Plate waste data were collected from three elementary schools (kindergarten through fifth grade) in one school district participating in the USDA’s school meal program in central Texas. Plate waste collections were conducted in two phases. Phase 1 was conducted in April and May 2012, pre-implementation of the new nutrition standards, while Phase 2 was conducted in October and November 2012, post-implementation of the new nutrition standards. This timing allowed us to evaluate plate waste from school lunches pre- and post-implementation of new school meal standards. This design corresponds to a natural experiment in that the same schools were involved prior to and after the implementation of the new standards for the NSLP. In this light, this study was observational in nature without any additional interventions.

By design, the research did not involve personal identifiers regarding the participants and was approved as exempt from committee review by the Texas A&M University Institutional Review Board.

All lunch periods for the sampled schools were scheduled by grade (kindergarten through 5). The same menu and portion sizes were served during all lunches in all three elementary schools. In all the schools, 30 minutes was allowed for lunch and recess was scheduled after the lunch. In order to meet the requirements for the new meal standard, the school district changed the recipes of the menu items and served a greater variety of vegetables; serving sizes were not changed, however. Schools used an eight-week menu cycle both pre and post-im- plementation.

For each data collection day, five to ten servings of each sampled entrée and vegetable were served on “test trays” by school nutrition staff. The “test trays” were used to measure pre-weights for each entrée and each vegetable for which plate waste was collected. On average, two or more options for entrées and vegetables were provided on a given day. Before each lunch period, separate trash bins were prepared for each entrée and vegetable and lined with pre-weighed trash bags. Students finished their meals and brought their trays to the research team members waiting at the designated stations. Uneaten entrées and vegetables were separated into the corresponding trash bin and the remaining contents of the tray were discarded in the cafeteria trash can for disposal. This process was repeated for each lunch period. Aggregated waste per menu item was weighed and recorded per grade at each school.

Altogether, plate waste data from 8430 students were collected―corresponding to 4145 students pre-imple- mentation and 4285 students post-implementation of school meal standards. A total of 144 observations of entrée/vegetable pairings repeated by grade and school (27 distinct pairings) were collected pre-implementation and 305 observations of entrée/vegetable pairings repeated by grade and school (56 distinct pairings) were collected post-implementation. Analyses of the respective entrée/vegetable pairings were conducted to determine plate waste and therefore, acceptability for particular entrée/vegetable pair subsequently.

Our data collection procedures allowed for plate waste to be examined for entrée and vegetable pairings as well as for each entrée and each vegetable separately. Plate waste (in grams) for each entrée and vegetable by grade and school was divided by the number of students choosing those items.

The ratio provided the plate waste per student (in grams), where j = 1 if main entrée and j = 2 if vegetable and k = 1, ∙∙∙, n is type of main entrée or type of vegetable.

Finally, the percentage of plate waste was calculated by dividing this ratio by the pre-weight (in grams) of the entrée or vegetable item, also measured in grams.

By knowing the number of participating children, plate waste was able to be expressed as a proportion or percentage.

The SAS statistical analysis system (version 9.4, 2013, SAS Institute, Inc, Cary, NC) was used to create a data file with food pairings, which served as variables in conducting descriptive and statistical analysis. Mean values and standard deviations of plate waste were calculated for entrée/vegetable pairings and separately for entrées and for vegetables, for pre- and post-implementation of new school meal standards. The popularity of entrées and of vegetables was calculated as a mean number of students selecting the particular item on the data collection days. T-tests were used to determine the differences in means in popularity of entrées and vegetables, and differences in means of plate waste for entrée/vegetable pairings. The level of significance chosen was at α = 0.05.

