INTRODUCTION
Adequate dietary intake is important for normal growth and development, maintaining health and well-being, reducing the risk of illness and injury, and optimizing sports performance in junior athletes [1]. Diet is routinely assessed by nutrition professionals to evaluate whether an individual is achieving specific health and sports nutrition targets [2]. However, the accurate nutritional assessment of athletes in practical situations is difficult [3].
In epidemiological studies, food frequency questionnaires (FFQs) have been used as effective measures for assessing habitual dietary patterns. As dietary habits vary greatly by ethnicity as well as social and cultural backgrounds, FFQs must be tailored to target populations [4-6]. The Japanese diet has unique characteristics [7], and, to date, FFQs have been developed and validated for different populations in Japan [4]. Additionally, factors affecting food choice differ between athletes and the general population. Athletes engage in carbohydrate loading before competitions or consume high-protein diets to build muscle. Many athletes choose specific foods to improve their performance or physique [8], and foods typically listed in the FFQs may differ from the usual food choices of athletes. Although the FFQs developed using databases involving nonathletes were validated for use by athletes [9-11], they may not be useful in assessing the diet of Japanese athletes.
Ishikawa-Takata et al. [12], in contrast, used a database of athletes’ food intake to develop a new FFQ for Japanese athletes (FFQJA) and found moderate validity for all nutrients; the quartile indicated good joint classification. This previous validation study targeted 78 athletes (age: 15-22 years) [12]. However, it is unclear whether this dietary assessment method is useful for athletes younger than 15 years. According to previous studies, the age at which children can accurately report their own food intake is approximately 12 years [13]. The cognitive abilities required to self-report food intake include an adequately developed concept of time, a good memory and attention span, and knowledge of the names of food [14]. It is unclear whether athletes aged <15 years (i.e., junior high school athletes) can answer the FFQJA accurately.
Junior high school athletes have very high nutritional requirements owing to intensive daily training, in addition to their energy and nutrient needs for growth and development [15]. Junior high school athletes, particularly endurance athletes, sometimes suffer from medical complications related to undernutrition or diet [15]. Dietary interventions are reportedly useful for maintaining overall health and athletic performance in junior high school athletes [16,17]. Therefore, to perform proper dietary interventions or provide appropriate dietary support to Japanese junior high school athletes, it is important to estimate their habitual nutrient and food intake.
To the best of our knowledge, no studies have verified the usefulness of FFQs developed from athlete food databases for junior high school athletes. Therefore, we investigated whether the FFQJA, originally developed and validated to estimate the dietary intake of Japanese athletes, is also valid for estimating and ranking the intake of junior high school athletes. We hypothesized that the FFQJA would demonstrate acceptable validity for estimating nutrient intake among junior high school athletes.
METHODS
Participants
We recruited junior high school athletes (13-15 years old) through announcements posted on bulletin boards at soccer clubs in central Saitama Prefecture, Japan. In this study, the inclusion criteria were participation in regular training sessions to improve performance and regular participation in competitions. In the present study, “regular training sessions” were defined as participation in sports training for 3-4 days per week for> 1 year. A total of 113 junior high school athletes satisfied the inclusion criteria. The data of 49 were excluded from the analyses (44 did not complete all 3 days of dietary records [DRs], and five did not submit their FFQJA). The remaining 64 (56.6%) were ultimately included in the analysis. Before the study, the purpose and methodology of the study were explained orally and in writing to the participants, and we obtained the athletes’ assent to participate and written informed consent from their parents/guardians. This study was approved by the Human Research Ethics Committee of Kagawa Nutrition University, Saitama, Japan (Approval No. 465).
Study design
The surveys were conducted during a 1-month period from April to May 2023. Participants completed the DRs on 3 nonconsecutive days: 1 weekday with sports practice, 1 weekday without practice, and 1 weekend day with sports practice, with approximately 3 days between each recording (Figure 1). Each participant completed the FFQJA immediately after completing their third DR to evaluate their dietary intake [9,18]. The interval between the third DR and the FFQJA administration was within 1 week.
