UNDER EMBARGO UNTIL NOVEMBER 7, 2014, 12:01 AM ET

Podcast available online at www.jneb.org Research Brief Nutritional Comparison of Packed and School in Pre-Kindergarten and Kindergarten Children Following the Implementation of the 2012–2013 National School Program Standards Alisha R. Farris, MS, RD1; Sarah Misyak, PhD1; Kiyah J. Duffey, PhD1; George C. Davis, PhD2; Kathy Hosig, PhD, RD3; Naama Atzaba-Poria, PhD4; Mary M. McFerren, EdD1; Elena L. Serrano, PhD1

ABSTRACT Objective: Approximately 40% of children bring a packed lunch to school. Little is known about the quality of these lunches. This study examined the nutritional quality of packed lunches compared with school lunches for pre-kindergarten and kindergarten children after the implementation of 2012–2013 National School Lunch Program standards. Methods: The researchers collected observational data for packed and school lunches from 3 schools in rural Virginia for 5 consecutive school days and analyzed them for macro and micro nutrients. Results: Of the 1,314 observations collected; 42.8% were packed lunches (n ¼ 562) and 57.2% were school lunches (n ¼ 752). Energy, fat, saturated fat, sugar, vitamin C, and iron were significantly higher whereas pro- tein, sodium, fiber, vitamin A, and calcium were significantly lower for packed lunches than school lunches. Conclusions and Implications: Packed lunches were of less nutritional quality than school lunches. Additional research is needed to explore factors related to choosing packed over school lunches. Key Words: packed lunch, school lunch, children, NSLP, nutrition (J Nutr Educ Behav. 2014;46:621-626.) Accepted July 30, 2014.

INTRODUCTION with decreases from 31.8 million such as 24-hour recalls and 1- to 3-day average daily in 2011 to 31.6 observations and were conducted Over 50 million children attend million in 2012,5 which translates with a variety of ages ranging from elementary and secondary public into more children consuming kindergarten (K) to 12th grade.2,4,8-13 schools in the US each day.1 These chil- packed lunches. Whereas the NSLP is No studies reported data over a full dren spend a substantial portion of mandated to meet nutrition standards school week (5 days) with pre-K classes their waking hours and consume a sub- aligned with the 2010 Healthy attending public school and compared stantial portion of their daily calories at and Hunger-Free Kids Act,6,7 packed with the new NSLP standards. The school.2,3 For approximately 60% of lunches are not required to meet purpose of this study was to examine children, these calories are derived nutrition standards. the nutritional quality of packed from the National School Lunch To date, relatively few studies have lunches compared with school lunches Program (NSLP); the remaining 40% been published on the nutritional served to pre-K and K children are from packed lunches.4 During quality of packed lunches.2,4,8-13 attending public school over 5 consec- the past few years, there have been Existing studies used a variety of data utive school days after implementation significant shifts in NSLP participation, collection methods and timeframes of the 2012 NSLP standards.

