ARTICLE IN PRESS

Research Brief
Impact of a Pilot School-Based Nutrition Intervention on
Fruit and Vegetable Waste at School Lunches
Shreela Sharma, PhD, RD, LD1; Allison Marshall, MSSW, MPH2; Joanne Chow, MPH, RD3;
Nalini Ranjit, PhD2; Gregory Bounds, MPH3; Katherine Hearne, MPH, RDN, LD3;
Nan Cramer, RD, LD, SNS4; Amanda Oceguera, MS, RD, LD4; Alicia Farhat, MPH, RD, LD5;
Christine Markham, PhD3

ABSTRACT
Objective: To determine the preliminary impact of the Brighter Bites nutrition intervention on decreasing
fruit and vegetable (F&V) waste at school lunches among fourth- and fifth-grade children.
Method: This was a nonrandomized pre−post-controlled study in Houston and Dallas, TX. Two schools
received the Brighter Bites intervention (n = 76), and 1 comparison school (n = 39), during the 2017-2018
school year. Brighter Bites is a 16-week school-based nutrition intervention providing weekly distribution of
fresh F&V plus nutrition education. Main outcome measures were direct observation and weights to mea-
sure the number of F&V dishes selected at school lunches, amount of F&V wasted (gm), and related nutri-
ent waste (4 time points/child). Mixed-effects linear regression analysis was used to determine change in
F&V selection and waste over time.
Results: There was a significant decrease over time in proportion of F&V selected among those in the com-
parison school, but not the intervention schools (P < .001). Compared with children in the comparison
group, those receiving Brighter Bites showed a significant decrease in the amount of F&V wasted at each meal
(P < .001) and per item (P < .05) at the end of both 8 and 16 weeks of intervention. There were significant
decreases in waste of energy (kcal); dietary fiber (gm); vitamins B1, B3, and B6 (mg); total folate (mg); and B12
(mg) among those receiving Brighter Bites (P < .05).
Conclusions and Implications: Although absolute food or nutrient changes were small even when
significant, programs such as Brighter Bites may contribute to a healthy intake. Future studies are warranted
that include a larger sample size with a stringent, cluster-randomized control trial design and consideration
for other covariates.
Key Words: child dietary intake, fruit and vegetable consumption, nutrient waste, plate waste, school
lunch (J Nutr Educ Behav. 2019;000:1−9.)
Accepted August 8, 2019.

INTRODUCTION
1 Despite widespread national and inter-
Department of Epidemiology, Environmental Health, and Genetics, Michael and Susan Dell
national efforts to increase fruit and
Center for Healthy Living, University of Texas School of Public Health, Houston, TX
2 vegetable (F&V) consumption, chil-
Michael and Susan Dell Center for Healthy Living, University of Texas School of Public
dren of all ages consume less F&V
Health in Austin, Austin, TX
3 than recommended.1,2 Children from
Department of Health Promotion Behavioral Sciences, University of Texas School of Public
lower income households are more
Health, Houston, TX
4 likely to consume inadequate amounts
Houston Independent School District Food Service Support Facility, Houston, TX
5 of F&V compared with their higher-
Brighter Bites, Dallas, TX
income counterparts,3 which exacer-
This study was supported by Feeding Texas and the Texas Health and Human Services Com-
bates health disparities through the
mission. Feeding Texas and the Texas Health and Human Services Commission had no role
life course. Sufficient F&V consump-
in the design, analysis, or writing of this article.
tion is critical for healthy physical and
Conflict of Interest Disclosure: The authors’ conflict of interest disclosures can be found online
psychosocial development and func-
with this article on www.jneb.org.
tioning, especially during periods of
Address for correspondence: Shreela Sharma, PhD, RD, LD, Department of Epidemiology,
rapid growth in childhood and ado-
Environmental Health, and Genetics, Michael and Susan Dell Center for Healthy Living,
lescence.1 In children, higher F&V
University of Texas School of Public Health, 1200 Pressler, RAS E603, Houston, TX 77030;
intake is associated with reduced risks
E-mail: Shreela.V.Sharma@uth.tmc.edu
for chronic diseases and risk factors
Ó 2019 Society for Nutrition Education and Behavior. Published by Elsevier Inc. All rights
such as obesity, diabetes, hyperten-
reserved.
sion, high cholesterol, and nutrient
https://doi.org/10.1016/j.jneb.2019.08.002

Journal of Nutrition Education and Behavior  Volume 000, Number 000, 2019 1
 ARTICLE IN PRESS
2 Sharma et al Journal of Nutrition Education and Behavior  Volume 000, Number 000, 2019

