Exploring two food composition databases to estimate nutritional components of whole meals

• Tipo: Artigo em Revista Científica Internacional
• Data: 2021

Autores

Silva, M (Autor) Outra; Ribeiro, M (Autor) Outra; Viegas, Olga (Autor) FCNAUP; Martins, ZE (Autor) Outra; Faria, M (Autor) Outra; Susana Casal (Autor) FFUP; Pinto, E (Autor) Outra; Agostinho Almeida (Autor) FFUP; Pinho, Olívia (Autor) FCNAUP; Ferreira, Isabel (Autor) FFUP

Revista

Título: Journal of Food Composition and Analysis

Vol. 102

ISSN: 0889-1575

Editora: Elsevier

Indexação

ISI Web of Knowledge - 0 Citações

Scopus - 2 Citações

Outras Informações

ID Authenticus: P-00V-3TQ

DOI: 10.1016/j.jfca.2021.104070

Abstract (EN): The integration of foodomics data to explain the impact of diet on health requires a precise knowledge of nutrients composition of complex meals. This work assesses the adequacy of two food composition databases (FCDBs) for calculation of nutritional composition of whole meals, compared to the golden standard ¿lab chemical analyses¿ and search for predictive models to overcome some limitations of FCDBs. Six meals were designed by integrating healthy foods in a meal based on the ¿Western diet¿ pattern. The nutritional composition of each meal was i) chemically determined; ii) retrieved from the Portuguese food composition table (TCAP) and from iii) United States Department of Agriculture database (USDA). Compared to chemical analyses, both FCDBs significantly (p < 0.05) overestimate the amount of Na and vitamin B6; TCAP also overestimate the amount of Ca (p < 0.05), while USDA overestimate energy, fat, available carbohydrates, P, and Fe. Linear regression analyses were used to adjust nutrient values based on TCAP and USDA. Predictive models from both FCDBs were successfully obtained for reliable estimation of protein, PUFA, available carbohydrates, total carbohydrates, sugars, Zn, ß-carotene, vitamin E, riboflavin, and niacin in meals with a given uncertainty, which is provided by the respective correction factors. Those predictive models are limited to the range of theoretical values of meals studied. © 2021 Elsevier Inc.

Idioma: Inglês

Tipo (Avaliação Docente): Científica