Monte Carlo Modeling for Aflatoxin B1 Distribution in Pistachio Samples: A Prerequisite for Sampling Plan Validation

Abstract

AFB1 is the main aflatoxin found in pistachios, and as it is dangerous for health, the European Union (EU) has set a maximum level at 8 µg/kg in pistachios in 10 kg samples. Given that efficient sampling plans must be used in order to determine this contaminant level accurately, a mathematical validation of sampling plans has to be achieved with to Operating Characteristics curves. However, a prerequisite before drawing these curves is to assess the contaminant distribution within the food lot as well as to determine the variability between samples taken from the food lot. This is difficult to achieve for AFB1 in pistachios, as the contamination incidence rate is low, and concentrations differ greatly between food items. More precisely, the contaminant is heterogeneously distributed. Contrarily to current contaminant distribution assessment methods used on a regulatory basis and which are based directly on variability between samples, the method developed relied first on variability between individual pistachios, before computing the variability between samples, thus triggering more reliable results. This work assessed the distribution and incidence of aflatoxin in pistachio nuts, thanks to a Monte Carlo simulation, with pistachios split into categories, which were early split sound kernel, cracked sound kernel, and mouldy pistachios. A 30 kg sample size of Iranian pistachios for export to Europe was considered. Differences between good and bad lots were taken into account with lots split into 3 classes for mean contamination levels in export packages. Simulation results were in good accordance with observed contamination levels in Iran.