Evaluation of a School-Based Fruit and Vegetable Intervention Using a Digital Photography Method

Objective: The primary objective is to use digital photography of food to assess if a policy, systems, and environment (PSE) intervention increases the amount and variety of healthful fruits and vegetables consumed at lunch by low-income 5th graders. Design: Quasi-experimental design. Participants/Setting: The sample consisted of low-income 5 grade students in the Providence School District, n = 130 (treatment school, n=75, control school, n=55). Intervention: Both the treatment and control students were assessed at baseline and post intervention to determine amount consumed and variety of fruit and vegetable using a digital photography of food method. Within the treatment school, four out of the six classrooms agreed to participate in the 8-lesson Student’s Take Charge in-class intervention. Main Outcome Measures: Consumption of healthful fruits and vegetables (excluded French fries, tomato sauce, and fruit juice) in cups and variety from digital photographs. Analyses: Consumption differences were assessed using the Mann-Whitney U for between group changes and Wilcoxon signed-rank test for within groups. Pearson chi-square compared variety. Results: There was a difference between groups for change in consumption of fruit (p<0.01); within group analyses showed the treatment group decreased (M=0.12 cups, SD=0.46, p= 0.02) while the control group had a non-significant increase (M=0.12 cups, SD=0.49, p=0.20). There were no differences between or within groups for vegetable consumption (p>.05). Within treatment group variety of fruits decreased (Baseline: 59.3% with 1 or 2 fruits; Follow-up: 30.9%), and within the control group, variety of fruits increased (Baseline:11.7% with 1 or 2; Follow-up: 33.3%). Variety of vegetables decreased in the treatment group (Baseline: 49.4% with 1 or 2; Follow-up: 8.6%), and variety of vegetables was minimal at both time points within the control group (Baseline: 6.7%; Follow-up:8.3%). Although at baseline treatment children were more likely to have at least one fruit (59%) or vegetable (49%) than children in the control school (fruit 12%, vegetables 7%; p<.001), at follow-up most students in both groups had no fruit (6769%) or no vegetables (91-92%) on their trays (p>.05). Conclusions and Implications: The intervention was not successful in increasing consumption or variety of fruits and vegetables. The observed decrease in fruit consumption in the experimental group may be associated with different fruit options on observation days. Future research should explore changes in the eating environment to increase availability of preferred healthful fruit and vegetable options.


LIST OF TABLES AND FIGURE
Healthy Americans 1 . Low FV intake and lack of variety have been associated with higher body weight in children [2][3][4][5] . It is important for children to consume healthful fruits and vegetables 5 . Healthful fruits and vegetables are those low in saturated fat, sodium, and high in fiber, and excludes items such as French fries, fruit juices, and tomato sauce 5,6 .
Thirty one percent of children age 10-17 in the United States were overweight or obese in 2011-2012 7 . The prevalence was slightly lower at 28.3% of all children in Rhode Island being overweight or obese and increased with 38.6% Hispanic children being overweight or obese 3,7 . The Rhode Island Supplemental Nutrition Assistance Program Education (SNAP-Ed) has developed "Students Take Charge!" (STC), a policy, systems, and environment (PSE) program focusing on empowerment and a healthy lifestyle. STC is an in-class PSE curriculum for 5th grade students. Although process data from the pilot year of STC indicated that students had an increased FV consumption, outcome analyses failed to find any dietary changes in FV consumption 8,9 .
The lack of effect of STC on intake may be related to the use of a two-item FV survey to measure consumption 8 . Measurement of intake is difficult in children and particularly challenging among 5 th graders in low-socioeconomic populations 10,11 . Digital photographic methods such as the Digital Photography of Food Method (DPFM) have been validated to objectively measure consumption, but until now have not been used to assess the effectiveness of PSE interventions in low-income schools [12][13][14] .
The primary aim of this study was to use DFPM to determine if the STC intervention increased the amount of healthful fruits and vegetables consumed at lunch by low-income 5 th graders in an intervention school more than a control school. The secondary aim of this study was to determine whether the STC intervention increases the variety of healthful fruits and variety of vegetables chosen at lunch in the intervention school more than control school.