Table 1 provides characteristics of three elementary schools in our sample. These elementary schools were diverse based on race/ethnicity of the student population and percentage of students eligible to receive free/re- duced-price school meals. Nearly all students in school A were eligible for free/reduced price lunches. In contrast, one-third or less of the students in school C qualified for free or reduced-price lunches. Post-implementa- tion, a greater percentage of students in school B were eligible for free or reduced-price lunches compared to the pre-implementation period. The percentage of students eligible to receive free/reduced price meals from three elementary schools in our sample was about 66% pre-implementation and 68% post-implementation of nutrition standards. For the state of Texas, this percentage was 66% in 2012 and for the U.S. 50% in 2012 [16] . As reported in the School Nutrition Dietary Assessment Study-IV (SNDA-IV) report 64% of reimbursable lunches served in elementary school during 2009-2010 school year were served for free or at reduced prices. For our sample, we found that on average 74% and 76% of reimbursable lunches served were for free or at reduced prices, pre- and post-implementation, respectively. The percent of lunches sampled on the data collection days varied from 34% to 79%. Despite this variability, the data are representative of the population.

“Popularity” was defined as the mean number of students in our sample selecting the item on the data collection days. Popularity of the main entrées and vegetables, pre- and post-implementation of the new school meal standards, are reported in Table 2 and Table 3. On average, two or more different entrées and different vegetables selections were offered on a given day. Greater varieties of entrées and vegetables were served post-implemen- tation of the revised standards compared with pre-implementation. The five most popular vegetables were starchy vegetables, the majority of which were potatoes in various processed forms (mashed potatoes with gravy, oven baked French fries, tater tots, and potato wedges), both pre- and post-implementation. Steamed broccoli was the least popular vegetable both pre- and post-implementation. Chicken nuggets were the most popular entrée for both periods. Burgers (plain and with cheese) were also popular pre- and post-implementation. Pre-implementation, deli sliders were the least popular entrée, whereas the sunbutter sandwich was the least popular entrée post-implementation.

We hypothesized that there was a relationship between consumption of certain entrées and vegetables that would lead to less plate waste. Our hypothesis is that if students eat less of an entrée, then they might eat more of the accompanying vegetable(s). Similarly, if students ate more of the entrée (i.e., waste less of it), then they might have eaten less of the accompanying vegetable(s).

Values are means ± standard deviations.

Values are means ± standard deviations.

Figure 1 illustrates the relationship between plate waste of burgers and vegetables. As shown, burgers were a popular entrée pre- and post-implementation. Greater plate waste from burgers was observed when this popular entrée was paired with a popular vegetable like tater tots or oven-baked French fried potatoes―each having relatively low plate waste. In contrast, the burger was wasted less when paired with a less popular vegetable like sweet potato fries. The relationship between green beans and entrée plate is reported in Figure 2. Green beans were wasted the most when paired with chicken nuggets―the most popular and least wasted entrée, pre- and post-implementation. However, when green beans were paired with steak fingers―a less popular entrée―plate waste of the vegetable was reduced.

Total plate waste for all entrée and vegetable pairings in our sample pre- and post- implementation of new school meal standards, respectively, are provided in Table 4 and Table 5.

Overall, the pairing of deli sliders with corn on the cob resulted in the highest combined plate waste (62.5%), and the pairing of corn dog on a stick with tater tots resulted in the lowest combined plate waste (21%) (Table 4). Plate waste for entrées was the lowest for chicken nuggets (7.6%) paired with green beans and highest for deli sliders (44.3%) served with corn on the cob. Serving deli sliders as a main entrée and corn on the cob as a

^aValues are mean % ± standard deviation; ^bRanked from smallest to largest total percent wasted; ^cHere n in the number of observations for each entrée/vegetable pairing.

^aValues are mean % ± standard deviations; ^bRanked from smallest to largest total percent wasted; ^cHere n in the number of observations for each entrée/vegetable pairing.

vegetable side dish resulted in a notable amount of plate waste for both entrée (44.3%) and vegetable (80.8%). Although chicken nuggets had the lowest percent plate waste when served with green beans, the percentage plate waste for green beans was among the highest (69%) and was higher than the plate waste for chicken nuggets and mashed potatoes and gravy.

Four out of five pairings that had the lowest overall plate waste involved white potato products. Entrées and vegetables pairings with the least overall plate waste involved the most popular entrées and the most popular vegetables for elementary school students.

Five most wasted pairings of entrée and vegetables were corn on the cob (80.8%) served with deli sliders (44.3%), hamburger or cheeseburger (37.4) with ranch style beans (78.5%), pizza (27.1%) with corn on the cob (81.3%), pizza (28.8%) with green peas (78.1%) and barbeque on a bun (39.1%) with corn on the cob (63.0%).

