Comparison of different exposure assessment methods to estimate the long-term dietary exposure to dioxins and...

Long-term exposures to dioxins (PCCD/F and dioxin-like PCBs) and ochratoxin A were calculated using food consumption data of the European concise database combined with concentration data of the Netherlands (NL) using a deterministic approach. To refine these assessments, exposures were also calculated using three long-term exposure models, observed individual means (OIM), Iowa State University Foods (ISUF), and betabinomial-normal (BBN) models, combined with individual food consumption data of NL. BBN and ISUF correct the variation in long-term exposure for the within-person variation, whereas OIM calculates the mean exposure over the days in the food consumption survey. Exposures obtained with the concise database were highest, and those obtained with OIM higher than with BBN and ISUF. Contribution of the major sources of exposure differed between the concise database and the three models. Given the constraints of the concise database, exposures obtained with this database should be interpreted as a first tier assessment. Preferably, refined assessments using models that correct the variation in long-term exposure for the within-person variation combined with individual food consumption data and national concentration data should be used to assess the long-term exposure. We recommend the use of BBN since it can model exposure distributions that depend on covariates.