Food composition database reliability in calculations of diet offers

Abstract

This study analysed food offers and compared food composition of daily menus of two most common hospital diets, pancreatic and diabetic diet. In diets that are concentrated on the minimisation of some food components (as minimisation of carbohydrate content in diabetic diet or fat intake minimisation in the pancreatic diet), use of a reliable Food Composition database (FCDB) is an important factor. By use of four FCDBs, Croatian, Danish, USDA and the hospital FCDB, the food composition was analysed. Seven daily offers per two seasons were analysed (seasons: Spring/Summer & Autumn/Winter). Descriptive statistics as well as multivariate tools were used to investigate differences in the food composition, when different FCDB’s were the basis of calculations. Multivariate analysis detected food composition differences in booth diets, when different FCDBs were used (content of energy, proteins, sodium, magnesium, iron & zinc) as well as for offers in different seasons. To be sure that the differences are significant we applied the Bland-Altman test. The results showed different biases and ranges – but none of the food component showed to be significantly different, what would lead to a potential rejection of some of the used FCDB. Those results are a strong confirmation that the Bland-Altman test ensures objective conclusions; like which FCDB is applicable for food composition analysis and menu planning.