Comparative Study of a New Dietary Screener to Assess Food Groups of Concern in Children

Abstract

Background:

Although there are several valid and reliable dietary screeners to measure child intake patterns, there is a paucity of brief assessment tools targeting under- and overconsumed foods.

Objective:

To compare the Fueling Learning through Exercise study (FLEX) dietary questionnaire, a screener designed to assess consumption patterns in third to fifth graders, to a validated dietary assessment tool.

Methods:

The FLEX dietary questionnaire was developed to assess fruit, vegetable, snack, and beverage consumption and was compared to the Block Kids Food Screener (BKFS). Correlations were analyzed using Pearson correlation coefficient. Agreement was assessed using Bland-Altman plots.

Results:

The sample (n = 63) had mean age of 9.9 years (SD 0.7). Most participants were non-Hispanic white (70%) and eligible for free/reduced price lunch (57%). Correlations between food group categories were significant for all groups (P < .05) except fruits (r = 0.51) and sugar-sweetened beverages (SSBs) (r = 0.21). We found moderate-to-strong correlations between reported vegetable, salty snack, sweet snack, total beverage, milk, and fruit juice consumption (0.62,0.59, 0.69,0.47,0.48, and 0.46, respectively). The FLEX screener reported systematically higher mean servings per day (0.24–1.1) compared to the BKFS (0.05–0.51).

Conclusion:

Based on these correlations, the FLEX dietary questionnaire performs similarly to a validated tool in assessing intake of under- and overconsumed food groups in a diverse third to fifth grade population. Overall serving size discrepancies are likely due to more relevant food items on the FLEX questionnaire and a more child-friendly format. This study highlights the need to update older diet screeners to reflect current child consumption patterns.

Background

Despite continued efforts to improve child and adolescent diets at the local and national level, research demonstrates that children in the United States do not consume enough fruits, vegetables, or milk to meet the recommended daily servings as advised by the Dietary Guidelines for Americans (DGA).¹ At the same time, sweetened beverages and highly processed snacks are consumed in quantities higher than recommended. Accurate measurement of children’s dietary intake is critical to understanding the role of nutrition in short- and long-term health. While methods such as the 24-hour recall are regarded as the most accurate, these approaches are time-intensive and costly.² Food recall techniques are more feasible within school and other community research settings; however, issues with portion size estimation,^3,4 attention span,⁵ and memory^6,7 make administering recalls a challenge with children. In addition, few researchers have examined the reliability of food recalls in minority or underrepresented child populations.^8–11 Given the high burden of collecting dietary information with children, there is a need for valid, reliable, culturally appropriate, and brief instruments to gather data on specific food groups that are commonly under- or over-consumed by children.¹⁰

The Fueling Learning through Exercise (FLEX) study, a cluster-randomized controlled trial conducted at elementary schools in Massachusetts, was designed to evaluate the impact of 2 physical activity programs on a variety of health measures.¹² The intervention targeted third to fifth grade children from diverse, low-income school districts. Dietary intake was assessed using the Block Kids Food Screener (BKFS), which has been validated in children aged 8 to 17 years.^8,13,14 However, field testing demon strated that the BKFS format was difficult for FLEX students to comprehend, and foods included on the survey were not culturally appropriate. The most recent iteration of the DGA includes dietary recommendations based on specific food groups to choose or limit, signaling a shift from nutrient to food pattern–based recommendations.¹ The FLEX researchers therefore developed a food group–targeted dietary screener to assess underconsumed foods (fruits, vegetables, unsweetened beverages) and overconsumed foods (sweet and salty snacks and sweetened beverages) using a child-friendly format and relevant food categories for diverse child populations.¹⁵ “Food groups of concern” is used to refer to these commonly under- or overconsumed food groups. Both the FLEX dietary questionnaire and BKFS were administered to a subsample of the students enrolled in FLEX during a study visit. Although the FLEX questionnaire is adapted from validated measures,^11,14,16 it has not been compared to existing food frequency tools.

Objective

Given that there is a dearth of existing tools that adequately capture specific food groups of concern in children, evaluating newly developed dietary screeners is of high public health utility. Hence, the aim of this short communication is to describe comparison of the FLEX dietary screener, a brief food frequency questionnaire (FFQ) designed to assess fruit, vegetable, sweet and salty snack, and beverage consumption in third to fifth graders, to the BKFS.