Design
The PSE intervention, "Students Take Charge!" (STC) utilized a 2x2 quasiexperimental design. STC is a research study that was piloted in the previous year and modified based on feedback reported by Lepe et al. in their process evaulation 9 . The current STC classroom-based program consisted of eight lessons taught by SNAP-Ed educators. The objective was to empower low-income elementary students to increase FV consumption and to choose a variety of FVs. Table 1  DPFM allows for objective measurement of consumption and variety of FVs without relying on self-report. Within the treatment school, four out of the six classrooms agreed to participate in the in-class intervention. Two classrooms within the treatment school declined to participate in the education component, but were exposed to environmental changes such as the recipe day in the cafeteria and promotional posters. The treatment group with education received an 8-lesson curriculum, selected a recipe for a fruit or vegetable item for school lunch, and participated in organizing an all-school taste testing during school lunch on recipe day. The two classrooms that did not participate in the education, participated in taste testing a recipe in the cafeteria and were assessed using DPFM at the two time points. The control school did not receive any intervention.
Demographic information was collected from the students in the treatment group with education and the control school only. Demographic information was not collected for the treatment group without education.

Research Question/Hypothesis
All students were from the fifth grade, and all comparisons are baseline to follow-up: i. Primary: Students in the intervention group will increase the amount of healthful FVs consumed at lunch more than students in the control group.
ii. Secondary: Students in the intervention group will increase the variety of healthful FVs selected at lunch more than students in the control group.
iii. Exploratory: Within the intervention school, the four classes that participated in the STC intervention will increase the amount of healthful FVs consumed at lunch more than the two classes that did not receive the intervention.

Subjects
Providence, Rhode Island is a diverse city 15  Primary and secondary aims included subjects from control and intervention schools, and the exploratory aim only included 5 th graders from the intervention school.
The intervention school had six, 5 th grade classrooms, four of which agreed to have their students participate in the school-based intervention, n=75. DPFM data from the other two classrooms were collected, n=34. These students were exposed to the school-wide taste testing on recipe day and signage promoting FV but did not receive the in-class education, STC. The control school had three 5 th grade classrooms that were assessed within two weeks of the intervention school data collection points, n=55. Each classroom had around 25 students however due to absences every student's tray was not photographed, resulting in an overall analytical sample of 164 students. Only students with four pictures, two at baseline pre and post meal, and two at follow-up pre and post meal were included for hypotheses 1 and 3, and only students with one baseline pre-meal photo and one follow-up pre-meal photo were included for hypothesis 2. For hypothesis one, n=75 from the treatment school with education and n=55 from the control school.
For hypothesis two, n=81 from the treatment school with education and n=60 from the control school. For the exploratory hypothesis, n=75 from the treatment school with education and n=34 from the treatment school without education group.

Student Eating Environment
Students had access to FV via two avenues in the cafeteria: on the tray line and in the garden carts. Providence Public School District (PPSD) schools provide students with a garden cart in all cafeterias. The garden cart is a salad bar that consists of an assortment of FVs available to the students without any restriction on quantity. Students were free to select how much and how many FV they would like from these carts. The garden cart placement is different from school to school.

Procedure
For the STC program, key stakeholders identified two schools in the PPSD. Once the two schools agreed, they were randomly selected as the treatment or control. Students in both schools completed demographic surveys at the same time as the intervention school and were assessed using digital photography during meals at pre and post intervention. This study was approved by the University of Rhode Island Institutional Review Board.

