Project

Bayesian modelling of dietary intake of pesticides using monitoring and survey data

In order to prevent the occurrence of unwanted substances in our food, these substances are monitored at various places in the food chain (e.g. sea ships, auctions, shops).

A challenging question is how to combine monitoring data with food consumption data in order to estimate the exposure risks posed to consumers. Traditionally the analysis of such risks by food safety authorities has been in terms of deterministic estimates of exposure, but currently there is a shift in thinking from deterministic to probabilistic approaches. However, currently available probabilistic approaches require data of sufficient quality and quantity, which are often not available. Bayesian models hold a promise to allow an efficient use of prior knowledge and an efficient pooling of parameters. In this study we explored the possibilities of Bayesian models for exposure assessment in order to prepare more realistic risk analysis methods in the field of food safety. In this analysis the risks associated with the possible presence of pesticides in vegetables and fruits was taken as example. Bayesian models for food consumption, weight and for concentrations of Iprodione were developed from survey and monitoring data, and combined to obtain posterior intakes of Iprodione.
Fig 1 – Observed consumption of four commodities by 6000 Dutch on one given day.
Fig 1 – Observed consumption of four commodities by 6000 Dutch on one given day.
Fig 2 – Amount of the pesticide Iprodione found in samples of four vegetables (histogram, blue bars, and bayesian model, red line).
Fig 2 – Amount of the pesticide Iprodione found in samples of four vegetables (histogram, blue bars, and bayesian model, red line).
Fig 3 - Posterior intakes of Iprodione by the Dutch population in one given day, based on the consumption of four commodities containing the pesticide. Only 4.3% of the population will have a positive intake, and x% have intakes above the Acute Reference Dose.
Fig 3 - Posterior intakes of Iprodione by the Dutch population in one given day, based on the consumption of four commodities containing the pesticide. Only 4.3% of the population will have a positive intake, and x% have intakes above the Acute Reference Dose.