Reproducibility and validity of a Chinese food frequency questionnaire used in Taiwan.

A 64-item Chinese food frequency questionnaire (FFQ) combined with open questions on types of staples and cooking oil most frequently consumed was designed for a prospective study in Taiwan to appraise participants' usual intake. We examined its reproducibility and validity. The form was administered three times at three-month intervals by face-to-face interview to 83 senior college students majoring in nutrition, in order to recall their usual dietary intake over the past six months. They also completed five-day (includes both weekends) diet records (DR) after each interview. Averaged intake levels of most nutrients assessed by the three FFQs were slightly higher than those of 15-day DR. Intraclass correlation coefficients for nutrient intakes assessed by three FFQs three months apart ranged from 0.37 for saturated fat to 0.82 for alcohol (average: 0.52). Averaged Pearson correlation coefficients between the unadjusted and energy-adjusted nutrient intakes measured by DR and by the third FFQ (which asked about diet during the six months encompassing the diet records) were 0.40 and 0.35, respectively, not including vitamins A and C. These correlations were higher (average: 0.47) after adjusting for the daily variation of the diet records. On average, 50% subjects were correctly classified into the same tertiles by both methods; 11% of the subjects were misclassified to extreme categories. These data indicate that this FFQ for Chinese-speaking people in Taiwan is reproducible and provides a useful measure of intake for many nutrients over a six-month period.