deficiency.4 Thus, promoting ade- school year before the beginning of and 1 in district B) received Brighter
quate intake of F&V among children, Brighter Bites programming (baseline, Bites programming for the first time
especially those from lower-income wave 1), the midpoint of the program in the 2017−2018 school year; con-
families, remains important.3,4 (end of 8 weeks in the fall, wave 2), currently, the comparison school (in
Over 30 million children rely on before the beginning of the spring district B) was not receiving Brighter
the National School Lunch Program programming (wave 3), and the end Bites and had never participated in
(NSLP) for nutrient and energy intake of the 16 weeks of the Brighter Bites it. Inclusion criteria for students
during the school day, 20 million program (end of spring, wave 4). All included (1) being enrolled in Brighter
meals of which are free lunches to fourth- and fifth-grade children in Bites in the 2017−2018 school year
low-income children.5 Students are participating intervention schools for the first time (for intervention
required to select at least 1 fruit or veg- were exposed to Brighter Bites, but schools), (2) participating in the
etable per lunch meal under current only those who consented to partici- NSLP at the school, and (3) being
regulations.6 Unfortunately, previous pate in the study were measured. This enrolled in fourth or fifth grade in
plate waste studies demonstrated that study was approved by the University the 2017−2018 school year.
children often do not consume the of Texas Health Science Center Com- A total of 115 students were
F&V selected from the lunch lines.7−11 mittee for the Protection of Human recruited and provided consent at
Plate waste during school lunch results Subjects. the 3 schools (intervention school 1:
in nutrient loss and unnecessary costs n = 44; intervention school 2: n = 32;
for schools and the NSLP.8,10,12 Brighter Description of Brighter Bites comparison school: n = 39). Two
Bites is an evidence-based school Intervention classrooms per grade were targeted
health promotion program that com- for recruitment. Recruitment strate-
bines access to fresh F&V with nutri- Brighter Bites is an evidence-based pro- gies consisted of presentation to
tion education in school and at home gram13 currently being disseminated school leadership and a parent invi-
to increase preference and intake of nationwide through the Brighter tation letter sent home for participa-
F&V among children. The current Bites 501c3 nonprofit organization. A tion in the study. Of those in the
study tested the preliminary impact of detailed description of the program is participating grades, recruitment
Brighter Bites on (1) increasing the available elsewhere;13 briefly, Brighter rates were 19.1%, 17.5%, and 39% in
number of F&V dishes selected by chil- Bites is a 16-week school-based nutri- intervention schools 1 and 2 and the
dren at school lunch, (2) decreasing tion program grounded in Social Cog- comparison school, respectively.
the amount of F&V wasted during nitive Theory constructs.14 It consists Written informed consent was
school lunch by weight and percent- of 3 main components: (1) weekly dis- obtained from parents, and verbal
age (proportion of F&V wasted), and tributions of 50 servings of fresh assent obtained from children. At
(3) decreasing F&V-related nutrient donated F&V sourced from local food each time point, all children were
waste per lunch meal among elemen- banks sent home with parents; (2) measured for 5 days (the entire
tary schoolchildren in Houston and nutrition education, which includes school week). Children who had
Dallas, TX. the evidence-based Coordinated <3 days of school lunch measure-
Approach to Child Health program in ments at baseline were excluded
schools,15,16 and parent education from subsequent measurements.
METHODS through bilingual nutrition hand- Finally, parent−child attendance to
Plate Waste Study Design books and recipe cards; and (3) weekly Brighter Bites distributions was
recipe demonstrations at produce obtained from Brighter Bites.
A nonrandomized pre−post-controlled pickup time. Results of Brighter Bites
trial design was employed for this evaluation demonstrated significant Data Collection Measures
study. Trained data collectors mea- improvements in the intake of F&V
sured participating students’ school among participating children and A total of 15 data collectors, who
lunch in the cafeteria at 4 1-week parents and improvements in the were university graduate students,
time points per child (Monday home nutrition environment.13 were trained by study investigators
through Friday) during the 2017−2018 to conduct plate waste measures for
school year. A convenience sample of Recruitment and Participants the study. Before lunch, cafeteria
3 schools participated in the study: staff provided data collectors with
2 schools receiving the Brighter Bites A convenience sample of 3 schools samples of all available choices of
program and 1 comparison school from 2 public school districts, 1 in F&V options for that day. Across all
(not participating in Brighter Bites). At Dallas, TX (district A) and another in 3 schools, 30 minutes was allocated
each time point, data were collected Houston, TX (district B), was recruited for the lunch period. For each day of
every day of the week (Monday to participate in the study. The measurement, before the lunch period,
through Friday). The Brighter Bites pro- schools were selected based on com- trained data collectors recorded the
gram is implemented over 16 weeks parable racial and ethnic composition type and number of F&V items
in the school year: 8 weeks in the and the percentage of the student offered, as indicated in production
fall semester and 8 in the spring, population eligible for the free or records provided by the school food
respectively. The 4 time points of data reduced-priced lunch program. Two and nutrition service staff, obtained 3
collection included the start of the elementary schools (1 in district A sample portions of each selection,
 ARTICLE IN PRESS
Journal of Nutrition Education and Behavior  Volume 000, Number 000, 2019 Sharma et al 3

