Development of a food frequency questionnaire to estimate habitual dietary intake in Japanese children

Long-term eating habits affect lifestyle-related diseases including obesity, diabetes, and cardiovascular diseases. In recent years, obesity has become prevalent not only among adults but also in children in Japan [1, 2]. Those who are obese in childhood tend to remain obese as adults [3‐6]. When children are overweight, they are more likely to develop metabolic syndrome later in life [7]. Furthermore, the longer an individual is overweight, the greater their risk of cardiovascular disease [8]. Various factors contribute to obesity, including physical inactivity, an irregular and unbalanced diet, and over-eating [9]. Dietary habits are formed during childhood [10]. To prevent adult obesity, it is desirable that individuals acquire appropriate dietary habits in childhood. Habitual dietary intake among children should be assessed to evaluate childhood dietary problems, enabling the correction of any bad dietary habits.

In general, a food frequency questionnaire (FFQ) is used in large-scale epidemiological studies. FFQs are often used because they are more cost-effective and easier to administer than weighed dietary records (WDR) or 24-hour recall [11]. There have been a number of reports using FFQs to assess dietary intake in adults [12, 13]. A FFQ originally developed for adults is applied to the assessment of dietary intake in children; however, it has been reported that the children's intake is overestimated when assessed using the FFQ for adults [14‐17]. Since it is necessary to develop specific FFQs for the group being studied (to accurately reflect socioeconomic, cultural, and seasonal differences), it would therefore be useful to develop a specific FFQ for children [18]. However, to date no FFQ has been developed and tested in Japanese children [12].

In Japan, some FFQs incorporate raw single food or food groups, such as meat, fish, egg and vegetables [12]. However, a typical Japanese meal normally consists of steamed rice, soup, main dishes, and multiple side dishes. These dishes include various food items and seasonings, and are served as chopped foods that are mixed and/or cooked. This meal format is similar in other Asian countries. Given the mixed nature of these meals, it is difficult for Asian subjects to estimate their intake of single ingredients in the meals prior to cooking, especially if the subjects did not prepare or cook their own meals [19, 20]. In order to enable the subjects to answer a FFQ more accurately, a questionnaire as presented below must be developed. The FFQ includes questions about both individual food items and mixed dishes. Therefore, we developed a FFQ specifically for Japanese children, which included questions about both individual food items and mixed dishes based upon the typical eating habits of normal Japanese children.

A general WDR is quantified either by weighing or determining volumes using a household measuring tool, such as standard measuring cups and spoons, and a ruler for measuring dimensions. Usually, general WDR performers weigh the raw ingredients [23]. Our WDR was designed to make it easier to assess food portions and composition because we used a digital cooking scale as an index of the size of the dish. We confirmed that the Spearman's correlation coefficient between the general WDR and our WDR were 0.88, 0.81, 0.65, 0.88, and 0.50 for energy, protein, fat, carbohydrate, and sodium intake, respectively, in the preliminary investigation. In the Japanese diet, seasonings such as soy sauce are important dietary elements; however, it is difficult to estimate the intake of these foods from dietary photographs alone. Our WDR was able to give a moderate estimate of sodium intake. To obtain the necessary open-ended data from children, we conducted our WDR instead of using a general WDR. The open-ended data generated by our WDR identified the FOOD types that contribute most importantly to the total absolute intake of a nutrient by the group as a whole. An advantage of this open-ended approach is that important contributors to nutrient intake are unlikely to be missed. Many arbitrary decisions, however, must be made regarding the collapsing of variables, as the open-ended methods are typically coded in much finer detail than would be appropriate for items on a questionnaire [11]. Therefore, we combined the multiple regression analysis to identify the FOOD types based on between-person variance with the contribution analysis. Finally, with two methods we obtained a list of 75 food items that are most important within the diets of Japanese children.

Willett reported the limit for the number of food items as 130 [11]. This is because the burden on subjects to respond to questions increases when there are many food items. Date et al. developed a FFQ from the basic dietary data of 805 Japanese adults, and adopted 122 items [19]. Another study by Tokudome et al. developed a FFQ from 351 Japanese adults, and included 102 items [24]. Cade et al. reported the distribution of several food items in which FFQs were used from 227 studies in a review, and reported that the median number of food items was 79 (range, 5-350 items) [13]. The number of food items in our FFQ is 75, which we consider to be the optimal number of items that can provide an accurate picture of dietary habits without becoming a burden to the study participants.

There were, however, two limitations associated with this study. First, our FFQ does not encompass the seasonal variation in the 75 food items, because the time when we conducted WDR for developing the food list was only one season. However, there are different fruits available in different seasons in Japan. To address this, we decided to add seasonal fruits for future investigation. In addition, because the subjects' parents/guardians freely decided the day for the WRD, it was possible that a special day (e.g. a holiday or other celebration where special foods are eaten) may have been selected. Therefore, when we explained the study method to the parents/guardians, we asked them to eat habitual diets on the day they selected for the WRD.

Finally, the coverage rates of energy and all nutrients were more than 90% except for Vitamin C. Therefore, the FFQ developed for children is considered to be a useful modality for estimating the intake of nutrients.

Using both a contribution analysis and a multiple regression analysis, we developed a 75-food item questionnaire from the study involving 586 children. In future, we will verify the reproducibility and validity of this FFQ.