters, users may input their own data (Oldring 2013b). All this information is combined in a probabilistic tool, which employs Monte Carlo iterations to estimate exposure to different FCMs (Oldring et al 2013b). Results of different runs were found to match down to 2 to 3 decimal places, suggesting high internal model reproducibility (SmithersPira workshop, 2013). In order to verify the model, Oldring and colleagues compared the probabilistic FACET results of bisphenol A (BPA) exposure for British adults aged 19-64 to a refined deterministic estimate (2013a). The deterministic model was based on minimum and maximum levels of extractable BPA supplied by industry, and included only those material groups where BPA-containing FCMs accounted for more than 50% of market share (Oldring 2013a). Consumption data was based on the probabilistic model. Oldring and colleagues confirmed that their probabilistic estimate of overall BPA exposure lay in between refined deterministic estimates using the same software (135 ng/kg bw/day versus 37 to 322 ng/kg bw/day) (Oldring et al 2013a). The European Food Safety Authority’s (EFSA) recently published draft estimate of total BPA exposure amounted to 132 ng/kg bw/day ( Oldring et al. concluded that as their probabilistic model using FACET only addressed light metal packaging, it could overall be considered conservative for estimating BPA exposure. According to the authors, the BPA case study demonstrated the FACET model’s internal consistency, as well as its robustness when certain probabilistic parameters are replaced with deterministic estimates.

The FACET tool is the first pan-European model estimating exposure to food contact substances using probabilistic modeling. As such, it develops exposure scenarios for any percentile of consumers (e.g. average and 95th percentile high consumers) and for different age groups. However, its protectiveness for vulnerable population groups remains unclear. Notably, FACET provides a list of substances used in FCM, which is significantly more comprehensive that other lists that are currently available in Europe. Whilst the tool is not designed to carry out compliance work, it may be used by manufacturers to estimate likely exposure scenarios for both existing and new substances. It may serve manufacturers in their supportive documentation and risk assessment of FCMs currently not specifically regulated under European law.

In spite of its strengths, the tool also has some limitations and weaknesses of concern. Firstly, it has to be noted that FACET does not include secondary packaging, and thus, it is not adequate to estimate exposure to contaminants migrating out of secondary packaging materials (i.e. the migration of mineral oils from paper and board). Without detailed data on migration from secondary packaging, a probabilistic exposure tool is generally inadequate to model migration. Secondly, the model is dependent on encrypted and confidential National food consumption data. Data quality may vary widely from country to country. Thirdly, while the identity of substances covered by the model can be extracted, data on usage and migration is proprietary and confidential (Oldring et al 2013b). The confidentiality of the data hinders an independent peer-review assessment of the data’s quality. Finally, the lack of access to the raw data on chemical presence and migration levels hinders academic research from benefiting of the tool for hypothesis development, or for confirming exposure measurements for known substances.

*Even though these are known weaknesses, these National surveys compile the best data available to date and European authorities rely on them for risk assessment and risk management.

References

Oldring, P. et al (2013a). “” Food Additives& Contaminants: Part A.

Oldring, P. et al. (2013b). “” Food Additives & Contaminants: Part A.

EFSA (July 25, 2013). “”

Seiler et al. (2013). “” Food Additives and ContaminantsPart A (published online January 23, 2014).