Validity and reproducibility of a semi-quantitative food frequency questionnaire for Korean adults with obesity

This study examined the validity and reproducibility of a newly developed 129-item semi-quantitative food frequency questionnaire (FFQ) for Korean adults with obesity.

Ninety-seven Korean adults aged 19-64 years with obesity participated. To assess validity, FFQ estimates were compared with 4 seasonal sets of 3-day dietary records (3DRs). Reproducibility was assessed by administering the FFQ twice, as FFQ1 and FFQ2. Agreement was evaluated using quartile cross-classification, weighted kappa statistics, and crude and energy-adjusted Spearman correlation coefficients. Bland-Altman plots evaluated individual-level agreement. Nutrient and food-group intakes were compared between FFQ1 and FFQ2.

The FFQ demonstrated modest relative validity for ranking individuals. Compared with mean 3DR values, same-or-adjacent quartile agreement for energy and 10 nutrients ranged from 68.0% to 82.5%, and crude correlation coefficients ranged from 0.30 to 0.56. Averaging FFQ1 and FFQ2 improved validity metrics, including the energy-adjusted correlation for sugar intake (r=0.54). Bland-Altman plots showed relatively small mean differences for most nutrients, although the limits of agreement were wide and flavonoid intake was generally overestimated. Reproducibility was good, with same-or-adjacent quartile agreement of 81.4% to 90.7% between FFQ1 and FFQ2 and nutrient correlation coefficients of 0.50 to 0.69. Similar consistency was observed for food and dish categories.

The tailored 129-item FFQ for Korean adults with obesity showed modest validity for ranking individuals and good reproducibility. Although caution is needed when estimating precise absolute intake at the individual level, this FFQ may support epidemiological studies of diet, obesity, and related chronic diseases in Korean adults.