Data collection
The data collection procedures were the same for all three aims. A fixed method was developed and included four cellular phone tripods. The tripods were 14 inches from the table, and a cellular phone was attached and placed at a 45-degree angle. All persons involved were trained on how to prepare the tripod and phone prior to data collection consisting of a written explanations as well as 1-2 verbal meetings prior to data collection, and an overview prior to data collection while at the school. Practice photographs were taken during training and feedback provided until assessors were proficient. The cellular phone cameras were all calibrated prior to data collection in order to insure consistency of the quality of photo and size.
For the intervention group and 2 non-participating classes, data collection occurred at baseline and after the 8th STC lesson, approximately 4 months after baseline.
The control group data were collected within a 2-week period of intervention group data.
At both baseline and follow-up, photos using the DFPM method were taken before eating ("pre-meal") and after eating ("post-meal").
On each day of data collection, SNAP-Ed researchers confirmed enrolled students in each classroom and assigned ID numbers to new students. Each student was given his or her unique lanyard prior to going to the cafeteria for lunch. The research photographers explained the data collection process to the students. The students were then read a script to explain the data collection process. The same protocol was conducted on subsequent days until all classrooms had data collected.
In the cafeteria in both schools, students are seated by classroom and each classroom was assigned two tables. There were two research photographers per table; each photographer photographed the student's tray across from him/her until the entire tables' photographs were collected. Researchers were given a diagram to help follow protocol as well as the layout of each cafeteria.
During data collection, lists of FVs available to the children were collected each day. The "garden cart" was photographed each day, and the foodservice staff provided a list of pre-plated FVs. In addition each item was referenced according to size (i.e. pieces, 1 whole, or converted to cups compared to a reference photo plated on a scale).

Intervention
STC was an 8-lesson PSE (policy, systems, and environment) intervention focused on fruit and vegetable intake with 5 th grade students in the PPSD. The curriculum included 8 lessons (see Table 1). STC encouraged the consumption of healthful fruits and vegetables. The curriculum did not include French fries or juice as part of a healthful diet.
This program allowed students to "have a say" in what they are being served in-school, and allowed them the opportunity to try fruits and vegetables in a different way.

Data Analysis
FVs served were identified. Items that were pre-plated followed standardized recipes with standardized portions, and items chosen from the "garden cart" were identified as whole pieces of fruit or vegetables, prepackaged portions in cups, or, for salad items, in relation to a weighed measure and converted to cups. For the purposes of this study, French fries, fruit juice, and tomato sauce were excluded 6 . The visual estimation for consumption is based on the protocol of the FoodWise Project (outlined in  Within the intervention school, the exploratory hypothesis looked at the amount of fruit and amount of vegetables consumed by the four 5 th grade classes with education compared to the classes that did not receive education. The same method to assess consumption for the primary hypothesis was used.

Statistics
Continuous data were assessed for normalicy; consumption data were not normally distributed (kurtosis > 2) thus non-parametric statistics were utilized for the primary and exploratory hypotheses. Mann-whitney U was used in order to analyze between group changes, and Wilcoxon signed-rank tests were conducted for within group change. In addition, Wilcoxon signed-rank tests were used to compare consumption between groups at baseline. For hypothesis two, variety, Pearson chi-square was used to analyze data. Demographic data at baseline were compared between groups using Pearson chi-square or student's t-tests.

