OBJECTIVE: To test reproducibility and validity of a food frequency questionnaire based in food groups developed for adults living in Southern Brazil. METHODS: Two food frequency questionnaire (food frequency questionnaire 1 and food frequency questionnaire 2) and three 24-hour dietary recalls (24h-R) were applied to 128 participants (20-69 years). The foods were grouped in 19 groups previously defined. Analyses were performed for crude data and adjusted for energy. RESULTS: Statistically significant differences in the estimates of intake between two food frequency questionnaire were observed for the group of leafy vegetables and legumes. The energy-adjusted correlation coefficients between questionnaires ranged from 0.53 (sausages and hams) to 0.85 (alcoholic beverages and low-fat dairy). As to the validity, both questionnaires showed higher and lower estimates of intake in relation to 24-hour dietary recalls. The mean correlation coefficient between each food frequency questionnaire and 24h-R was 0.39 (food frequency questionnaire 1) and 0.40 (food frequency questionnaire 2). The correlation coefficients were equal to or greater than 0.40 for nine food groups in the food frequency questionnaire 1 and for eleven food groups in the food frequency questionnaire 2. The average classification percentages in the same tertile, by the two methods, were 49% (food frequency questionnaire 1) and 48% (food frequency questionnaire 2). The weighted kappa values ranged from -0.04 (non-alcoholic beverages) to 0.69 (milk and light derived) for both food frequency questionnaire. CONCLUSION: This food frequency questionnaire based food group represents a good alternative for assessing the food habits, with the advantage of having a short list of food items.
Adult; Validation studies; Diet surveys; Reproducibility of results