and averaged the weights for each serving of fruit or vegetable was Furthermore, overall attendance at
offering to generate standard weights. unconsumed, it was assumed that its Brighter Bites weekly produce distribu-
Across all schools, children chose weight was equal to that of the stan- tions was 64% (77% attendance rate
F&V items from individual containers dardized sample plate. Plate waste in intervention school 1 and 50% in
preserved by cafeteria staff. Before the percentage was calculated using the intervention school 2).
lunch period, on each day of measure- equation: ([1 weight of food Table 1 shows the baseline
ment, data collectors also provided remaining  weight of standardized descriptive statistics for F&V plate
tags with identifier numbers to partic- sample plate] £ 100). Nutrient loss waste measurements, overall and
ipating children in their respective estimations for each specific nutri- stratified by schools and school dis-
classrooms. Children wore these tags ent were calculated using plate waste tricts. Overall, on average, 4.7 § 1.51
throughout the lunch period, and as percentage and were adjusted to the unique varieties of F&V were avail-
they came through the lunch line standardized sample serving size. able to children at the school lunches
with their trays after having selected Mixed-effects linear regression mod- each day. Of these, each day, chil-
foods, their trays were labeled with els were used with group £ time dren chose an average of 1.2 § 0.6
same identifier numbers. During interaction terms to test for varieties and tried 0.9 § 0.7 different
lunch, data collectors recorded the between-group changes over time. F&V. Stratified analysis shows signifi-
type and number of F&V items Children were tracked individually; cant differences in the variety of F&V
selected by each participating child. analysis was at the child level. Plate offered, selected, and tried among
After lunch, data collectors calculated waste measures were clustered the 3 schools (P < .001).
the amount consumed by the student within the child, and change in F&V At baseline, children wasted an
by weighing the total amount of each plate waste measures and related average 59.5% of the F&V that they
F&V food item wasted on participat- nutrients wasted at the child level chose for school lunch. At baseline,
ing students’ plates using Schuler were assessed over time.18,19 although there were significant differ-
Scientific digital scales (SSP-1502, Although a cluster analysis adjusting ences among schools for the average
Englewood, CO). The F&V were for school as a random effect would amount of F&V selected (P < .001) and
weighed to the nearest 0.01 g. Inter- have been the preferred analysis the amount of F&V wasted (P < .001),
rater reliability was high (100% technique, given the small sample there were no significant differences
agreement) and an average weight size at the school level, this was not among schools in the average propor-
difference of 0.012 g was found performed; instead, school was tion of F&V wasted (P = .91). Table 1
between raters (r = .99). Prior field- adjusted for as a covariate in the anal- also shows the baseline descriptive sta-
based studies showed that mobile ysis. All analyses were performed tistics for the percent F&V wasted strat-
electronic data systems have greater using Stata software (version 14.2 (Sta- ified by various foods. The F&V most
accuracy compared with secondary taCorp, College Station, TX, 2016). wasted were legumes and deep yellow
manual data entry.17 Hence, for this The level of significance was defined and dark green leafy vegetables,
study, a mobile data collection as P < .05. Goodness of fit tests were whereas the least wasted foods were
application in AppSheet (version used to determine model fit. baked par-fried potatoes; 9% to 45% of
10.0, AppSheet, Inc, Seattle, WA, foods were 100% wasted (ie, uneaten).
2017) was developed and used in RESULTS Table 2 shows the results of
real time by trained data collectors mixed-effects linear regression analy-
to record all F&V selections and the All 3 participating schools had ≥90% sis. Whereas there was no change
amount of F&V waste data electroni- of children participating in the free among those in the intervention
cally. and reduced-price lunch program, group, there was a significant decrease
which indicated that a majority of in the proportion of F&V dishes
Data Analysis the children were from low-income selected by children in the comparison
families. At the school level, on aver- group at all waves of measurement (P
Descriptive statistics including means, age, 77.3% and 13.7% of children in < .001). Results also showed that, mea-
SDs, and frequency distributions the intervention schools and 81.5% sured against those in the comparison
were computed at baseline. One-way and 12.6% of those in the compari- school, at the end of the intervention
ANOVA was used to assess for base- son school were Hispanic or African (wave 4), children in the intervention
line differences in F&V waste among American, respectively. The overall schools had a significant decrease in
the 3 schools. All plate waste data retention rate of participating chil- the amount of F&V wasted at each
and recipes for the F&V provided by dren in the study was 79% (data not meal (b = 32.06; 95% confidence
school districts were entered into shown). After the initial measure- interval, 48.9 to 15.2; P < .001) and
Nutrition Data System for Research ment week (wave 1, baseline), 24 stu- per item (b = 28.03; 95% confidence
software (Nutrition Data System for dents were removed who did not interval, 39.7 to 18.4; P < .001).
Research, University of Minnesota, meet the initial requirement of ≤3 Total amount of energy regarding kcal,
Minneapolis, MN, 2016). For foods school lunch measurements owing carbohydrate (gm), and protein (gm)
such as whole fruit (eg, whole to absences or home lunches. Reten- wasted also decreased significantly
bananas, whole uncut apples) or tion rates at waves 2, 3, and 4 were among those participating in Brighter
unopened packaged fruits or vegeta- 84.1% for the intervention school 1 Bites measured against those in the
bles (eg, fruit cups), if the entire and 81.3% at intervention school 2. comparison school at the end of 8
 4
Table 1. Baseline Data on F&V Selection and Waste at School Lunches, Brighter Bites Plate Waste Study, 2017−2018

Sharma et al
Average F&Va Items Average F&Va Items Percentage wasted 100%
Variable Available Selected Average F&Va Items Tried ([B / A] £ 100) Wasted
Overall (for all schools) 4.7 § 1.5 1.2 § 0.6 0.9 § 0.7
Intervention school 1 6.1 § 1.1 1.0 § 0.6 0.8 § 0.7
Intervention school 2 4.4 § 0.8 1.3 § 0.5 1.0 § 0.6
Control school 3.2 § 0.4 1.4 § 0.5 1.0 § 0.7
Pa <.001b <.001b .01b
Average F&V Amount Average F&V Amount Average Amount Consumed
Selected (g) (A) Wasted (g) (B) (g) (A − B)
Overall (for all schools) 103.2 § 87.7 82.5 § 68.7 55.1 § 54.7 59.5 § 35.5 15.0
Intervention school 1 143. 9 § 105.1 101.0 § 84.0 64.3 § 65.9 59.5 § 33.7 15.0
Intervention school 2 91.9 § 64.5 71.5 § 53.9 47.7 § 44.0 60.5 § 37.1 16.0

