ABSTRACT
The objective of this study was to validate a food frequency questionnaire (FFQ) used to estimate energy and selected nutrient intake in a Brazilian population with various ethnic backgrounds. Validity of intake estimated using the FFQ was tested among 55 Brazilian women, namely 26 Caucasians, 15 of Japanese descent, and 14 others. The FFQ was originally developed for use in a case-control study of breast cancer conducted in Sao Paulo. Dietary records (DRs) recorded in two seasons were used as references. Intake of energy and 24 nutrients were calculated using the USDA and Japanese food composition tables. Validity and reproducibility were evaluated using Spearman's correlation coefficients. Results showed that intake of chicken/poultry, eggs and legumes were overestimated by the FFQ compared to the DR, whereas that of pork and fat was underestimated. Further, intake of folate, fiber and isoflavones was overestimated by the FFQ whereas that of energy, fat, carbohydrate alpha-carotene and lycopene was underestimated. Energy-adjusted correlation coefficients between nutrient intakes estimated with the FFQ and DR were high for isoflavones (0.76), calcium (0.50), and vitamin C (0.49). In contrast, validity varied from moderately high to low for energy and other nutrients. In conclusion, validity of the FFQ for estimation of the intake of selected nutrients among Brazilian women with varied ethnic background was moderately high.
