Peanut Traces in Food: A Probabilistic Risk Assessment Based on the French MIRABEL Survey

The risk of reactions due to the unintentional presence of allergens, such as traces in packaged products, remains difficult to characterize. The aim was to assess the risk regarding unintended traces of peanut in packaged food products in peanut allergic patients using original data from the MIRABEL survey. We developed an integrated Bayesian probabilistic risk model based on relevant data including consumption of a panel of selected products with and without precautionary labelling by peanut allergic patients, and their individual threshold dose at oral food challenge (OFC). 785 patients (<16 years: 86%) were included in the survey. Data on OFC and food consumption were available for 238 and 443 patients, respectively. For eight food categories with precautionary labelling (30%) or without (70%), the risk was nil (no peanut traces). For chocolate tablets and spreads, the risk was not significantly different from zero. For appetizers, from the different models and including uncertainty intervals, the mean estimated risk was between 38 reactions for 1 000 000 eating occasions and 55 reactions for 10 000 eating occasions. For the 1% lowest dose reactors at OFC, the estimated risk was between 8 reactions for 10 000 and 71 reactions for 1000 eating occasions. According to these results, the allergic risk related to peanut traces in packaged food products was only significant for the most sensitive allergic consumers of appetizers. If the link between food consumption and threshold dose is not taken into account, individual variability could be overlooked, and the risk underestimated. These findings need to be confirmed by larger and representative studies including non-packaged products.