Methods

Study Population and Data Source

This comparative study uses postintervention study visit data from the FLEX study. Participant recruitment and data collection occurred during 2 study waves (wave 1 and wave 2). Data used for this study come from wave 1 of data collection. Further details of the FLEX study protocol are described elsewhere.¹² In brief, FLEX aims to evaluate the impact of 2 school-based physical activity programs on third to fifth graders’ moderate-to-vigorous physical activity and cognitive performance. Apart from 1 school from a district with 34% low-income students, all enrolled districts had greater than 40% student eligibility for free or reduced price lunch and/or 40% non-Caucasian students. Data collection took place at 3 time points (baseline, midpoint, and postintervention) over the course of approximately 1.5 years. Data collection for the present study occurred during wave 1 in the 2014 to 2015 school year (n = 6 schools). The institutional review board (IRB) of Tufts University (Medford, Massachusetts), as well as individual school district IRB/research boards where required, approved the study.

Food Frequency Questionnaires

During wave 1 baseline and midpoint data collection, the BKFS, a validated tool for assessing child and adolescent eating patterns,^8,13,14 was used to assess dietary intake. The screener contains 39 questions in Scantron form, asking about frequency (how many days) and quantity (amount per day) of specified foods eaten over the past week. For food groups such as sweetened beverages, examples of included items are listed (eg, “drinks like Coke or 7-Up, Sunny Delight, Hawaiian Punch, or aguas frescas). Portion sizes are matched to standard serving size equivalents per food type (1 glass, 2 glasses, and 3 glasses for juice and milk; 1 bottle, 2 bottles, and 3 bottles for sodas; 1/2, 1, and 2 for fruits and vegetables; a few, small bag, and large bag for chips; and a little, some, and a lot for other snacks). Field administration demonstrated that the format of the BKFS was challenging for FLEX participants, and the food list was not relevant to usual consumption patterns in the study population. Therefore, the FLEX dietary questionnaire was developed and tested with a subset of fourth and fifth grade participants at the wave 1, postintervention visit. Development of the tool took into consideration the length of the entire study (a 2-year period, including third graders at baseline of Wave 2) and was therefore intended for use by third to fifth grade participants.

The FLEX dietary questionnaire is adapted from several validated FFQs.^11,14,16 The survey includes 39 questions about the frequency (number of times in the past week) and amount (a little, some, and a lot) of foods commonly consumed by the study population.^17–19 Food categories were developed based on previously used categories in validated FFQs, as well as over- and underconsumed foods identified by the DGA for this population.^1,14 Similar to the BKFS, examples of foods included in each category are given under most categories, with additional clarification added as needed. For example, the fruit-flavored sweet drinks category reads, “Fruit-flavored sweet drinks (such as lemonade, fruit punch, flavored teas, Capri Sun, Snapple, Sunny D, and Kool-Aid)” and the starchy vegetables category reads, “Starchy vegetables (such as corn, green peas, potatoes, plantains, cassava, and taro) DO NOT COUNT POTATO CHIPS OR FRENCH FRIES. Portion sizes are matched to standard serving size equivalents (a little = 1/2 serving; some = 1 serving; a lot = 1.5 servings). The questionnaire is not designed to assess total dietary intake, instead asking about categories of foods to promote fruits, vegetables, and unsweetened beverages and those to limit sweet and salty snacks and sugar-sweetened beverages (SSBs).

Procedure

A study subsample of 63 students completed both the BKFS and FLEX dietary questionnaire at the postintervention data collection time point from March 21 to March 28, 2016. Trained research assistants administered the 2 instruments on the day of data collection during school hours. Instructions were read aloud to students in small groups and then screeners were self-administered in the presence of a research assistant. Questions were read aloud individually to students as needed.

