Following the core food (CF) models of the USA and France reported in the 2000s, this study presents the first oriental CF model, for total diet studies in Taiwan. First, we combined the four latest national dietary recall surveys from 2005 to 2012, data sets from the Nutrition and Health Survey in Taiwan (NAHSIT) covering all ages (1 to 65+ years old). In total, 3,098 NAHSIT food codes from 349,497 records of responses from 7,580 individuals were aggregated within a three-tiered scheme into 237 CFs in 54 subcategories and 16 categories. As with Western foods, Taiwanese foods were mainly aggregated by food type and major ingredient, processing method, cooking method or serving habit. In addition, we propose several categories typical to Taiwanese foods: (1) processed foods (e.g. rice paste, rice noodle, bean curd); (2) mixed foods (e.g. tomato fried with egg, Guangdong porridge) and stuffed staple foods (e.g. rice cake, zongzi); (3) Chinese Dim Sum (e.g. dumpling, salted bun); and (4) animal offal (e.g. pig liver, chicken kidney) and by-products (e.g. chicken head, pig foot). Systematic nomenclature and coding methods are also proposed. A computerised web model was developed using a graphical user interface (GUI) for automatic CF aggregation and inquiry of various consumption estimates according to the residing area, age group, and gender of consumers. Point estimates of weighted means were calculated for consumption rates (CR) based on the whole group (WG) or consumers only (CO). The CR and body weight (BW) of each individual were matched to calculate mean intake per unit body weight (g/kg bw/day). A general sample design procedure with the CF model is suggested for a total diet study (TDS) on target hazards. Strategies are also discussed for improving the quality of food matching in sample designs.
Keywords: Core foods; food aggregation; food matching; sampling design; total diet study.