Validity and Reproducibility of a Food Frequency Questionnaire for Dietary Assessment in Japanese People.

Abstract

A 76-item food frequency questionnaire (FFQ) was developed to investigate nutritional epidemiology in urban residents in Japan. The authors prepared two food models-a life-size three-dimensional model and a life-size two-dimensional photograph-to assess the FFQ portion size. The validity of the FFQ was verified using the two food models by comparing them with 16-d weighted dietary records (WDRs). Validation was conducted by comparing the FFQ1 findings with those obtained with the WDR, which is regarded as the gold standard, and reproducibility was verified by comparing the findings from FFQ2 and FFQ1. After completion of the WDR, the participants were randomized into two groups. In one group, the FFQ was conducted using life-size three-dimensional models (3D-FFQ) to estimate the portion size. In the other group, the FFQ was administered using life-size photo collection (2D-FFQ). Regarding validity, the median values (range) of Pearson's correlation coefficients for the energy and nutrient intake of the 32 items by the WDR and FFQ1 were r=0.53 (0.30-0.68) in the 3D-FFQ and r=0.57 (0.33-0.87) in the 2D-FFQ. When FFQs with 2D or 3D food models and two different portion sizes were compared with regard to the intake of certain food groups, energy, and nutrients, both the 2D-FFQ and 3D-FFQ provided good correlation coefficients with the WDR.