評估台灣飲食頻率問卷效度之初探研究

Pilot study of the validity of food frequency questionnaires in Taiwan

目標：評估本研究室研發兩種飲食頻率問卷之效度（validity），分別為食材為主飲食頻率問卷（food and ingredients-based food frequency questionnaire, FIFFQ）以及餐次結構食譜為主飲食頻率問卷（meal and dish-based food frequency questionnaire, MDFFQ），亦探討熱量校正（energy calibration）前後的差異。方法：150位受試者執行141項之MDFFQ後，5個月後再執行161項之FIFFQ，並以24HDR校正兩種FFQ，並評估FFQ校正前後的相對效度（relative validity）。結果：以平均相關係數而言，FIFFQ與24HDR營養素校正前後為0.03與0.25，六個食物類別校正前後為0.17與0.32。MDFFQ與24HDR營養素校正前後為0.03與0.25。以平均一致性比例而言，FIFFQ和24HDR營養素校正前後為35%與41%，食物類別校正前後為37%與43%。MDFFQ和24HDR營養素校正前後為33%與41%。以FFQ和24HDR的加權kappa結果而言，FIFFQ營養素校正後Kw範圍為0.00-0.39，MDFFQ校正後為0.02-0.43。結論：不論在營養素或食物種類的相關性與一致性，熱量校正後的FFQ普遍優於校正前，且兩者與24HDR之效度結果為可接受的（acceptable）。因此，建議以飲食頻率問卷探討飲食與其他結果變項（如：疾病）關聯時，熱量調整（energy adjustment）或使用定量飲食評估工具（如：飲食回憶或飲食紀錄法）進行熱量校正可能是必須的。

Objectives: The main purpose of this study was to assess the validity of two food frequency questionnaires: the food and ingredients–based food frequency questionnaire (FIFFQ) and the meal and dish–based food frequency questionnaire (MDFFQ). The effect of energy calibration on validity was also examined. Methods: A total of 151 participants completed the 141-item MDFFQ and then the 161-item FIFFQ 5 months later. Validity was assessed by comparing the questionnaires according to 24-hour recall (24HDR) and after energy calibration. Results: The average of the correlation coefficients between the FIFFQ and 24HDR for 28 nutrients was .03, which increased to 0.25 after energy calibration; for six food groups, it was 0.17 or 0.32 after energy calibration. The average of the coefficients of the correlation between the MDFFQ and 24HDR was 0.03, which increased to 0.25 after energy calibration. Before and after energy calibration, the average percent agreement between the FIFFQ and 24HDR was 35% and 41% for 28 nutrients and was 37% and 43% for 6 food groups, respectively. The average percent agreement between the MDFFQ and 24HDR was 33% or 41% after energy calibration. The weighted kappa ranged from -0.01 to 0.40 for the FIFFQ and 24HDR and from 0.02 to 0.43 for the MDFFQ and 24HDR after energy calibration. Conclusions: The FIFFQ and MDFFQ exhibited acceptable correlations and validity after energy calibration. Therefore, the use of these food frequency questionnaires to assess dietary patterns may require energy adjustment or quantitative dietary measurements such as recall or records for calibration purposes to improve their association with outcome variables.