RESULTS
There were no statistically significant differences between group by age, gender, or ethnicity (p>0.05) (see Table 3). The primary hypothesis of this study was to see if the treatment group with education increased the amount of cups of fruits and amount of cups of vegetables consumed at lunch more than the control group from baseline to follow-up. Data analysis protocol for defining consumption can be seen in Table 2. At baseline, both groups had a low intake of fruits and vegetables; the treatment group consumed an average of 0.26 cups of fruit and 0.03 cups (~1/2 tablespoon) of vegetables and the control school consumed an average of 0.11 cups of fruit and 0.04 cups of vegetables. As shown in Table 4, there was a significant difference between groups (p<0.01) for fruit consumption with the control school increasing fruit consumption more than the treatment school.
There was a significant decrease of 0.12 ± 0.46 cups within the treatment school (p=0.02); the control school increased fruit consumption by 0.12 ± 0.49 cups but this was not statistically significant (p=0.2). There were no significant differences between (p=0.13) or within groups (treatment school; p=0.41, control school; p=0.71) from baseline to follow-up for vegetable consumption. The treatment school had a nonsignificant increase in vegetable consumption by 0.01 cups while the control school had a non-significant decrease in vegetable consumption by 0.01 cups from baseline to followup.
The secondary hypothesis of this study was to see if the treatment group increased the variety of fruits and vegetables at lunch more than the control school from baseline to follow-up. Variety is defined as the number of different items on the tray for fruits and number of different items for vegetables. At baseline, there was a statistically significant difference in fruit variety between the treatment and control school (X 2 =33.29, p<0.001); a smaller proportion of treatment students had no fruit on their trays (40.7%) than control students (88.3%) (see Table 5). At follow-up, there were no statistically significant differences for fruit variety (X 2 = 0. 22 The treatment group decreased variety of fruits from baseline to follow-up, but there was no change in the control group ( Table 6). The control group increased variety of vegetables but there was no change in the treatment group. A substantial proportion of students 27 to 60% of students had no fruit on their trays at either time point and 48 to 88% of students had no vegetables on their trays at baseline and follow-up.
The exploratory group in the treatment school did not receive formal education, but were exposed to the recipe tasting and the posters around the school. Comparing students in the two classes that did not receive education to the four classes that did, there were no differences between groups for fruit (p=0.32) or for vegetables (p=0.37) as seen in Table 7. As reported above, the treatment group decreased fruit consumption with no change in vegetable consumption. There were no changes within the exploratory group, students consumed 0.27 cups of fruit at baseline, 0.30 cups of fruit at follow-up, and 0 cups of vegetables at both time points.
Due to the significant differences at baseline for both fruit and vegetable variety, further tests were conducted excluding pre-plated items from both the consumption and the variety variables. Tables 8 through 10 show the consumption and variety without preplated items, garden cart only. Within the treatment and control school, a majority of FV consumption was of the garden cart items. The treatment school consumed a total of 0.26 cups of fruit at baseline, 0.20 cups were garden cart items, and 0.06 cups were pre-plated item, and at follow-up, 0.14 cups of fruit were consumed, 0.09 from the garden cart, and 0.06 were pre-plated. Within the control school, all fruits at baseline and follow-up were consumed from the garden cart. Vegetables were minimally consumed at both time points, and a majority were from the garden cart as seen in Table 8. The treatment school consumed a total of 0.03 cups of vegetables at baseline, 0.03 cups were garden cart items, and 0 cups were pre-plated item, and at follow-up, 0.04 cups of vegetables were consumed, 0.02 from the garden cart, and 0.02 were pre-plated. All vegetable consumption from the control school were garden cart items. Variety decreased in groups when pre-plated items were excluded ( Excluding pre-plated items, there are no within group changes ( Table 10).
The decrease of fruit consumption by the treatment school can be seen in Figure   1. At both schools the only fruits served both at baseline and follow-up were oranges, apples, and pears. The greatest differences were for oranges. At baseline in the treatment school 11 students consumed all or a portion of an orange but only 4 students consumed oranges at follow-up. In the control school, 1 student consumed all or a portion of an orange at baseline and 5 at follow-up.
In addition to oranges, pears and apples (see Figure 1), there were other FVs provided at different time points. The treatment school did not provide bananas at baseline or follow-up. Although the control school did not provide bananas at baseline the school provided them at follow-up (n=13). The pre-plated items were also not consistent at baseline or follow-up within or between schools. This is illustrated by changes in pre-plated fruit items in the treatment school. Blueberries (n=6) and a frozen peach cup (n=6) were pre-plated at baseline accounting for a total of 4.4 cups of fruit consumed. At follow-up, pre-plated applesauce (n=8) and a strawberry cup (n=2) were provided and students consumed a total of 3.9 cups of these pre-plated fruits.

