P772 Development and validation of tools to assess food additive intake: the ENIGMA study

Abstract

Background: Processed food additives are widely used to change food consistency, appearance and shelf life. In the Food Agriculture Organization/ WHO International Food Standards CODEX additives are deemed non-toxic or carcinogenic, but their functional impact is unknown. The global pandemics of metabolic and inflammatory bowel diseases have occurred in parallel with widespread additive use. Additives have been causally linked to microbiota changes and mucus layer destruction. A validated measure of food additive intake does not exist. We report the development and validation of 2 food additive measurement tools. Methods: Questionnaire design: Two dietitians working in Australia and Hong Kong created a database of food additives (n = 10) implicated in IBD, the CODEX food-categories they are permitted in, and their maximum suggested permissible concentration (mg/kg). Food categories were condensed into 27 food lists, with examples. Intake in early life (part 1) and recently (part 2) were assessed. Part 1 comprised 39 dichotomous questions on breast-feeding, home and processed food consumption up to age 18. Part 2 assessed frequency of consumption for the 27 food lists in the preceding 12 months. Forward-backward translation into Hong Kong Chinese was undertaken. Pilot testing: 31 individuals assessed understandability. Validation: A new cohort of 26 individuals undertook the tool twice, 2 weeks apart, to assess reproducibility. Cohen's' Kappa-co-efficient was used to assess percent agreement for part 1 questions. Intraclass correlation coefficient (ICC) was used to assess the agreement between the total annual frequencies of the food lists. Results: Pilot testing: Participants reported difficulty recalling food intake, estimating portions and confusion around certain terms. Instructions were therefore added for estimating food intake. Validation cohort: Respondents judged the questionnaires easy to understand and complete. The average kappa-coefficient for part 1 questions was 0.5. Eighteen per cent of questions had slight to fair correlations, 36% had moderate correlations, and 46% had substantial to almost perfect correlations. Researchers expect moderate correlations in measures of remote diet intake. For part 2 the ICC for total, annual frequency of the 27 food lists was 0.888 (p < 0.001), indicating good reliability. Conclusions: Two tools (part 1 and 2) have been developed and validated, in two major languages and cultures, which reproducibly assess early-life and recent intake of food additives. These can be applied to individuals to assess this important emerging field of the relationship between food additive intake and disease.