Assessment of rare dietary risks based on analysis of food combinations

Probabilistic methods, in particular Monte Carlo methods, have become widely used in assessment of dietary risks from plant protection products. However, if the critical exposure occurs rarely, estimating its probability with commonly used Monte Carlo approaches can require an unrealistically big number of iterations. A simple method proposed in this paper, referred to as food combination analysis (FCA), finds out subsets of input values necessary for occurrence of a critical exposure event. In particular, for a critical event to occur consumption of a certain combination of contaminated foods could be required. Sometimes by finding the probability that such a food combination is consumed one could directly get an acceptable estimate of the risk, without Monte Carlo simulations. The method performs especially well if available data sets of consumed amounts of foods and residue concentrations of a chemical contain a large fraction of zeros. Based on a literature example, it is shown that the probability of the critical exposure estimated with the FCA could be more than 10 times lower than the estimate of a Monte Carlo approach with 50,000 iterations. The present approach also provides a platform for adaptation and development of more sophisticated methods to estimate low dietary risks.

Research highlights► Effective method for probabilistic risk assessment without Monte Carlo simulations. ► Accounting for large fractions of zeros in consumption and residue data. ► Finding food combinations that could cause a critical exposure event. ► Identification of possible measures to reduce a dietary risk. ► Prerequisite for advanced methods to estimate probabilities of rare dietary risks.