DRs
Self-administered DRs have been used in previous studies of FFQs in Japanese populations [11,12,19] because most participants have sufficient knowledge and are capable of completing DRs independently. Regarding the number of days required for DRs, Murakami et al. [20] reported that ≥ 2 nonconsecutive days, including holidays, are appropriate. In the present study, DRs were completed on 3 nonconsecutive days within 2 weeks (2 weekdays and 1 weekend day), and the average daily intake was calculated.
For this validation study, registered dietitians devised the DRs, developed a manual of potential issues arising from weighing errors or omissions in recording, and conducted training to standardize survey techniques. Trained registered dietitians then explained the DRs to the study participants and their guardians, and the participants were given a dietary log, a ruler, and a scale. Participants and their guardians were instructed to record food and drink names, weight, or size, or a combination of these, in detail. Participants and their guardians were also asked to take photographs of all foods and drinks consumed. When processed food and drinks were consumed, the participants kept food labels. For dine-out and take-out foods, the name and weight of each food item were determined from the product labels and food photographs, along with the DRs. For food groups mentioned in the DRs that were not listed in the Food Composition Table, we used similar food items to calculate nutrient values. For participants who had school lunch on weekdays, details were obtained from the daily menu, and the weight of each food item was calculated. Participants then recorded the amount of food they consumed for each dish on a scale of 0%-100%. Details of the DRs were checked by registered dietitians or well-trained students enrolled in a registered dietitian course face-to-face at the time of submitting their records.
To calculate daily energy and nutrient intake according to the Standard Tables of Food Composition in Japan (Eighth Revised Edition, 2020) [21] and food group, the nutrient calculation software (Excel Eiyou Plus, Version 1.2; Kenpaku-sha, Tokyo, Japan) was used. Before the use of the program, standardization was conducted between three registered dietitians to ensure the consistency and reliability of the analyzed data.
FFQJA
The FFQJA was used to assess food intake over 2 weeks. The participants and their guardians were provided with instructions to complete the FFQJA. During this survey, participants were shown pictures of food items listed in the FFQJA to help them estimate their approximate food intake. In recalling their habitual diets, the participants indicated a portion size and weekly frequency for each food group. In this study, participants and their guardians were requested to complete the FFQJA together. A previous study showed that when children aged 4-14 years completed the FFQ together with a parent, the data obtained were among the most accurate [22,23]. The FFQJA included 62 questions covering food consumption, estimated energy intake, and the intake of 10 nutrients and 17 food groups [12].
Participants and their guardians provided information on portion size and frequency for each food item. Consumption frequency was reported separately for breakfast, lunch, dinner, and supplements. The consumption frequency of beverages and sweets was reported on a daily basis. In addition, answers to foods listed in other categories (eaten at least once a week) were obtained through a free-response format.
The FFQJA used typical values and/or natural units reported in the literature as standard portion sizes for foods in the DRs. The amount of food consumed was reported via open-ended responses regarding the number of times the standard portion size was consumed. The frequency of food consumption was classified into four categories: almost every day, three to four times per week, once or twice per week, and never or rarely. The FFQJA assessed food intake over 2 weeks, with “never or rarely” defined as less than once a week. Participants submitted the FFQJA to Musashigaoka Junior College. Daily energy and nutrient intakes were calculated using the specialized program developed for the FFQJA [12]. One registered dietitian performed the data entry and calculations.
Other measurements
Age was self-reported. Participants’ height was measured using a stadiometer (YHS-200D; YAGAMI, Inc., Nagoya, Japan), while body weight and body fat percentage were measured using the Inbody 470 analyzer (InBody Japan, Inc., Tokyo, Japan). Lean body mass was calculated by subtracting the amount of body fat from the body weight. Height and body weight were measured according to the standard protocol proposed by the International Society for the Advancement of Kinanthropometry (ISAK) [24] by ISAK-accredited anthropometrists.
