Validity of a food frequency questionnaire in a population with high alcohol consumption in Japan

Alcohol consumption has a relatively large impact on energy intake in drinkers, and several studies reported different dietary habits from non-drinkers. However, few studies have investigated the influence of alcohol consumption on the validity of the Food Frequency Questionnaire (FFQ). To investigate its influence, we conducted a validity test in a population with high alcohol consumption.

The study subjects were 66 residents living on an island in the south-western part of Japan. We conducted the FFQ and 12-day-weighed dietary records (12d-WDRs) in each 3 day of each 4 season. We calculated Pearson correlation coefficients (CCs) and agreement rates according to quartile classification after adjusting for energy.

The intake energy (kcal) estimated from 12d-WDRs and FFQ was 1,641 and 1,534 in women, and 2,093 and 1,979 in men, respectively. The cumulative percentage contribution of the alcohol energy was 6.7% in men. De-attenuated, log-transformed Pearson's median CCs between the nutrients quantified with the 12d-WDRs and FFQ were 0.51 in women and 0.38 in men. The CCs for carbohydrate and saturated fatty acids intake of men were lower than those in the previous Tokai study using the same FFQ. The findings in agreement rates were consistent with the Tokai study.

This study suggested that the FFQ can be used for epidemiological studies using categorical comparisons in this population, although the underestimation of carbohydrates and other nutrients in the FFQ should be taken into consideration.