DISCUSSION
This study found that students consumed an average of 0.03 cups (~1/2 tablespoon) of vegetables during school lunch, and consumption appeared to be affected by options available during mealtime. Consumption was slightly better for fruit averaging 0.20 to 0.21 cups. To our knowledge, this is the first PSE intervention to be assessed using a DPFM method. Although the hypothesis that the PSE intervention would improve consumption was not supported, this study aids in understanding the school lunch environment and how it plays a role in eating behavior.
There was a significant change in fruit consumption between groups but no between-group change in vegetable consumption. However, the difference in fruit consumption ± 0.12 cups was small and, as described below, was likely due to changes in fruit offered at different time points. Mean consumption of fruit at baseline was higher at the treatment school than the control school, but this was reversed at follow-up. Perry  The decrease in fruit consumption may be partially explained by different items provided at the different time points. The types of fruits served at baseline and follow-up were not consistent. Children selected and consumed canned fruit in juice in the treatment school at baseline, however canned fruit in juice was not an option at that school at follow-up, which accounted for some of the decrease. The pre-plated item at follow-up for the treatment school was applesauce or strawberries, and students consumed 0.5 cups less of these pre-plated items than the canned fruit in juice offered at baseline. At the control school, bananas were not provided at baseline but were provided at follow-up.
Children appeared to chose and consume bananas frequently. A study within a Farm to School participating Wisconsin school found that canned fruits in juice were wasted less than whole fruits, while cooked vegetables were wasted more than raw 27 .
Although there were no statistically significant differences in variety from baseline to follow-up between groups, it is important to note that 61% of the total sample did not have a fruit on their tray at baseline, 69% did not have a vegetable at baseline, 68% did not have a fruit at follow-up, and 92% did not have a vegetable at follow-up.  The FVs consumed are assessed based on a percentage that is missing from the prephoto (i.e. pre-plated 4 oz. cup of berries, 25% left in post-meal photo = 1 oz. berries at post, 4oz-1oz = 3 oz. consumed).

Step 2
The graduate student researcher will then use the known pre-plated portions, or reference photos from the "garden cart" to analyze amount on the tray.
Step 3 The graduate student researcher will use the post-meal photo to analyze the amount consumed. This will be done by referencing the pre-meal photo and using the criteria from the FoodWise project to estimate the amount left on the tray.
Step 4 Consumption will be calculated by subtracting the amount in the pre-meal photo and post-meal photo.

Background of National School Lunch Program (NSLP)
The NSLP is responsible for feeding nutritious, well-balanced meals to more than 31 million children each day in the public school system 2,3 . Research reports that students who participate in the NSLP and the School Breakfast Program (SBP) may consume up to 47% of their daily nutrients from these items provided by the school 4  When foodservice workers were interviewed, they reported many barriers to following the new regulations such as increased labor cost, minimal understanding of the current ruling, and the lack of understanding about the new NSLP guidelines, its goals, and its need for participation/support at multiple levels on the part of parents, teachers, school staff, stakeholders, and foodservice workers 7 . Districts who had more support with NSLP tended to be districts reporting greater success with implementing the new NSLP guidelines 7 .

Fruit and Vegetable Consumption
FVs are high in fiber, water, and nutrients that are not energy dense. Consuming the recommended amounts of FVs is associated with lower total energy intake/density and increased satiety 8 . Epidemiological studies have shown a positive association between increased FV intake and decreased risk of obesity later in life 9 . Furthermore, children who consume diets rich in FVs are more likely to maintain these habits into adulthood and decrease the likelihood of excessive weight gain in adulthood 10 .
According to recent data, the U.S. population on average does not consume the recommended amounts for fruits and vegetables 11 . Children, 9-13 years old consume an average of 1.1 cups of fruit a day compared to the recommendation of 1.5 -2.0 cups 11 .
For vegetables, children ages 9 to 13 eat an average of 1.0 to 1.1 cups of vegetables a day compared to the recommendation of 1.5 to 3.0 cups per day 11 .

Variety of Fruits and Vegetables
Many epidemiological and cohort studies support the benefits of consuming adequate amounts of FVs. FVs contain nutrients essential for healthy body function and growth including Vitamin A, C, and K, potassium, magnesium, and phytonutrients, all which are currently under-consumed in the United States today [11][12][13][14] . Based on the 2015-2020 Dietary Guidelines, the U.S. population also does not meet the recommended intake for any subgroups of vegetables, indicating a lack of variety as seen in Table 1 11 .  As seen in Table 1, male and female children meet less than 50% of the weekly recommended average intake of any subgroup. About one-third of the intake of fruits comes from fruit juice, and the remaining two-thirds from whole fruits (which includes cut up, cooked, canned, frozen, and dried fruits) 11 . Potatoes and tomatoes are the most commonly consumed vegetables, with potatoes accounting for 21 percent and tomatoes 18 percent of vegetables consumed 11 . This is of concern due to potatoes being a starchy vegetable that is often consumed in its high fat, high sodium form, French fries 11 .
Specifically, lower income Americans consume more calorically dense foods than their higher income counterparts and are at a higher risk for disparities due to limited access to resources 15 . This increases the risk of disease due to poor nutritional quality.