Similar to our findings for the pre-implementation period of nutrition standards, four out of five entrée and vegetable pairings that had the lowest overall plate waste involved potato products (tater tots, mashed potato with gravy, French fries) as side dishes for the period post-implementation of the new school meal standards (Table 5). Broccoli florets served with a hot dog on a bun were among the five least wasted entrée and vegetable pairings.

Post-implementation of the new standards, the mean combined percent plate waste for entrée and vegetable pairings increased slightly from 40.4% to 43.5% but this increase was not statistically significant (Table 4 and Table 5). Mean percent plate waste for vegetables, both pre- and post-implementation of the new school meal standards, were over 50% (52.1% for pre-implementation and 57.7% for post-implementation). Entrées were wasted less compared to vegetables. Mean entrée plate waste increased by less than 1% post- compared to pre-implementation (28.6% for pre- and 29.3% for post-implementation).

The results reported in Table 4 and Table 5 raise a set of questions regarding the energy and nutrients intake from the most and the least wasted entrées and vegetables and their pairings. The nutrient profile for every entrée and vegetable sampled was provided by school district. The mean energy and nutrients intakes per student from entrée and vegetable pairing were calculated using total energy and nutrients provided per serving and the percentage of plate waste. Table 6 and Table 7 report mean energy and nutrients (dietary fiber, sodium, iron, vitamin A, vitamin C, calcium, protein, fat, and saturated fat) intake per student and mean energy and nutrients available to students from entrée/vegetable pairings from school meals for the ten most and least wasted entrée and vegetable pairings, pre- and post-implementation of the new standards, respectively². For the ten most wasted pairings, the mean combined waste from entrée/vegetable pairings was 51% pre-implementation and 60% post-implementation versus for the ten least wasted pairings, the mean combined waste from entrée/vegetable pairings was 31% both pre-implementation and post-implementation of the new school meal standards.

With respect to the ten least wasted pairings, more calories were offered post-implementation (377 kcal/student) compared to pre-implementation (247 kcal/student) and consequently more calories were consumed post- (260 kcal/student) compared to pre-implementation (240 kcal/student). Less sodium was offered post- (800 mg/student) compared to pre-implementation (889 mg/student). No notable differences in the amount of saturated fat offered were observed between pre- and post-implementation. In general, more dietary fiber, iron, vitamin A and C, calcium and protein were available to students from entrée/vegetable pairings served post-implementation compared to pre-implementation of the new school meal standards. Specifically, more dietary fiber were available to students during the post- compared to pre-implementation (5 g/student versus 3.2 g/student) and noticeably more vitamin C was available to students from food pairings post- compared to pre- implementation (16.5 mg/student versus 7.3 mg/student).

Similar patterns of nutrients available to students from the entrée/vegetable pairings were observed for the ten most wasted food pairings. Noticeable increases in the amounts of vitamin A and C as well as calcium available to students were observed for the post- compared to pre-implementation period.

After the changes in school meal standards, the minimum and maximum energy required for lunch for kindergarten through fifth grade are between 550 - 650 kcal. For our sample of the ten least and most wasted pairings, mean energy available from entrée/vegetable pairings was 365 kcal/student pre-implementation and 382

^aMean energy or nutrient intake from the combined entrée and vegetable pairing; ^bMean energy or nutrient available from the combined entrée and vegetable pairing.

^aMean energy or nutrient intake from the combined entrée and vegetable pairing; ^bMean energy or nutrient available from the combined entrée and vegetable pairing.

kcal/student post-implementation. The target requirement for sodium for lunch is ≤1230 mg that started in SY 2014-2015. For our sample, the mean sodium available from entrée/vegetable pairings from school meals were 938 mg/student pre- and 895 mg/student post-implementation. The new standards require that saturated fat should be less than 10% total calories. Saturated fat available to students from entrée/vegetable pairings in our sample was less than 1% of total calories available to students in both the pre- and post-implementation periods of the new school meal standards.