Food allergy population thresholds: an evaluation of the number of oral food challenges and dosing schemes on the accuracy of threshold dose distribution modeling.

For most allergenic foods, limited availability of threshold dose information within the population restricts the advice on action levels of unintended allergenic foods which should trigger advisory labeling on packaged foods. The objective of this paper is to provide guidance for selecting an optimal sample size for threshold dosing studies for major allergenic foods and to identify factors influencing the accuracy of estimation. A simulation study was performed to evaluate the effects of sample size and dosing schemes on the accuracy of the threshold distribution curve. The relationships between sample size, dosing scheme and the employed statistical distribution on the one hand and accuracy of estimation on the other hand were obtained. It showed that the largest relative gains in accuracy are obtained when sample size increases from N=20 to N=60. Moreover, it showed that the EuroPrevall dosing scheme is a useful start, but that it may need revision for a specific allergen as more data become available, because a proper allocation of the dosing steps is important. The results may guide risk assessors in minimum sample sizes for new studies and in the allocation of proper dosing schemes for allergens in provocation studies.