Socioeconomic Status (SES) and FV Variety/Consumption
SES during childhood has been shown to be a strong predictor of adult health outcomes 16 . Two common indicators used to classify adolescent SES is parental education and parental income 17 . A longitudinal study with 896 adolescents found that high-income families reported greater accessibility to FV at home compared to their lowincome counterparts (Healthy Eating Index (HEI) FV subscale score 5.0 vs. 4.1, p<0.001) 18 . Results suggested a large reason low SES adolescents eat less FV than high SES adolescents is due to the decreased access in the home setting 18 . Likewise, this longitudinal study found that adolescents of higher education parents reported having greater preferences for FV, greater knowledge of FV recommendations, and stronger intentions to meet dietary FV guidelines 18 . Low SES is not only a national issue but also a local problem in Rhode Island.
Thirty-seven percent of the children living in Providence, RI live below the poverty line, 33% receive SNAP benefits (government assistance for purchasing food), and 88% are eligible for free or reduced meals 19 .
Access to FVs is limited for low-income families 20 . Based on the HEI 2005, lowincome families have lower component scores for total fruits and a statistically significant lower score for total vegetables (p<0.05) as compared to their higher income counterparts 20 . Specifically, low-income families have lower consumption of dark leafy greens and orange fruits and vegetables 11 . The Dietary Guidelines report that a majority of vegetable consumption by low-income populations consists of starchy vegetables such as; potatoes, corn, and peas 11 .

FV and Obesity
Childhood obesity has been linked to a high prevalence of metabolic syndrome in children, that rises with increased obesity 21 . Obesity is defined as being at or above the 95 th percentile on the BMI-for-age growth chart by the Centers for Disease Control (CDC) for children under 18 years old 22,23 . In Rhode Island, 28.3% of children age 10-17 are overweight or obese 24 . Overweight and obesity prevalence in Rhode Island vary by race and ethnicity. Hispanic children from core cities are more likely to be overweight or obese when compared to non-Hispanic white children not living in a core city [25][26][27] .
Increased consumption of fruits and non-starchy vegetables is inversely related with weight change 28 . Specifically, this study observed better weight management with the consumption of each extra daily serving of fruit, and an increase in total vegetable intake was also associated with prevention of weight gain 28 . However, an increased intake of starchy vegetables such as corn, peas, and potatoes was associated with weight gain 28 .
One study with children at risk of obesity 8-12 years old showed that an increase in FV intake may lead to a decrease of energy dense foods leading to weight management and decreasing the incidence of obesity 13,14 . Furthermore, children who have healthy dietary habits in adolescence, such as consuming FVs, have a higher likelihood of carrying these habits into adulthood and decreasing their risk of obesity in adulthood [8][9][10] .

