Validation of a semi-quantitative Food Frequency Questionnaire for dietary assessment among adults in Northwest China.

Abstract

Accurately assessing habitual diet is essential for nutritional epidemiology. While food frequency questionnaires (FFQs) are a practical tool for large-scale studies, they require population-specific validation due to regional dietary variations. This study aimed to validate a culturally tailored semi-quantitative FFQ for an understudied population in Northwest China, utilizing a digital Chronic Disease Follow-up Platform (CDFP) to enhance data collection and participant engagement. A culturally specific 62-item FFQ was developed based on local dietary surveys. The FFQ was administered twice (FFQ1, FFQ2) 3 months apart to assess reproducibility. Relative validity was evaluated against the average of three non-consecutive 24-h food records (3d-FR). The CDFP was employed to manage the entire validation study workflow, including recruiting, sending automated short-message reminders for uploading photographs of food records, and scheduling follow-up telephone calls for the completion of two FFQ administrations (FFQ1, FFQ2). Nutrient intakes were calculated using the China Food Composition Database. Statistical analyses included correlations, intraclass correlation coefficients (ICCs), cross-classification into quartiles, weighted kappa statistics, and Bland-Altman plots. The FFQ demonstrated good to excellent reproducibility, with a mean ICC of 0.71 between FFQ1 and FFQ2. Energy-adjusted correlation coefficients between the FFQ1 and FFQ2 ranged from 0.46 to 0.82 (mean: 0.70). For validity, cross-classification showed 77.1% of participants were assigned to the same or adjacent quartile on average, with gross misclassification into opposite quartiles averaging 5.2%. Bland-Altman analysis indicated acceptable agreement, though a systematic overestimation by the FFQ was observed for energy and most nutrients. However, we also observed modest to fair validity correlations (mean crude r = 0.31 and energy-adjusted r = 0.22), with several nutrients showing non-significant associations in crude analyses. The developed FFQ shows reasonable reproducibility and acceptable relative validity for ranking nutrient intakes rather than for estimating absolute intakes among adults in northwest China. The integration of the CDFP proved to be an effective strategy for avoiding loss to follow-ups and ensuring high-quality dietary data collection. Further validation studies in community-based settings with larger and more balanced samples are required.