Food Group Selection

To compare the FLEX dietary questionnaire to the BKFS, foods reported on both surveys were first grouped into common categories. Each category comprised the most similar items that appeared in both surveys (Table 1). The fruits category included apples, pears, bananas, grapes, citrus fruit, packaged fruit, and applesauce; vegetables included nonstarchy vegetables, starchy vegetables, pulses, and beans; salty snacks included chips, popcorn, French fries, and tater tots; sweet snacks included cookies, pastries, ice cream, and other frozen desserts; and the total beverages category included white milk, flavored milk, 100% fruit juice, and soda. The milk category included both flavored milk and white milk. We were unable to use flavored milk and white milk as separate categories because this distinction was not made in both the BKFS and FLEX questionnaires. The SSB category included regular soda and fruit-flavored sweetened drinks; diet soda was excluded since it is not listed on the BKFS. Foods not covered on the FLEX dietary questionnaire, such as grains and protein, were excluded from the analysis. Water, sports, and energy drinks are included on the FLEX questionnaire, but were omitted from the validation analysis since they are not included on the BKFS.

Table 1.

Mean Servings of Foods Consumed Daily, as Reported by the Block Screener and FLEX FFQ, Validation Subsample at Postintervention.

	Raw Means		Transformed Means
	Block Screener	FLEX FFQ	Block Screener	FLEX FFQ
Food groups (n)	Mean (SD)	Mean (SD)	Mean (SD)	Mean (SD)	Correlation Coefficient
Fruitsa (62)	0.51 (0.52)	0.73 (0.77)	0.69 (0.29)b	0.74 (0.40)b	0.5l
Total vegetablesc (60)	0.37 (0.39)	0.59 (0.78)	0.59 (0.32)b	0.64 (0.43)b	0.62d
Salty snackse (61)	0.17 (0.23)	0.43 (0.53)	0.45 (0.25)b	0.62 (0.3l)b	0.59d
Sweet snacksf (62)	0.22 (0.26)	0.31 (0.38)	0.38 (0.26)g	0.45 (0.33)g	0.69d
Total beveragesh (61)	0.51 (0.43)	1.10 (0.85)	0.64 (0.3l)g	0.96 (0.42)g	0.47d
Milki (63)	0.27 (0.31)	0.49 (0.53)	0.41 (0.3l)b	0.56 (0.42)b	0.48d
Fruit juice (63)	0.20 (0.26)	0.34 (0.39)	0.35 (0.28)g	0.46 (0.37)g	0.46d
Sugar-sweetened beveragesj (59)	0.05 (0.09)	0.24 (0.30)	0.13 (0.l8)b	0.38 (0.3l)b	0.2l

Abbreviations: FLEX, Fueling Learning through Exercise; FFQ, food frequency questionnaire.

^aIncludes apples, pears, bananas, grapes, citrus fruit, packaged fruit, and applesauce.

^bCube root transformation.

^cIncludes nonstarchy vegetables, starchy vegetables (excluding French fries and tater tots), and beans.

^dP < .05.

^eIncludes chips, popcorn, French fries, and tater tots.

^fIncludes cookies, pastries, ice cream, and other frozen desserts.

^gSquare root transformation.

^hIncludes white milk, flavored milk, 100% fruit juice, sweetened fruit drinks, and soda.

ⁱIncludes flavored milk.

^jIncludes regular soda and sweetened fruit drinks and excludes diet soda.

Statistical Analysis

Servings per day were calculated by dividing the total servings reported on both surveys by 7. Observations more than 4 standard deviations from the mean were dropped from the analysis as outliers before carrying out variable transformations. Each outlying observation was individually labeled as missing so that it would be excluded from analyses, while still retaining the participant’s other food category observations. Therefore, sample size varied slightly between correlation tests. All dietary variables were positively skewed due to a significant number of zeroes reported on each category; therefore, dietary variables were transformed for use in correlation analyses. A square root transformation was used for the sweet snack, fruit juice, and SSB variables while a cube root transformation was performed for all other variables (fruits, total vegetables, salty snacks, total beverages, and milk). Transformed food group variables were individually compared to assess correlations between the 2 dietary instruments. Correlations were categorized as strong, moderate, or small according to the classification proposed by Cohen,²⁰ where R > 0.5 is considered strong, 0.5 > R < 0.3 is considered moderate, and R < 0.3 is considered small. Bland-Altman plots were used to assess overall agreement between the 2 instruments. Analyses were performed using Stata version 13 (Stata-Corp; College Station, Texas).

Results

Descriptive Statistics

As presented in Table 2, the study sample included 63 fourth and fifth graders with a mean age of 10.0 years (SD 0.77). The majority were girls (58.7%), non-Hispanic white (69.8%), and eligible for free- or reduced price lunch (57.1%).