ARTICLE IN PRESS
Control school 72.8 § 67.9 67.9 § 50.4 49.7 § 44.9 58.5 § 36.5 14.0
Pa <.001b <.001b .01b .91 .90

Journal of Nutrition Education and Behavior  Volume 000, Number 000, 2019
Stratified by type of fruit and vegetable
Citrus fruit 88.4 § 9.7 52.0 § 32.8 37.8 § 33.9 58.5 20
Fruit (excluding citrus) 99.4 § 27.2 56.4 § 40.4 43.5 § 35.3 55.2 27
Avocados and similar 90.7 § 0.0 83.1 § 6.1 7.7 § 6.1 91.6 14
(Includes avocado in
guacamole)
Dark green vegetables (eg, 63.0 § 16.9 43.2 § 25.9 19.5 § 16.3 64.8 18
broccoli, collards, romaine,
spinach)
Deep yellow vegetables (eg, 77.7 § 11.4 60.8 § 17.1 16.0 § 12.8 78.8 34
carrots, winter squash,
sweet potatoes, pumpkin)
Tomato 74.0 § 16.1 50.5 § 28.6 23.2 § 20.6 66.2 12
White potatoes 203.8 § 57.0 110.6 § 73.1 96.0 § 85.2 58.0 13
Baked par-fried potatoes 70.1 § 9.00 32.2 § 28.8 39.4 § 24.8 43.3 9
Other starchy vegetablesd 137.9 § 81.3 87.9 § 68.3 71.4 § 76.4 58.7 23
Legumes (cooked dried 102.2 § 10.4 78.8 § 21.8 19.7 § 25.3 81.2 45
beans)
Other vegetablesd 53.1 § 27.6 35.4 § 33.1 19.0 § 23.4 59.9 33
F&V indicates fruits and vegetables.
a
Mean differences were analyzed by 1-way ANOVA; bSignificance at P < .05; cIncludes vegetables in salads, soups, stews, stir-fry, and similar mixed dishes (eg, corn,
immature lima beans, lentil sprouts, peas); dIncludes vegetables in salads, soups, stews, stir-fry, and similar mixed dishes (eg, beets, cabbage, mung bean sprouts, sum-
mer squash).
Notes: Data are shown as n § SD. Study population was fourth- to fifth-grade students enrolled across 3 schools. Two schools receiving the Brighter Bites intervention and
1 was the usual care comparison school (intervention school 1: n = 44 children; intervention school 2: n = 32 children; comparison school: n = 39 children).
 ARTICLE IN PRESS
Journal of Nutrition Education and Behavior  Volume 000, Number 000, 2019 Sharma et al 5

Table 2. Changes in F&V Selection, Waste and Related Nutrients Waste, Brighter Bites Plate Waste Study, 2017
−2018

Wave 1 Wave 2 Wave 3 Wave 4

Observations, n (Baseline) (End of Fall) (Start of Spring) (End of Spring)
Intervention group, n 295 308 263 310
Control group, n 130 121 96 102
Proportion of F&V sides
selected per meala
Intervention group, mean 0.2 (0.1) 0.2 (0.2) 0.2 (0.2) 0.2 (0.2)
(SD)
Control group, mean (SD) 0.4 (0.2) 0.3 (0.1) 0.3 (0.2) 0.4 (0.1)
Net changes in proportion of Reference 0.1 (0.1 to 0.1) 0.1 (0.1 to 0.2) 0.1 (0.0−0.1)
F&V side dishes selec-
ted,b b (95% CI)
P <.001c <.001c .001c
Amount of F&V wasted per
meal, g
Intervention group, mean 88.9 (74.5) 79.7 (62.2) 78.2 (57.6) 63.0 (51.2)
(SD)
Control group, mean (SD) 67.9 (50.4) 68.1 (53.7) 55.8 (43.3) 74.6 (54.1)
Net changes in weight of Reference 8.1 (−24.3 to 8.2) 1.4 (−15.9 to 18.6) 32.1 ( 48.9, 15.2)
food wasted,b b (95% CI)
P .33 .88 <.001c
Amount wasted per F&V
item, g
Intervention group, mean 60.6 (51.7) 50.7 (40.4) 50.9 (45.7) 43.0 (40.6)
(SD)
Control group, mean (SD) 46.0 (38.3) 48.8 (37.5) 40.3 (34.3) 58.1 (42.3)
Net changes in weight of Reference 11.5 ( 21.5 to 1.50) 3.3 (−14.0 to 7.4) 28.0 ( 39.7 to 18.4)
food wasted,b b (95% CI)
P .02c .55 <.001c
Total energy wasted, kcal
Intervention group, mean 58.9 (61.3) 46.0 (45.8) 49.3 (48.5) 37.5 (38.2)
(SD)
Control group, mean (SD) 46.6 (42.3) 48.7 (43.0) 32.6 (41.0) 45.6 (45.4)
Net changes in energy Reference 14.1 ( 27.4 to 0.9) 4.6 (−9.5 to 18.6) 19.6 ( 33.3 to 5.9)
wasted,b b (95% CI)
P .04c .52 .005c
Total carbohydrates wasted, g
Intervention group, mean 12.6 (11.2) 10.2 (9.7) 11.0 (9.7) 8.7 (8.0)
(SD)
Control group, mean (SD) 10.0 (8.8) 10.0 (8.3) 7.4 (9.1) 10.6 (10.4)
Net changes in total carbo- Reference 2.28 (−5.0 to 0.4) 1.1 ( 1.7 to 4.0) 4.3 ( 7.0 to 1.5)
hydrates wasted,b b (95%
CI)
P .09 .44 .002
Total protein wasted, g
Intervention group, mean 1.0 (1.5) 0.8 (1.4) 0.7 (1.0) 0.5 (0.7)
(SD)
Control group, mean (SD) 1.0 (1.6) 1.2 (1.9) 0.9 (1.8) 1.2 (1.8)
Net changes in protein Reference 0.4 ( 0.8 to 0.0) 0.2 ( 0.6 to 0.2) 0.6 ( 1.0 to 0.2)
wasted,b b (95% CI)
P .05c .31 .003c
Total fat wasted, g
Intervention group, mean 1.0 (2.2) 0.6 (1.1) 0.7 (1.8) 0.4 (1.4)
(SD)
Control group, mean (SD) 0.7 (1.1) 0.8 (1.3) 0.2 (0.3) 0.2 (0.5)
Reference 0.6 ( 1.0 to 0.2) 0.1( 0.3 to 0.6) 0.1( 0.6 to 0.3)
(continued)
 ARTICLE IN PRESS
6 Sharma et al Journal of Nutrition Education and Behavior  Volume 000, Number 000, 2019