School Eating Environment
There are many factors that can influence a child's meal patterns. Research conducted in elementary schools participating in the NSLP found that the classes that had recess before lunch had a higher consumption of FVs compared to students who had recess after lunch 4 . Another study found that children who received recess prior to lunch increased their fruit consumption by 5.1% 29  Research shows that there are certain foods that are typically accepted by children more than others such as bananas and French fries 4,[32][33][34] . With the current NSLP guidelines, reimbursable meals are required to have at least one serving of fruit or vegetables. One study showed that pre-plated vs salad bar items are still wasted at the same frequency as salad bar items at meals 32 . Salad bar items are those that are selfserved and self-selected by the child 32 . Their study found canned fruits in juice were wasted less than whole fruits, while cooked vegetables were wasted more than raw 32 .
Providing students with a variety of choices both hot and cold may to increase school FV consumption 3 .
One approach to create change in in the school lunchroom is CAN (Convenient, Attractive, Normal) approach that has been studied be Wansink and colleagues 35 . The strategy focuses on making food more Convenient in the lunchroom, this can be done by changing the location where healthier food is served or by pre-packaging items 35 . The Attractive component focuses on displaying the healthier foods in more appealing ways 35 .
The last part of this strategy is Normal 35 . Normal can be achieved by using the power of suggestion to make the healthy choice seem more socially acceptable 35 . This can be achieved by having a standardized location on each child's tray for a fruit or a vegetable.
The Smarter Lunchroom Movement is an initiative that was designed to help achieve the CAN approach 36,37 . The Smarter Lunchroom Movement changes are simple and low-cost that can easily transform the school environment to promote healthy choices 36 . Changes include displaying whole fruits in attractive bowls or baskets instead of hotel pans, creating descriptive names for FVs, and politely prompting students to select a fruit or a vegetable 37 . Studies assessing the Smarter Lunchroom Movement have found that through this approach, FV sales have increased by 20% in schools 38 .

Reporting Methods for Amount and Variety Consumed of FV
Dietary intake can be difficult to assess in children for a number of reasons. Some common instruments include food frequency questionnaires (FFQ), 24-hour recalls, surveys, dietary records, weighed measures, visual estimation, and digital photography.
This section will discuss strengths and limitations of different dietary intake tools commonly used to assess FV consumption in children.

Self-Reporting
The most common dietary intake method is self-reporting dietary intake. Selfreporting includes the use of comprehensive FFQ, 24-hour recalls, and brief surveys such as SNAP-Ed Fruit and Vegetable Checklist. In children, these methods can be difficult due to their limited cognitive ability, difficulty estimating portion sizes, reliance on caretaker to estimate portions, and limited attention span 1,39,40 .
Comprehensive FFQs have been used in many studies, but they are long and tedious for young subjects and may require assistance by an adult. These surveys includes items from all food groups in order to capture the habitual intake of the subjects, but this increases the subject burden for studies focusing on fruits and vegetable consumption.
The FFQ is a tool that obtains the average intake of items on a day-to-day basis, and therefore may not be as sensitive to change in daily consumption in cups as dietary recall based methods 41 . Nevertheless, FFQs can be self-completed, and are suitable for large scale studies with children [41][42][43] . In order to obtain an FFQ from a child, participation from caregivers are often necessary due to a child's limited long-term memory 1 .
Twenty four-hour recalls are considered the gold standard for self-reported intake.
The 24-hour recall has a low-respondent burden, and can be administered over the phone 42 . However, children ages 8 to 10 rely on the caregiver since a child may not be able to quantify food 1 . Limitations to this method include dependence on the subject's memory, bias in reporting "good/bad" foods, difficulty in estimating portion size, and a single recall is not a good measure of usual diet since it only captures one 24-hour period 1,42,44 .

SNAP-Ed uses the SNAP-Ed Fruit and Vegetable Checklist as a self-reporting
tool. This survey assesses the number of times fruits and vegetables were consumed in the previous day. This survey has been adapted from the 2-item FV screener 45 . There are six different response choices ranging from "0" to "5 or more times a day". The checklist includes five other items to assess types and quantities of FV consumed on the previous day. However, a limitation to this instrument is its lack of sensitivity to change. There were no changes from pre to post intervention in the pilot year of STC 45 . Specific problems with this survey include memory required to assess previous day's intake and confusion about classification of FVs 45 . These surveys are administered in class as a group in English, which requires the subject to be literate in English to follow along.
Lastly, students have difficulty accurately remembering what and how much they consumed on the previous day. As there are many limits to self-reporting, more objective measures are warranted.