Table 2.

Demographic Characteristics of the Fueling Learning Through Exercise Study Population, Validation Subsample at Postintervention.

	Postintervention (n = 63)
Sex, n (%)
Boys	26	(4l.3)
Girls	37	(58.7)
Age (years), mean (SD)	9.9	(0.8)
Race/ethnicity, n (%)
Non-Hispanic white	44	(69.8)
Non-Hispanic black	2	(3.2)
Mulitracial/Asian/American Indian/other	6	(9.5)
Declined to respond	4	(6.5)
Free/reduced price lunch eligibility, n (%)
Yes	36	(57.l)
No	23	(36.5)
Declined to respond	4	(6.4)
Maternal education, n (%)
High school degree or less	l0	(l5.8)
Some college or associate’s degree	24	(38.l)
Bachelor’s degree or above	24	(38.l)
Declined to respond	5	(7.9)

Abbreviation: SD, standard deviation.

Diet Screener Comparison

Overall mean intake was highest for total beverages (BKFS: 0.51, FLEX: 1.10), fruits (BKFS: 0.51, FLEX: 0.73), and total vegetables (BKFS: 0.37, FLEX: 0.59), while it was lowest for SSBs (BKFS: 0.05, FLEX: 0.24), sweet snacks (BKFS: 0.22, FLEX: 0.31), and fruit juice (BKFS: 0.20, FLEX: 0.34) as presented in Table 1.The FLEX dietary questionnaire estimates of consumption were consistently higher than that of the BKFS. Relative servings on both surveys were higher for fruits, vegetables, and milk compared to sweet snacks, salty snacks, and SSBs. The correlations were strong for fruits (0.51), vegetables (0.62), salty snacks (0.59), and sweet snacks (0.69), moderate for total beverages (0.47), milk (0.48), and fruit juice (0.46),²⁰ and small for SSBs (0.21). The correlations were significant (P < .05) for all food groups except fruits and SSBs. Both screeners reported a significant number of zeroes in all categories (Table 3). The fruit and SSB categories had the largest discrepancies between zeroes, with a particularly high discrepancy for SSBs (n = 36 and n = 17 for the BKFS and FLEX questionnaire, respectively); zeroes were similar for all other categories. Examination of Bland-Altman plots (Figure 1) indicated that there was good agreement between the 2 instruments for all food and beverage categories.

Table 3.

Number of Zeroes in Each Food Group as Reported by the Block Screener and FLEX FFQ.^a

Food Groups, Servings Per Day	Block Screener	FLEX FFQ
Fruits	4	l0
Total vegetables	9	l3
Salty snacks	9	7
Sweet snacks	l0	l3
Total beverages	3	2
Milk	l4	l3
Fruit juice	l5	l8
Sugar-sweetened beverages	36	l7

Abbreviations: FLEX, Fueling Learning through Exercise; FFQ, food frequency questionnaire.

^aN = 63.

Figure 1.

Bland-Altman plots showing mean agreement (—————) and 95% limits of agreement* (——) between transformed variables of servings of food groups per day measured on the Block Kids Food Screener for ages 8 to 17 years (BKFS) and FLEX dietary questionnaire (FLEX). *95% limits of agreement are calculated as mean of the difference between the BKFS and FLEX ±2SD. **Cube root transformed variables. Ŝquare root transformed variables. FLEX indicates Fueling Learning through Exercise.

Discussion

The FLEX dietary questionnaire was developed in response to a need for a relevant, simple to administer assessment tool to measure under- and overconsumed food groups in diverse child populations. Dietary patterns that are high in fruits, vegetables, and unsweetened beverages and low in sweet and salty snacks and SSBs are linked to improved health outcomes and lower weight status in children.^1,21–23 This study demonstrates that the FLEX questionnaire has good comparability to a validated tool in its ability to assess vegetable, sweet snack, salty snack, and total beverage consumption. Compared to similar screeners, the FLEX dietary questionnaire has several important strengths. The instrument was designed to measure specific food groups of interest to yield estimates of key nutrient-dense and nutrient-poor foods. The targeted nature of the screener makes it suitable for time-pressured data collection, such as within school or afterschool environments; although both screeners contain the same number of questions, the FLEX survey is separated into distinct food categories to address issues of attention span and concentration in younger age groups and is useful to address research questions involving dietary patterns. The FLEX questionnaire food list was updated to include commonly consumed items (eg, flavored waters, energy drinks, dried fruits, crackers, nuts, and seeds) within this population.^17–19 The FLEX dietary questionnaire took approximately 10 to 15 minutes, on average, for the students to complete, while the BKFS took 15 to 20 minutes on average. Additionally, a portion of the children required extra support from research assistants to properly complete the BKFS (due to skipping rows). This demonstrates that the format of the FLEX screener may be more appropriate for children in this age-group.