Table 2. (Continued)

Wave 1 Wave 2 Wave 3 Wave 4

Observations, n (Baseline) (End of Fall) (Start of Spring) (End of Spring)
Net changes in total fat
wasted,b b (95% CI)
P .006c 0.63 .52
Total dietary fiber wasted, g
Intervention group, mean 1.8 (2.0) 1.5 (1.8) 1.7 (1.8) 1.3 (1.4)
(SD)
Control group, mean (SD) 1.5 (1.5) 1.5 (1.5) 1.5 (2.4) 1.8 (2.4)
Net changes in fiber Reference 0.2 ( 0.8 to 0.3) 0.1 ( 0.6 to 0.5) 0.7 ( 1.2 to 0.2)
wasted,b b (95% CI)
P .35 .75 .009c
Vitamin B1 (thiamine) wasted,
mg
Intervention group, mean 0.045 (0.05) 0.042 (0.05) 0.035 (0.04) 0.025 (0.03)
(SD)
Control group, mean (SD) 0.041 (0.03) 0.043 (0.04) 0.034 (0.05) 0.048 (0.07)
Net changes in vitamin B1 Reference 0.003 ( 0.02 to 0.010) 0.002 ( 0.02 to 0.01) 0.03 ( 0.04 to 0.01)
(thiamine) wasted,b b
(95% CI)
P .63 .77 .001c
Vitamin B2 (riboflavine)
wasted, mg
Intervention group, mean 0.036 (0.03) 0.026 (0.02) 0.026 (0.02) 0.022 (0.02)
(SD)
Control group, mean (SD) 0.034 (0.04) 0.038 (0.04) 0.028 (0.03) 0.029 (0.03)
Net changes in vitamin B2 Reference 0.01 ( 0.02 to 0.005) 0.003 ( 0.01 to 0.005) 0.008 ( 0.02 to .001)
(riboflavine) wasted,b b
(95% CI)
P .001c .41 .06
Vitamin B3 (niacin) wasted, mg
Intervention group, mean 0.47 (0.52) 0.35 (0.35) 0.37 (0.40) 0.27 (0.32)
(SD)
Control group, mean (SD) 0.37 (0.35) 0.35 (0.33) 0.25 (0.30) 0.43 (0.40)
Net changes in vitamin B3 Reference 0.09 ( 0.19 to 0.02) 0.03 ( 0.09 to 0.14) 0.26 ( 0.37 to 0.15)
(niacin) wasted,b b (95%
CI)
P .11 .70 .001c
Vitamin B6 (pyridoxine, pyri-
doxyl, and pyridoxamine)
wasted, mg
Intervention group, mean 0.07 (0.08) 0.06 (0.06) 0.06 (0.07) 0.05 (0.06)
(SD)
Control group, mean (SD) 0.09 (0.12) 0.08 (0.10) 0.07 (0.12) 0.10 (0.15)
Net changes in vitamin B6 Reference 0.0007 ( 0.03 to 0.02) 0.009 ( 0.02 to 0.04) 0.03 ( 0.06 to 0.005)
wasted,b b (95% CI)
P .95 .57 .02c
Total folate wasted, mg
Intervention group, mean 13.3 (21.6) 14.9 (27.1) 12.3 (17.2) 9.5 (14.1)
(SD)
Control group, mean (SD) 12.9 (17.2) 15.1 (23.1) 17.1 (30.4) 16.3 (30.8)
Net changes in total folate Reference 0.33 (-6.7,6.0) 5.1 (−11.8 to 1.7) 6.9 ( 13.5 to 0.4)
wasted,b b (95% CI)
P .93 .14 .04c
(continued)
 ARTICLE IN PRESS
Journal of Nutrition Education and Behavior  Volume 000, Number 000, 2019 Sharma et al 7

