RESUMO
Testing the specific migration limits of all substances intentionally added to polymer material according to European Union (EU) regulation is a time-consuming and expensive task. Although mathematical modeling offers an interesting alternative, it can significantly overestimate the migration in situations which are strongly conservative due to significant uncertainty in transport properties. In addition, its application is of little use for end-users or enforcement laboratories, which do not have access to the formulation. This paper revises the paradigm of migration modeling by combining modeling with deformulation experiments and iterative modeling in the framework of decision theory. The complete approach is illustrated for polyolefins in contact with 50% ethanol for eight typical migrants, including hindered phenolic antioxidants and low molecular weight surrogates. Results from a French ACTIA project on the identification of formulation fingerprints and on the prediction of partition coefficients with alcoholic and aqueous stimulants is described. When the true migration was close but still lower than the limit of concern, the proposed compact decision tree, including up to four sources of uncertainty, showed that the chance of demonstrating compliance was about 3 : 4 in the presence of one source of uncertainty, whereas it fell below 2 : 4 and 1 : 4 with two and three sources of uncertainty, respectively. The recommendations for further food packaging safety surveys and future developments are discussed.