Sampling design by the core-food approach for the Taiwan total diet study on veterinary drugs

The core-food (CF) approach, first adopted in the United States in the 1980s, has been widely used by many countries to assess the exposure to dietary hazards at a population level. However, the reliability of exposure estimates (C × CR) depends critically on sampling methods designed for the detected chemical concentrations (C) of each CF to match with the corresponding consumption rate (CR) estimated from the surveyed intake data. In order to reduce the uncertainty of food matching, this study presents a sampling design scheme, namely the subsample method, for the 2016 Taiwan total diet study (TDS) on veterinary drugs. We first combined the four sets of national dietary recall data that covered the entire age strata (1-65+ years), and aggregated them into 307 CFs by their similarity in nutritional values, manufacturing and cooking methods. The 40 CFs pertinent to veterinary drug residues were selected for this study, and 16 subsamples for each CF were designed by weighing their quantities in CR, product brands, manufacturing, processing and cooking methods. The calculated food matching rates of each CF from this study were 84.3-97.3%, which were higher than those obtained from many previous studies using the representative food (RF) method (53.1-57.8%). The subsample method not only considers the variety of food processing and cooking methods, but also it provides better food matching and reduces the uncertainty of exposure assessment.