Table 2. (Continued)

Wave 1 Wave 2 Wave 3 Wave 4

Observations, n (Baseline) (End of Fall) (Start of Spring) (End of Spring)
Vitamin B12 (cobalamin)
wasted, mg
Intervention group, mean 0.007 (0.03) 0.006 (0.02) 0.004 (0.02) 0.0002 (0.002)
(SD)
Control group, mean (SD) 0.0001 (0.0005) 0.0001 (0.0005) 0.00004 (0.0003) 0.00009 (0.0005)
Net changes in vitamin B12 Reference 0.0009 (−0.006 to 0.004) 0.002 (−0.008 to 0.003) 0.006 ( 0.01 to 0.001)
(cobalamin) wasted,b b
(95% CI)
P .71 .34 .01c
Vitamin C (ascorbic acid)
wasted,d mg
Intervention group, mean 5.6 (5.9) 6.6 (9.2) 6.8 (9.1) 6.1 (10.8)
(SD)
Control group, mean (SD) 6.9 (7.3) 6.1 (6.2) 3.0 (4.4) 4.1 (4.8)
Net changes in vitamin C Reference 1.7 ( 0.6 to 4.0) 4.9 (2.5 to 7.4) 3.0 (0.7 to 5.4)
wasted,b b (95% CI)
P .15 <.001c .01c
Vitamin E (total a-tocopherol)
wasted, mg
Intervention group, mean 0.33 (0.32) 0.33 (0.39) 0.38 (0.41) 0.26 (0.32)
(SD)
Control group, mean (SD) 0.25 (0.32) 0.23 (0.31) 0.09 (0.12) 0.18 (0.19)
Net changes in vitamin E Reference 0.02 ( 0.07 to 0.12) 0.20 (0.10 to 0.29) 0.008 ( 0.10 to 0.09)
(total a-tocopherol)
wasted,b b (95% CI)
P .60 <.01c .83
Vitamin K (phylloquinone)
wasted, mg
Intervention group, mean 7.3 (13.3) 7.1 (12.8) 6.9 (9.5) 5.6 (11.5)
(SD)
Control group, mean (SD) 8.1 (15.9) 7.4 (22.5) 3.8 (6.8) 4.2 (7.0)
Net changes in vitamin K Reference 0.8 (−2.9 to 4.5) 4.1 (0.2 to 8.1) 2.5 (−1.3 to 6.3)
(phylloquinone) wasted,b
b (95% CI)
P .67 .04c .20
Calcium wasted, mg
Intervention group, mean 14.3 (21.6) 13.7 (14.3) 13.7 (14.7) 8.9 (10.2)
(SD)
Control group, mean (SD) 16.1 (19.5) 12.9 (13.4) 9.9 (14.9) 12.0 (15.4)
Net changes in calcium Reference 2.5 (−1.9 to 6.8) 4.6 ( 0.06 to 9.2) 1.7 (−6.2 to 2.8)
wasted,b b (95% CI)
P .27 .05 .45
Iron wasted, mg
Intervention group, mean 0.30 (0.33) 0.29 (0.39) 0.25 (0.27) 0.19 (0.25)
(SD)
Control group, mean (SD) 0.40 (0.47) 0.39 (0.50) 0.27 (0.51) 0.38 (0.54)
Net changes in iron Reference 0.01 ( 0.10 to 0.12) 0.08 ( 0.03 to 0.20) 0.07 ( 0.18 to 0.03)
wasted,b b (95% CI)
P .85 .15 .18
CI indicates confidence interval; F&V, fruits and vegetables.
a
Proportion of F&V side dishes selected per meal is F&V side dishes selected per meal divided by F&V side dishes available
per meal; bCoefficient from interaction term wave £ intervention in mixed-effect linear regression models with plate waste
measures clustered within child, adjusted for school as a covariate in the analysis; cSignificance at P < .05; dSummation of
weight of food wasted per meal, assuming weight of food wasted equals weight of sample food item if it was totally unconsumed.
Notes: Study population was fourth- to fifth-grade students enrolled across 3 schools. Two schools receiving the Brighter Bites
intervention and 1 was the usual care comparison school (final sample size for pre−post analysis = 91 students).
 ARTICLE IN PRESS
8 Sharma et al Journal of Nutrition Education and Behavior  Volume 000, Number 000, 2019