Although the FLEX diet screener systematically overestimated food group servings compared to the BKFS, the observed correlations were statistically significant and moderate to high in all categories apart from fruits and SSBs. Bland-Altman plots demonstrated good agreement between the 2 instruments. Some variability between the 2 instruments may be explained by the large number of reported zero servings, particularly for SSBs. Over 50% (36) of participants reported zero SSB consumption on the BKFS, compared to just over 25% (17) of participants who reported zero SSBs on the FLEX questionnaire, suggesting that the BKFS did not fully capture SSB consumption. The large discrepancy in reported mean SSB servings (0.05 vs 0.24 for the BKFS and FLEX questionnaire, respectively, Table 1) may be explained by the additional SSB categories included on the FLEX questionnaire; the FLEX screener asked about sweetened fruit drinks and sodas in 2 separate questions, while the BKFS included both types in 1 question. Recall may be improved by seeing more examples of drink options. There is also some evidence that children are more likely to report consuming foods that they enjoy, a phenomenon that is particularly noticeable with SSBs and snacks.²⁴ However, relative servings of foods to encourage (fruits, vegetables, and milk) were higher than foods to discourage (sweet and salty snacks and SSBs) on both surveys, suggesting a reporting bias for healthier foods that has been noted in similar age groups.²⁵

Reported mean servings per day were higher across all food categories on the FLEX questionnaire relative to the BKFS. This is likely due to the different construction of the 2 questionnaires, with the BKFS designed to assess total dietary intake, and the FLEX screener targeted to capture only specific foods categories. The BKFS asks about mixed dishes and may capture information about foods such as fruits and vegetables within these categories; this may lead to relatively lower estimates of foods when only questions specific to categories like fruits and vegetables are considered. The FLEX questionnaire included more questions for each food category than the BKFS since it aimed to capture all foods consumed within categories, which would theoretically improve recall of items. Other differences may be due to discrepancies in serving size descriptors between surveys: The FLEX questionnaire framed all portions as “a little, some, and a lot,” while the BKFS included several descriptors (“a little, some, and a lot”; “1 glass, 2 glasses, and 3 glasses”; “a few, small bag, and large bag”).

The Scantron format of the BKFS may have resulted in lower estimations within the FLEX study population. Data collectors reported that the students struggled with this “fill in the bubble” structure. The FLEX questionnaire was designed to mimic more familiar layouts that the students encounter regularly at school, asking participants to check the boxes that best fit their consumption pattern and providing clear, bolded instructions above each section. It should also be noted that reported consumption patterns on both screeners were low for all categories compared to national estimates,^17,26,27 suggesting that issues of recall, attention span, and portion size estimation noted in similar studies were present.^3–7

This study had several limitations. The small sample size (63 participants) may not yield enough statistical power to draw rigorous conclusions within certain food groups. However, this sample size is comparable to similar validation studies.^8,10,14 Additionally, this sample was made up of predominantly non-Hispanic white children (69.8%), which limits generalizability to other ethnic populations. The inherent differences between the 2 screeners make direct comparison a challenge. Further research is needed to determine the validity and reliability of the FLEX questionnaire within a larger population and against gold standard assessment methods such as 24-hour recalls. In the interim, the FLEX questionnaire appears to perform similarly to the BKFS, with the additional benefit of shorter administration time and more relevant food categories within an age-appropriate format.

Conclusion

The results indicate that the FLEX dietary questionnaire performs similarly to the BKFS in assessing food groups of concern in diverse elementary school-aged children. Overall correlations and agreement between the 2 instruments were high. Discrepancies between the 2 screeners highlight the need to update older instruments to reflect current consumption patterns in the United States and adhere to child-friendly formats. Future research is needed to validate the screener against gold standard dietary assessment methods.