Statistical analyses
To evaluate data normality, the Shapiro-Wilk test was used. Data were expressed as medians and interquartile ranges (IQRs) because most variables were not normally distributed. In 102 studies of FFQs, Lombard et al. [25] reviewed statistical tests that are applied to evaluate the validity of dietary intake assessment methods and concluded that the use of multiple statistical tests is helpful in analyzing the comprehensive validity of a dietary assessment method. To determine the validity of the FFQJA, nutrient intakes and intakes from each food group calculated from DRs and the FFQJA were compared with the use of crude Spearman’s correlation coefficients (CCs), energy-adjusted CCs, and intraclass CCs (ICCs). The residual method proposed by Willett et al. [26] was used to calculate the energy-adjusted variables. Crude CCs and energy-adjusted CCs were used to validate the individual ranking estimates. The difference and percentage differences were calculated to determine the validity of the mean estimates. The percentage differences were calculated according to the formula (FFQJA − DRs)/DRs × 100. The Wilcoxon signed-rank test was used to compare the FFQJA and DRs in terms of mean intakes of energy, nutrients, and food groups. To evaluate the degree of intake agreement between the two methods, we used the following levels of ICCs: very good (median ICCs ≥ 0.8), good (0.60-0.79), moderate (0.40-0.59), fair (0.20-0.39), and poor (≤ 0.19) [27].
To estimate the categorization validity, the participants in the validation study were divided into four categories according to the intake of each nutrient. Subsequently, we used the kappa statistic to evaluate the degree of coincidence between categories regarding the two survey methods. The degree of agreement for the weighted kappa values (κw) was characterized as poor (κw, <0.20), acceptable (κw, 0.20-0.59), and good agreement (κw, ≥ 0.60) [27]. We then compared the proportions of participants classified into the same, same and adjacent, and completely opposite categories based on the two methods (Figure 2). We considered p values of <0.05 statistically significant. All statistical analyses were performed with JMP software, version 14.3.0 (SAS Institute, Inc., Cary, NC, USA).
RESULTS
Participant demographics
The median age, height, and weight of participants were 13 years (IQR, 12-14 years), 161.1 cm (IQR, 153.9-166.4 cm), and 47.8 kg (IQR, 41.0-53.6 kg), respectively. Similarly, median body mass index, 18.5 kg/m2 (IQR, 17.0-20.1 kg/m2); median body fat percentage, 10.1% (IQR, 8.1%-13.0%); and median lean body mass, 37.8 kg (IQR, 33.8-43.7 kg).
Validation of the FFQJA for assessing nutrient intake
The median crude CC for nutrient intake was 0.483; the CCs ranged from 0.350 for retinol to 0.708 for iron (Table 1). The median energy-adjusted CC was 0.477 (range: 0.340 for vitamin C to 0.722 for calcium). Compared with the DRs, the FFQJA tended to underestimate intake, with a median difference of −7.2% for nutrient intake. Iron (23.7%) and dietary fiber (27.7%) were the nutrients showing differences of −20% or more between the two methods, while retinol (39.2%) showed the largest positive difference (≥ 20%). The median ICC was 0.469 (range: 0.118 for retinol to 0.719 for vitamin B2).
Validation of the FFQJA for assessing the intake of each food group
The median crude CC of each food group was 0.386 (range: −0.077 for sugar to 0.684 for dairy products; Table 2). The median energy-adjusted CC was 0.384 (range: −0.122 for sugar to 0.744 for dairy products). In comparison with the information in the DRs, the FFQJA tended to underestimate intake. A median difference of −17.6% was observed for nutrient intake; for 9 of the 17 food groups, the differences were −20% or more. In contrast, fruits (69.0%) and sweets (124.8%) were the food group that showed the difference of ≥ 20% between the two methods. The median ICC was 0.214 (range: −0.292 for seasonings to 0.734 for dairy products).
Cross-classification analysis
The median κw for nutrient intake was 0.350 (range: 0.200 for retinol to 0.525 for iron; Table 3). The median proportions were 41% for the same category, 83% for the same or adjacent category, and 3% for the completely opposite categories.
