Validity of single vs. multiple FFQs in ranking nutrient intake among Japanese adults: the Tohoku Medical Megabank Project.

Abstract

Background and objectives: Repeating food frequency questionnaires (FFQs) within the same population was reported to improve the validity of correlation coefficient (CC). However, the enhancement of validity in ranking agreement remains underreported. Herein, we assessed the validity of energy and nutrient intake es-timates using single and multiple FFQs and their ability to rank individuals.

Methods and study design: 213 men and women aged ≥20 years were recruited from the residents participating in the Tohoku Medical Mega-bank Project (TMM) cohort studies; three FFQs were conducted in November each year from 2019 to 2021, with 12-day weighted food records (WFRs) as the reference method. Spearman's rank CCs were calculated between single or multiple FFQs estimates and those obtained through the 12-day WFR. Additionally, the ranking agreement was compared based on cross-classification.

Results: CCs between intake estimated using a single FFQ and 12-day WFR were moderate for several nutrients, with median CCs of 0.52 for men and 0.48 for women. CCs for multiple FFQs were slightly higher than that of single FFQ, with median CCs of 0.59 for men and 0.56 for women. Regardless of the number of FFQs, the proportion of subjects classified into the opposite extreme category was ≤5% for most nutrients.

Conclusions: A single FFQ used for adults in the TMM cohort studies showed moderate validity. Estimates from multiple FFQs improved the accuracy slightly; nevertheless, this indicates that relying on a single FFQ is unlikely to result in a serious misclassifica-tion compared to using intake data from multiple FFQs over a relatively short period.