An analysis of the incremental value of retaining brand-level information in food consumption databases in estimating food additive intake

In the Dietary and Nutritional Survey of British Adults (DNSBA) food consumption database, brand level intake data are recorded for 2197 subjects over 7 days. This study set out to examine the incremental value of such brand level data in food consumption studies to estimate additive intake. The food codes in the DNSBA database were re-arranged into 14 food categories within each of which were varying numbers of both sub-categories of foods and of brands. Intake of the 97.5th percentile for brands and sub-categories were compared with intakes at the 97.5th percentile of the appropriate food category. Taking ≥ 60 consumers as the minimal sample size for which a 97.5th percentile statistic can be calculated, only 6% of the 1363 brands listed had sufficient data to work with. Of the 14 food categories, six had a sub-category with a 97.5th percentile in excess of that of the category but none exceeded 1.3 × 97.5th percentile intake of the category. Of the 85 brands for which there were ≥ 60 consumers, only 11 had 97.5th percentile intakes in excess of that statistic for the relevant food category and none exceeded a multiple of 1.3 times the category 97.5th percentile intakes. Unless dietary surveys are very large and carried out for longer periods of time, there is little value in retaining food consumption data at brand level. If 1.3× the intake of an additive from a given food category at the 97.5th percentile does not exceed the ADI, no sub-category or brand appears to do so. This may provide a valuable technique in crude estimates of food additive intake.