Digital Photography
Digital photography provides a quick and unobtrusive method to estimate food intake in cafeteria settings [46][47][48] [50][51][52] . Another study by Williamson et al. (2003) also showed high correlation of digital photography to weighed and visual estimation of portion sizes 48 .  55 . This study did not report consumption of FV, but found that modification to the school cafeteria is feasible and has the ability to positively influence children's food consumption 55 . The use of DPFM found statistically significant decreases in total fat selected at lunch (Wisemind: -60 ± 10.6; p=0.03; LA Health: -78 ± 10.4; p<0.0001) and
This decrease in total fat intake may be due to the increased availability of healthier more nutrient dense items such as FVs 55 . The findings of this study support the hypothesis that modifying the lunch environment can positively impact healthier choices, and supports the decision to change the NSLP guidelines to the current 2012 standards 55 .
Hubbard et al. evaluated whether a Smarter Lunchroom intervention could be adapted to increase the selection of FVs for students with intellectual and developmental disabilities 56 . This quasi-experimental study used baseline and follow-up intervention DPFM data 56 . Data collection occurred 5 days at baseline and 5 days at follow-up. Days were matched based on menu items in order to ensure items offered were identical at both time points 56 . There are a total of 644 trays analyzed and subjects, n=43, ranged from 11 to 21 years old with disabilities attending a residential school in Massachusetts 56 . The 3-month intervention occurred from March to June 2012, and capitalized on environmental changes such as moving fruits to the front of the service line and providing items in separate, attractive bowls 56 . This study found that after the 3-month environmental changes, the daily consumption of total fruits increased by a mean of 0.18 cups (p=0.008), canned fruits increased by 0.13 cups (p=0.02), and whole fresh fruits increased by 0.05 cups (p=0.38) 56 . Selection of raw vegetables significantly decreased by 0.16 cups (p=0.001), but intake of total vegetables increased by 0.07 cups (p=0.14) 56 .
Plate waste significantly decreased for vegetables (p=0.03) 56 . What this tells us is that the students selected vegetables they were more likely to eat at follow-up, consumed more, and wasted less 56 57 . This study used percent increment in order to interpret data and did not report consumption or selection in cups 57 . The percentage of students selecting fruits significantly increased from 54% of students to 66% (p<0.05) and consumption of fruits remained high pre and post policy implementation 57 . Furthermore, this study found that fruit consumption increased by 9% for each additional fruit that was offered at meal-time 57 . Post-policy implementation, vegetable selection dropped from 68% to 52% (p<0.05), however, students ate 20% (p<0.05) more vegetables post-policy implementation, lowering vegetable waste 57 . This study also looked into which fruits and vegetables were most popular 57 . Based on baseline and follow-up data, 88% of fruit cups, 78% of bananas, 70% of oranges, 56% of pears, and 48% of apples served were consumed 57 . For vegetables, 72% of potatoes (excluding fried), 65% of corn, 46% of beans, 42% of salad, and 38% of broccoli served was consumed 57 . This study indicates that the NSLP updated guidelines have led to more nutritious meals and increased fruit selection without increasing plate waste of FVs 57 .
DPFM is a reliable, valid tool in order to look deeper into the breakdown of school meals 48,58,59 . Researchers have been able to use DPFM in a wide variety of settings, including the school lunchroom in order to quantify consumption of school lunch. The findings from these studies suggest the DPFM is an appropriate tool to use amongst low-income 5 th graders, and is reliable and effective in the school lunch environment to assess fruit and vegetable consumption and variety 48,58,59 .

Conclusion
Accurately measuring dietary intake in children is important due to the low intake of FVs and rising incidence of overweight and obesity 1,11,19,45,53 . Children who do not consume the daily recommendations of FV are more likely to consume excess quantities of energy dense foods that can lead to overweight and obesity 8,13,60 . This review found that DPFM is a valid and reliable dietary intake tool in school aged children 46,48,50,58,61 .
DPFM is accurate within the school lunchroom setting, decreases participate burden, and is effective in measuring FVs 4,59,62 . Data shows that children consume an average of 0.4 cups of FVs at lunch, and consume a greater amount of fruits than vegetables 3,4,46,51 .
Process data from the pilot year of STC indicated that students had an increased knowledge in nutrition and perceived making changes but outcome data failed to find changes in FV consumption. This suggests that PSE interventions may need to use objective measures such as DPFM, but no studies have used this objective method for PSE outcome evaluation 45,53 .