weeks (P < .01) and the end of the fiber, vitamins, and folate are the retention rate with regard to plate
intervention (P < .01). With regard to building blocks of life and are critical waste measurement was lower in the
micronutrients, there was a significant for preventing chronic disease25−27; comparison school; more children
decrease in waste in the amount of die- strategies to promote consumption brought lunches from home, com-
tary fiber (gm) at wave 4; vitamins B1, and reduce waste of these key pared with the 2 intervention schools.
B3, and B6 (mg), total folate (mg), and nutrients should be employed. This differential attrition could have
vitamin B12 (mg) (P < .05). Conversely, Recent systematic reviews of the lit- biased the overall findings, likely away
there were greater decreases in vita- erature also indicated that school- from the null. Despite multiple
mins C and K (mg) wasted among chil- based interventions that involve reminders regarding the purpose of the
dren in the comparison school parents and improve the home envi- study, children bringing home lunches
measured against those participating ronment can support child energy- continued to be a challenge. This
in Brighter Bites (P < .05). balance behaviors and health.28 needs to be considered in longitudinal
Baseline data from the current plate waste assessments. Also, the
DISCUSSION study found that regardless of the attendance rate for Brighter Bites distri-
number of F&V selections offered, butions was lower in intervention
Overall, the results of this pilot plate children selected a small number of school 2 compared with intervention
waste study demonstrated a signifi- F&V at school lunches and wasted a school 1, which could have attenuated
cant, although small, decrease from substantial proportion of those the findings of the current study.
baseline to the end of the intervention selected. These results concur with Finally, a limitation is that changes in
in the amount of F&V wasted among those of Gase et al11: in an effort to F&V waste and nutrients lost might
children participating in Brighter Bites, characterize student receptivity to have resulted from changes in con-
compared with those not participating new menu offerings at school lunch sumption of other components of the
in the program. Moreover, the in Los Angeles, CA, those authors meal and not from the intervention
decrease in the amount of F&V reported that a significant propor- itself.
wasted at school lunches was signifi- tion of children did not consume
cant at the end of the 16-week inter- F&V offered at school lunch because
vention, but not at the midpoint they either did not select F&V or IMPLICATIONS FOR
(8 weeks); this potentially indicates threw away all of the F&V without RESEARCH AND PRACTICE
the need for longer-term exposure to taking a bite.
F&V to promote behavior change. School-based nutrition interventions
This study adds to the current Limitations may improve F&V intake among
body of literature on measuring F&V children during school lunch. Strate-
waste among children participating Limitations of the study include a gies such as those used by Brighter
in the NSLP20 by using objective small sample size with 1 comparison Bites to engage parents in a F&V co-
measures and measuring the type and 2 intervention schools that was op, teaching them how to use the
and amount of F&V wasted. These not powered for outcomes or cluster produce provided, and how to
results concur with those of other effects. This was a pilot study and was improve the home nutrition environ-
studies that assessed the impact of not statistically powered to detect ment, may hold promise in improv-
school-based interventions and pol- changes in the outcomes measured. ing healthy dietary intake among
icy change on plate waste at school Furthermore, sociodemographic data children from low-income popula-
meals and found that modifying the were not collected at the individual tions. However, future studies includ-
availability and accessibility of F&V child or family level, which could be ing a larger sample size with a more
in schools improved consumption potential covariates in the analysis. stringent, fully powered, 3-level,
and decreased waste at school Also, because students were clustered cluster-randomized, controlled trial
meals.21−23 The results of this study within schools, a 3-level multilevel design are needed. Future studies
also demonstrated significant, model adjusting for within- and might also assess the whole meal (vs
although small, reductions in waste between-school variance would have only F&V), consider other covariates
of nutrients such as dietary fiber and been the preferred method of analysis. such as school recess time, and assess
vitamins among children participat- However, because of the small sample dose−response relations between fre-
ing in Brighter Bites, measured size at the school level, a multilevel quency of exposure to F&V and food
against those in the comparison model adjusting for school-level vari- waste. By collaborating with commu-
group. Nutrient waste because of ance was not performed, which might nity partners and nutrition educators
food waste is a major issue in the US. have resulted in an underestimation of in implementing school-based nutri-
A study by Spiker et al24 using the SEs of regression coefficients and a tion programs that include provision
national data reported a significant subsequent potential overstatement of of healthy foods such as F&V plus
amount of F&V and related nutrient statistically significant findings. More- nutrition education, school districts
waste at the retail and consumer lev- over, factors such as the timing of may effectively reduce F&V waste in
els in the US amounting to dietary school recess (before or after lunch) school lunches and improve the die-
fiber waste equivalent to 23% of the may vary among schools and influence tary intakes of children, especially
recommended dietary allowance for findings. However, these data were not those from lower socioeconomic
Americans. Nutrients such as dietary collected in the current study. The populations.
 ARTICLE IN PRESS
Journal of Nutrition Education and Behavior  Volume 000, Number 000, 2019 Sharma et al 9

ACKNOWLEDGMENTS consumption, and waste. Am J Prev randomized trials: a review of recent