The median κw for intake of each food group was 0.275 (range, −0.055 for sugar to 0.535 for dairy products; Table 4). The median proportions were 34% for the same category, 77% for the same or adjacent category, and 6% for the completely opposite categories. The items with completely opposite ratios of ≥ 10% were sugar (22%) and nuts and seeds (17%).
DISCUSSION
The present study found that in the FFQJA, which was developed and validated to estimate the food intake of Japanese athletes, provided reasonably valid measurements of habitual nutrient and food group intake for junior high school athletes. However, participants underestimated or overestimated the intake of some nutrients and food groups.
The crude and energy-adjusted CCs for each nutrient intake and food groups found in this study were similar to those reported in studies in which other FFQs were administered to Japanese populations (the median of CCs for target nutrients range: 0.42-0.52) [4] and Japanese athletes (the median energy-adjusted CCs for nutrient intake was 0.478; range 0.270-0.584, and the median ICC was 0.369) [12]. Furthermore, the crude CCs for nutrient intake and each food group were higher than those reported in studies in which other FFQs were used for Japanese adolescent athletes (the median crude CCs for nutrient intake was 0.36, and for each food group 0.30) [11]. The number of questions in the FFQJA may have influenced these results. The FFQs used in previous studies for middle-aged and older adults consisted of> 400 questions, whereas the number of questions in the FFQJA is only 62. A review of validation studies in children and adolescents revealed that questionnaires with 20-60 items were associated with higher validity [28] than were those with more questions. Therefore, the limited number of questions in the FFQJA was more useful for junior high school athletes. These findings indicate that the FFQJA is a valid tool for estimating the nutrient intake of junior high school athletes.
Energy-adjusted CCs for calcium, iron, vitamin B1, and vitamin B2 and for the dairy products food group were high for junior high school athletes according to the responses on the FFQJA, particularly for calcium in this study (0.722). A systematic review of dietary assessment methods for adolescents to date indicated that the mean CC for calcium in eight studies was good (r = 0.58) [29]. In addition, the FFQJA results for Japanese athletes had a CC of 0.584 [12]. Our results surpassed those of previous Japanese population studies because high interindividual variability may have increased CCs. The IQR of calcium intake estimated from the DRs in this study (775-1294) was wider than that of Japanese athletes in another study (415-670) [12]. When the IQR is wide, much intraindividual variation is significant; the between-person variation in intake of a food group is therefore large, and the CC for validity is increased [4]. For these reasons, the CCs in this study may have been elevated. Our findings suggest that a FFQJA may be useful for ranking individual estimates of intake of nutrients and food groups among junior high school athletes; for calcium, iron, vitamin B1, and vitamin B2; and for the dairy food groups. In particular, CCs were greatly useful for estimating intake of calcium, iron, vitamin B1, vitamin B2, and dairy products.
In this study, junior high school athletes tended to underestimate or overestimate food intake on the FFQJA, more so for food groups than for nutrients. Underestimation and overestimation were notable for sugar (100.4%), nuts and seeds (100.0%), fruits (69.0%), sweets (−124.8%), seaweed (−91.7%), potatoes (−73.0%), and seasonings (−57.6%). The large difference in nuts and seeds and in seaweed can be attributed to their small intake amount, where even a minor difference resulted in a large percentage difference. In contrast, despite sufficient intake levels, the percentage difference was large for fruits, sweets, and potatoes. This was due to daily variations. All nutrient and food group intakes are reported on a daily variation [30]. Small differences in food eaten during the study period may have affected fruit, sweets, and potatoes intake because the study period for the DR (nonconsecutive 3 days) and FFQJA (2 weeks) was different in this study. In addition, the questions about sugar and seasonings are not included in FFQJA. This was because many seasonings are used in Asian cuisine, and athletes who do not cook for themselves may have difficulty accurately assessing their intake [12]. Therefore, when using FFQJA to assess the dietary intake of junior high school athletes, fruit, sweets, potatoes, and seasonings had high percentage differences. Our results showed that for some nutrients and food groups, absolute intake could not be accurately ascertained from the FFQJA for junior high school athletes, and the intake values from the FFQJA cannot be directly compared with that from the DRs. This result was consistent with findings from studies on athletes [10,12,31]. For these reasons, future studies may require modifications to the questionnaire items regarding sugars and seasonings.
