Standardization of Food Composition Databases for the European Prospective Investigation into Cancer and Nutrition (EPIC): General Theoretical Concept

EPIC is a prospective cohort study on diet and cancer involving 480 000 subjects from nine European countries. In order to establish the relationship between nutrient intakes and disease, standardized food composition databases are needed. In the absence of already existing comparable European nutrient database(s), an ad hoc approach was developed to standardize the EPIC databases. New matrices were built using information as collected from the EPIC study subjects in order to overcome the difficulty of reducing the systematic differences between food lists in national databases. In the EPIC databases, the foods lists, vertical axes, are based on information derived from standardized computerized 24-h diet recalls collected from 35 000 subjects. The criteria of selection and level of detail on foods reported in the EPIC databases are therefore highly standardized between countries. In addition, reported recipes are systematically broken down into ingredients to optimize the comparison between countries. For the nutrient list, horizontal axes, the number, mode of expression, definition, unit and methods of analysis are fixed. The compilation of nutrients, nutrient value cells, is carried out using standardized sources of nutrient data or algorithms. Depending on whether the food is common or country-specific, the same or country-specific source of values will be used. This approach addresses some methodological issues, which may have implications for future priorities and development of food composition tables.