Modelling the chronic non-cancer effects of mixtures of migrants using Cramer classes and quantitative models of uncertainty.

This paper presents a quantitative tool for assessing the toxicity of mixtures of substances that enter food from packaging materials (migrants). These estimates of mixture toxicity are believed to be conservative (the actual toxicity of the mixture will be less than the predicted estimates). These conservative estimates can be used to screen out mixtures of migrants having a low likelihood of causing adverse non-carcinogenic effects. The approach used in the tool is based on the concept of Cramer classes and existing models of mixture toxicity. The tool takes advantage of recent advances in modelling the uncertainty in estimates of safe doses of individual chemicals and mixtures of chemicals and provides both best estimates of safe doses of mixtures and confidence limits of those estimates. Lower confidence limits of the estimates of safe doses can be used to evaluate the adequacy of deterministic estimates of safe doses for the mixtures. Finally, the tool identifies components of a mixture that drive the mixture's risk. This information can be used to direct the development of more refined assessments of the mixtures' toxicity or strategies for risk mitigation. The tool can be used to assess mixtures of any number of migrants and including mixtures with migrants having little or no toxicity data. At the same time, the approach allows the use of toxicity information, when available, to improve the assessment of mixtures' toxicity. Two example applications of the tool are provided.