Validation and Comparison of Nutrient Intakes from Two Dietary Questionnaires Developed for the Multiethnic Cohort Study.

Background: Quantitative food frequency questionnaires (QFFQs) are often used to measure dietary intakes in large cohort studies but the impact of updating these questionnaires over time is not often examined.

Objective: This study compared nutrient intakes estimated from two different QFFQs to each other and to intakes calculated from three 24-hour dietary recalls (24HDRs).

Design: This study used a cross-sectional design.

Participants/setting Participants: Participants (N = 352) were members of the Multiethnic Cohort Study from five racial and ethnic groups (African American, Japanese American, Latino American, Native Hawaiian, and White) who lived in Hawaii and Los Angeles. They were recruited in 2010 and asked to complete two QFFQs, two months apart, and three 24HDRs in the time between completion of the QFFQs. One questionnaire had been developed for a baseline survey (baseline QFFQ) at the start of the Multiethnic Cohort Study during 1993-1996, and the other was updated for a follow-up study 10 years later (10-year QFFQ).

Main Outcome Measures: Daily intakes of energy and nine nutrients were estimated from both QFFQs, and from the average of three 24HDRs.

Statistical Analyses Performed: Pearson's correlation coefficients were calculated between log-transformed nutrient intakes from each QFFQ and the 24HDRs and between the two QFFQs overall, by sex, and by race and ethnicity.

Results: Correlations for the 10-year QFFQ with the 24HDRs (average = 0.45) were higher than for the baseline QFFQ (average = 0.41), although the differences were not statistically significant. The increase in correlations was particularly pronounced for Native Hawaiian and African American participants. When absolute values were adjusted for energy intake, the average correlations were higher at 0.57 for the baseline QFFQ and 0.58 for the 10-year QFFQ overall and this pattern was seen in most racial and ethnic subgroups. The average correlations between the two QFFQs were 0.73 for both absolute intakes and nutrient densities overall.

Conclusions: Correlations of nutrient intakes between the two QFFQs and 24HDRs were similar, and intakes from the two QFFQs were highly correlated. QFFQs updated for changes to the food supply may provide improved assessment for cohort studies that include diverse populations.