The degree of intake agreement for nutrients and food groups was moderate and fair, respectively, in this study. The median ICCs for nutrients or food groups obtained in this FFQJA were comparable with those for previous Japanese athletes [12], which suggests that the degree of coincidence between the FFQJA and DRs was within a moderate range for many nutrients and food groups. In this study, ICCs exceeded 0.6 for four nutrients (calcium, iron, vitamin B1, and vitamin B2) and for the dairy products food group. In previous studies of adolescent athletes, calcium and iron deficiencies have been reported [15,32]; according to one, iron stores were depleted in 31% of young male athletes and 57% of young female athletes [33]. The need for vitamins B1 and B2 increases the energy demand during physical exercise, which can alter the requirements [34]. For these reasons, the FFQJA, which enables the estimation of calcium, iron, vitamin B1, vitamin B2, and dairy product levels, might be useful in epidemiology studies that aim to identify the relationship between food intake and the condition of junior high school athletes.
The degree of coincidence of categorization among nutrients and food groups was acceptable in this study. The median percentages of participants classified into the same or adjacent categories were 83% for nutrient intake and 77% for each food group. These results showed similar or higher proportions in the same or adjacent category in comparison with results of previous studies of Japanese athletes [9,10,12]. The median proportions of the opposite category between the FFQJA and DRs for both nutrients and food groups were within 6%, although large missclassification were observed for sugar (22%) and nuts and seeds (17%). In this study, the proportions of each nutrient and food in the opposite category were similar to those in previous studies of Japanese athletes [9,10,12]. Accurate classification indicates the extent to which the dietary intake assessment method can rank participants correctly [25]. Our results suggest that the FFQJA could be useful for categorizing nutrients and food groups.
This study had several limitations. First, all participants were from the same soccer team and were not randomly selected; therefore, the participants could not be considered representative of the general Japanese population of junior high school athletes. Second, the FFQJA data were collected immediately after the third DRs in the main study, this study may have introduced memory bias into the FFQJA results. Therefore, the results of this study may be affected accuracy, it is difficult to determine at this point whether the impact is being overestimated or underestimated. Third, DRs were used as a reference method. DRs are prone to measurement error associated with portion size awareness. In addition, biomarkers were not used to validate the FFQJA in this study. Results of a previous review of the validity of dietary assessments in athletes suggested that energy intake is usually underestimated in comparison with energy expenditure [31]. A biomarker was used in a previous validation study to assess the accuracy of reported intake [35]. In the future, it will be necessary to use biomarkers to verify validity. Fourth, in this study, DRs were filled out on 2 weekdays and 1 weekend day; however, because training intensity and volume could not be controlled, these factors may have influenced DRs. Fifth, the reproducibility of the FFQJA could not be assessed because a dietary survey with the FFQJA could not be conducted before the first DRs. Future research is needed to assess the reliability of the FFQJA among junior high school athletes. Specifically, the actual FFQJA must be administered twice at an interval of several weeks to assess the consistency of dietary reporting over time to strengthen the validity and applicability of the FFQJA.
In conclusion, the validity of this FFQJA in junior high school athletes is comparable to that in adult athletes, for which it was originally developed. Therefore, this FFQJA might be useful for assessing habitual dietary intake in junior high school athletes, especially for calcium, iron, vitamin B1, and vitamin B2 and for the dairy products. Nonetheless, the underestimates or overestimates of the intake of some nutrients and food groups on the FFQJA suggest that caution is required when the mean dietary intake of junior high school athletes is estimated. The FFQJA could be used to screen for nutrient and food group deficiencies in intervention programs to assess nutrients or food groups and is a suitable tool for ranking individuals and categorization among Japanese junior high school athletes.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
