Studies on Validity and Reproducibility of Semi-quantitative Food Frequency Questionnaire for a local Community

通过食物频率问卷（FFQ）可以估算出个体食物的长期通常或平均摄入量。FFQ在流行病学研究中很有用，例如调查饮食与慢性病之间的关系。在过去的二十年中，日本已经开发了几个ffq并用于队列研究，但效度并不像欧洲和美国的ffq那么高。2003年，我们在国家长寿科学研究所流行病学部为当地社区开发了一种新的165项半定量FFQ。该FFQ的效度高于日本现有的FFQ，但在短期研究中，效度与欧洲和美国的FFQ几乎相等。通过比较每3个月的5个FFQ和12天的膳食记录（12DR: 3 days×4季节），我们检验了37种营养素的FFQ的长期有效性和可重复性。此外，我们还研究了饮食记录的季节性差异。经能量和去衰减调整后，每个FFQ与12DR之间营养物质的中位数相关系数范围为0.45-0.52（5次的中位数为0.50）。第一次FFQ与每次FFQ的中位相关系数为0.50-0.63（4次的中位相关系数均为0.60）。这是传统ffq中效度和重现性最高的。这项研究也证实了日本食物摄入量的季节性变化。由于该FFQ的长期疗效已被证明，我们可以在未来的大规模队列研究中使用该FFQ作为DR评估营养和食物摄入量的补充工具。

Long-term usual or average intakes of individual foods can be estimated by Food Frequency Questionnaire (FFQ). FFQ is useful in epidemiological studies such as an investigation of the relationship between diet and chronic disease. Several FFQs have been developed during the last two decades and used in cohort studies in Japan, but the validity was not high as FFQs in Europe and the United States. Because of the local characteristics in food intakes, we have developed a new 165 item semi-quantitative FFQ for a local community in the Department of Epidemiology, National Institute for Longevity Sciences in 2003. Validity of this FFQ was higher than those of existing FFQs in Japan and the validity was almost equal to those in Europe and the United States in a short-term study, nonetheless. We examined long-term validity and reproducibility of the FFQ in 37 nutrients by comparing between five FFQs in every three months and 12-day dietary records (12DR: 3 days×4 seasons). Furthermore, we examined seasonal differences of the dietary records. Median correlation coefficients of nutrients between each FFQ and 12DR adjusted for energy and de-attenuated ranged from 0.45-0.52 (median 0.50 of all 5 times). Median correlation coefficient between the 1stFFQ and each FFQ ranged 0.50-0.63 (median 0.60 of all 4 times). This was the highest validity and reproducibility among the conventional FFQs. Seasonal variability in food intakes in Japan was also confirmed in this study. As the long-term efficacy of this FFQ has been proved, we can use this FFQ in future large-scale cohort studies as complementary tool of DR for the assessment of nutrition and food intakes.