Med. 2014;46:388–394. methodological developments. Am J
The authors acknowledge Feeding 9. Buzby JC, Guthrie JF. Plate waste in Public Health. 2004;94:423–432.
Texas, Brighter Bites, and the Michael school nutrition programs. Report to 19. Shadish WR, Cook TD, Campbell DT.
and Susan Dell Center for Healthy Congress. Washington, DC: US Experimental and Quasi-Experimental
Living for their support of the study. Department of Agriculture, Economic Designs for Generalized Causal Inference.
They also acknowledge the Houston Research Service; 2002. Available at: 2nd ed. Boston, MA: Houghton Mif-
Independent School District and Dal- https://www.ers.usda.gov/webdocs/ flin; 2002.
las Independent School District for publications/43131/31216_efan02009. 20. Byker Shanks C, Banna J, Serrano EL.
their support of the study. pdf?v=41423. Accessed August 28, Food waste in the National School
2019. Lunch Program 1978−2015: A system-
REFERENCES 10. Byker CJ, Farris AR, Marcenelle M, atic review. J Acad Nutr Diet. 2017;
Davis GC, Serrano EL. Food waste in 117:1792–1807.
1. US Department of Health and Human a school nutrition program after 21. Just DR, Wansink B, Hanks AS. Chefs
Services, US Department of Agricul- implementation of new lunch pro- move to schools. A pilot examination
ture. Dietary Guidelines for Americans, gram guidelines. J Nutr Educ Behav. of how chef-created dishes can
2005. 6th ed. Washington, DC: US 2014;46:406–411. increase school lunch participation and
Government Printing Office; 2005. 11. Gase LN, McCarthy WJ, Robles B, fruit and vegetable intake. Appetite.
2. WHO. Global strategy on diet, physi- Kuo T. Student receptivity to new 2014;83:242–247.
cal activity and health. Updated 2018. school meal offerings: assessing fruit 22. Just D, Price J. Default options, incentives
http://www.who.int/dietphysicalac- and vegetable waste among middle and food choices: evidence from elemen-
tivity/childhood_consequences/en/. school students in the Los Angeles Uni- tary-school children. Public Health Nutr.
Accessed September 25, 2018. fied School District. Prev Med. 2014;67 2013;16:2281–2288.
3. Zarnowiecki DM, Dollman J, Par- (suppl 1):S28–S33. 23. Wansink B, Just DR, Hanks AS, Smith
letta N. Associations between predic- 12. Huang Z, Gao R, Bawuerjiang N, LE. Pre-sliced fruit in school cafeterias:
tors of children’s dietary intake and Zhang Y, Huang X, Cai M. Food and children’s selection and intake. Am J Prev
socioeconomic position: A system- nutrients intake in the school lunch Med. 2013;44:477–480.
atic review of the literature. Obes program among school children in 24. Spiker ML, Hiza HA, Siddiqi SM,
Rev. 2014;15:375–391. Shanghai, China. Nutrients. 2017;9:582. Neff RA. Wasted food, wasted
4. Ogden CL, Lamb MM, Carroll MD, 13. Sharma SV, Markham C, Chow J, nutrients: nutrient loss from wasted
Flegal KM. Obesity and socioeco- Ranjit N, Pomeroy M, Raber M. food in the United States and com-
nomic status in adults: United States Evaluating a school-based fruit and parison to gaps in dietary intake. J
1988−1994 and 2005−2008. NCHS vegetable co-op in low-income chil- Acad Nutr Diet. 2017;117:1031–1040.
Data Brief. 2010;50:1–8. dren: A quasi-experimental study. Prev e22.
5. School Nutrition Association. School Med. 2016;91:8–17. 25. Dahl WJ, Stewart ML. Position of the
meal trends & stats. Updated 2017. 14. Bandura A. Social Foundations of Thought academy of nutrition and dietetics:
http://www.schoolnutr ition.org/ and Action: A Social Cognitive Theory. health implications of dietary fiber. J
AboutSchoolMeals/School-Meal- Englewood Cliffs, NJ: Prentice-Hall; Acad Nutr Diet. 2015;115:1861–1870.
TrendsStats. Accessed January 20, 2019. 1986. 26. O’Neil CE, Nicklas TA, Fulgoni VL.
6. US Department of Agriculture, Food 15. Hoelscher DM, Springer AE, Ranjit Consumption of apples is associated
and Nutrition Service. SP 45-2013: N, et al. Reductions in child obesity with a better diet quality and reduced
Updated Offer versus Serve Guidance among disadvantaged school children risk of obesity in children: National
for the National School Lunch Pro- with community involvement: the Health and Nutrition Examination
gram and School Breakfast Program in Travis County CATCH trial. Obesity. Survey (NHANES) 2003−2010. Nutr
School Year 2013-2014. Updated 2010;18:S36–S44. J. 2015;14:48.
2013. Available at: https://www.fns. 16. Luepker R, Perry C, McKinlay S, et al. 27. Wu L, Sun D, He Y. Fruit and vegeta-
usda.gov/sp-45-2013-updated-offer- Outcomes of a field trial to improve bles consumption and incident hyper-
versus-serve-guidance-national-school- children’s dietary patterns and physical tension: dose−response meta-analysis
lunch-program-and-school-breakfast. activity. The Child and Adolescent of prospective cohort studies. J Hum
Accessed June 27, 2019. Trial for cardiovascular Health. Hypertens. 2016;30:573–580.
7. Cohen JF, Richardson S, Austin SB, CATCH collaborative group. JAMA. 28. Verjans-Janssen SRB, van de Kolk I,
Economos CD, Rimm EB. School lunch 1996;275:768–776. Van Kann DHH, Kremers SPJ, Gerards
waste among middle school students: 17. Cole E, Pisano ED, Clary GJ, et al. A SMPL. Effectiveness of school-based
nutrients consumed and costs. Am J Prev comparative study of mobile electronic physical activity and nutrition interven-
Med. 2013;44:114–121. data entry systems for clinical trials tions with direct parental involvement
8. Cohen JF, Richardson S, Parker E, Cat- data collection. Int J Med Inform. on children’s BMI and energy balance-
alano PJ, Rimm EB. Impact of the new 2006;75:722–729. related behaviors - A systematic review.
U.S. Department of Agriculture school 18. Murray DM, Varnell SP, Blitstein JL. PLOS ONE. 2018;13:e0204560.
meal standards on food selection, Design and analysis of group-
 ARTICLE IN PRESS
9.e1 Sharma et al Journal of Nutrition Education and Behavior  Volume 000, Number 000, 2019

CONFLICT OF INTEREST
S. Sharma serves on the Board of
Directors for Brighter Bites nonprofit
organization. This is an unpaid board
position. The rest of the authors
declare that they have no relevant
conflicts of interest.