台灣成人食材為主飲食頻率問卷之發展研究

Abstract

飲食頻率問卷(food frequency questionnaire, FFQ)收集受訪者攝取各項食物的頻率，用於評估長期的一般飲食狀況，為流行病學研究上重要的飲食評估法。本研究目的為以台灣人飲食為基礎發展出「台灣成人飲食頻率問卷(Taiwanese Adult Food Frequency Questionnaire, TAFFQ)」，以評估台灣成年人的飲食攝取狀況。

本研究使用三世代長期追蹤的660份二十四小時飲食回憶資料(24-hour dietary recall, 24HDR)，計算15種營養素(熱量、蛋白質、脂質、醣類、維生素A、胡蘿蔔素、維生素B1、維生素B2、菸鹼酸、維生素C、鈉、鉀、鈣、磷及鐵)累積貢獻百分比至90%的食物，作為TAFFQ的食物項目。由24HDR計算出的平均數及中位數訂定食物項目的中份，以天然單位或家用量器標示在食物項目旁，設計成半定量飲食頻率問卷(semi-quantitative food frequency questionnaire, SFFQ)。最後將設計完成的TAFFQ與本研究室先前研發之餐次結構(meal structure; meal-based)的食譜為主(dish-based)「個人飲食頻率問卷」一同進行預試(n = 43)，並與定量飲食資料(24HDR及飲食記錄法)比較相關性及一致性。

研究結果研發出109項食材為主(food-based)的自填式SFFQ，使用九種頻率(最低為從未或很少吃、最高為每日5次以上)，回憶期間為過去一年。TAFFQ的食物清單除了脂質(65%)及鈉(45%)的涵蓋率較低，其它的營養素涵蓋率佳(平均數= 89%、中位數= 95%、範圍= 45~99%)。問卷包含飲品甜度、烹調方式以及奶類的脂肪含量等補充性問題。最後放入開放式問題，詢問膳食補充劑的使用狀況、食物項目未列出的常吃食物以及意見回饋。

TAFFQ預試結果平均每日熱量攝取為1828大卡、蛋白質66公克(佔總熱量15%)、脂質59公克(29%)、醣類254公克(56%)。個人飲食頻率問卷平均每日熱量攝取為2756大卡、蛋白質100公克(14%)、脂質102公克(33%)、醣類353公克(51%)。斯皮爾曼等級相關性方面，兩種FFQ與定量飲食資料皆在維生素C、鉀、鈣、鎂、磷及膳食纖維達顯著相關(p＜0.05)。另外TAFFQ在醣類、維生素B1、維生素B2及膽固醇的相關性較佳，個人飲食頻率問卷則以熱量及維生素A較佳。使用布蘭德-奧特曼差異圖(Bland-Altman plot)比較兩種FFQ與定量飲食資料的一致性，發現熱量及三大營養素皆有超過93%落在一致性界線內。本研究設計之TAFFQ未來將進行信效度研究，可望應用於長期飲食追蹤的流行病學研究。

Food frequency questionnaires (FFQs) are one of the most important dietary assessment methods in epidemiological studies, collecting subjects’ food frequency information which is able to assess long-term dietary intake. The purpose of this study was to develop the Taiwanese Adult Food Frequency Questionnaire (TAFFQ) using data-driven approach to measure dietary intake of Taiwanese adults.

The food items of TAFFQ were selected from 660 days of 24-hour dietary recalls (24HDRs) of three longitudinal cohorts to identify foods that cumulatively contributed 90% to each following 15 nutrients: energy, protein, fat, carbohydrate, vitamin A, β carotene, thiamin, riboflavin, niacin, vitamin C, sodium, potassium, calcium, phosphorus and iron. TAFFQ was designed as a semi-quantitative food frequency questionnaire (SFFQ). The mean and median intakes were calculated from 660 days of 24HDRs to determine the standard portion sizes of food items. The standard portion sizes were specified in natural units or household measures on TAFFQ. The accomplished TAFFQ alone with the dish-based FFQ structured by meals developed previously were pilot tested (n = 43), and compared correlation coefficients and agreements with quantitative dietary data (24HDRs and diet records).

The self-administered 109-item food-based SFFQ was developed with nine frequency response section (from never or almost never to 5 or more times per day) to assess frequencies of food consumption over the past year. The food list of TAFFQ covered the population intake well (mean = 89%, median = 95%, range = 45~99%) except for fat (65%) and sodium (45%). The questionnaire included supplementary questions like added sugar in drinks, cooking methods and fat components in milk. The last parts of TAFFQ are open-ended questions inquiring about the use of dietary supplements, other foods frequently eaten and feedback.

The average daily intake of energy was 1828kcal, protein was 66g (15%E), fat was 59g (29%E) and carbohydrate was 254g (56%E) of TAFFQ. The evaluation of the dish-based FFQ on the average daily intake of energy was 2756kcal, protein was 100g (14%E), fat was 102g (33%E) and carbohydrate was 353g (51%E). Significant correlations were observed in vitamin C, potassium, calcium, magnesium, phosphorus and dietary fiber between 2FFQs and quantitative dietary data in Spearman rank correlation coefficients (p＜0.05). The correlations of carbohydrate, thiamin, riboflavin, and cholesterol performed better in TAFFQ, but the dish-based FFQ had better correlations in energy and vitamin A. The Bland-Altman plots showed acceptable agreements in energy, protein, fat and carbohydrate between 2FFQs and quantitative dietary data (＞93% within the limits of agreement). The reproducibility and validity study of TAFFQ are currently underway and would be applicable to longitudinal nutrition epidemiological research.