METHODS 1Department of Human Nutrition, , and Exercise, Virginia Tech, Blacksburg, VA Recruitment of Schools 2Department of Agricultural and Applied Economics, Virginia Tech, Blacksburg, VA 3Department of Population Health Sciences, Center of Public Health Practice and Research, The researchers contacted 8 elementary Virginia Tech, Blacksburg, VA schools in a rural area in Virginia 4Department of Psychology, Ben-Gurion University, Negev, Israel via e-mail and telephone. Of the 8 Address for correspondence: Alisha R. Farris, MS, RD, Department of Human Nutrition, schools contacted, 3 (37.5%) agreed Foods, and Exercise, College of Agriculture and Life Sciences, 338 Wallace Hall, Mail Code to participate. The County Public 0430, Virginia Tech, Blacksburg, VA 24061; Phone: (540) 818-0173; Fax: (540) 231-3916; School Research Office and each indi- E-mail: [email protected] vidual school administrator granted Ó2014 SOCIETY FOR NUTRITION EDUCATION AND BEHAVIOR permission to enter each elementary http://dx.doi.org/10.1016/j.jneb.2014.07.007 school. The Institutional Review Board Journal of Nutrition Education and Behavior Volume 46, Number 6, 2014 621 622 Farris et al Journal of Nutrition Education and Behavior Volume 46, Number 6, 2014 for Virginia Tech approved the study for a total of 24 items. Food and tests were carried out with Bonferroni and did not require child or parental beverage items selected represented adjustment of P < .004. Tukey's consent. commonly found items in elementary Honest Significant Difference test was packed lunches. School lunches were used to detect differences between Setting and Participants not chosen because of their uniformity groups. and ease of recording. Accuracy was The 3 elementary schools spanned 2 determined by dividing the number RESULTS counties: Montgomery, with a popula- of items accurately recorded by the tion of 95,194 composed of white total number of items (item identi- Nutritional Profile of Packed (87.9%), black (4.1%), and Hispanic/ fication ¼ 93.8%; portion estima- and School Lunches Latino (2.9%) individuals; and Giles, tion ¼ 92.1%). Interobserver reliability with a population of 16,928 composed was assessed with average pairwise A total of 1,314 lunches were observed, of white (97%), black (1.5%), and His- 42.8% of which were from packed percent agreement tests (JMP, version ¼ panic/Latino (1.3%) individuals.14 lunches (n 562) and 57.2% from 11, SAS Institute, Inc, Cary, NC, 2013). ¼ The 3 schools had free and reduced Observers demonstrated 90.7% agree- NSLP (n 752). Energy, carbohydrate, price school lunch participation rates fi fat, saturated fat, sugar, vitamin C, ment for item identi cation and fi of 33.3%, 46.6%, and 52.7%, respec- 86.8% agreement for portion estima- and iron were signi cantly higher tively. No identifying information for tion. This was consistent with previous for packed lunches compared with any student was collected. The authors school lunches whereas protein, so- research showing that trained observers fi selected pre-K and K students because dium, ber, vitamin A, and calcium with prior nutrition knowledge can fi young ages represent a malleable time accurately and reliably assess packed were signi cantly lower for packed to promote food acceptance.15-18 lunch contents and intake by direct lunches compared with school lunches observation in an elementary school (Table 1). The nutrient availability for Observational Protocol setting.19 children in both packed and school lunch groups almost entirely met the nutrition standards of the NSLP except National School Lunch Program me- Data Analysis nus need to meet nutritional require- that school lunches being 38 calories ments over the of 1 week; The school director for the below energy and 0.4 mg below iron therefore, observational data were region or the manager for the recommendations whereas packed collected in each elementary school elementary schools provided nutrient lunches were 1.5 g higher than fat 7 for 5 consecutive school days. All analyses for school lunches per food and 0.3 g higher than saturated fat rec- pre-K and K students in each school item. Packed lunch items were ommendations. were served NSLP meals consisting of analyzed (Nutritionist Pro Diet Anal- similar portion sizes. An observational ysis software, version 5.1, Axxya Sys- Nutritional Profile of Packed checklist reflecting the compo- tems, Stafford, TX, 2009) based on Lunches by School nents of the day's specific and serving sizes recorded at the time of commonly consumed items from observation. If an item was unavailable Differences among schools for the fi packed lunches were used to record in the Nutritionist Pro database, the nutritional pro le of packed lunches data on the presence of all food and nutrition facts label per product brand for protein, fat, saturated fat, sodium, served as part of the NSLP and type were used for analysis. US vitamin A, and calcium were not sta- fi and brought from home, with write- Department of Agriculture reference tistically signi cant (Table 2). Energy fi in sections for additional foods items were recorded when available. was signi cantly higher for packed that may have not been part of Students who brought a packed lunch lunches of students at the moderate the observational checklist. Each and purchased milk from the cafeteria (46.6%) free and reduced eligibility researcher was assigned a specific were classified as a packed lunch obser- level compared with both other grouping of students to observe for a vation. Students who participated in schools. Packed lunches of students lunch period, approximately 10 stu- the NSLP but also brought food from at the school with the low (33.3%) dents/researcher. home (n ¼ 5; 0.007% of total observa- free and reduced eligibility level fi tions) were excluded. were signi cantly lower for carbohy- Observer Training and Descriptive statistics were used drate, sugar, and vitamin C and significantly higher for iron and fiber Reliability to describe the nutritional quality of packed vs school lunches. Sha- than packed lunches at both other Undergraduate and graduate nutrition piro-Wilk test was used to determine schools. students were recruited and trained as whether the data were parametric. observers in direct observation to assess The researchers carried out compari- Food Categories in Packed and lunch contents, specifically visual item sons of mean quantities of macronutri- School Lunches identification and portion size estima- ents and micronutrients (calories, tion. The training was conducted by a protein, fat, saturated fat, carbohy- Packed lunches were less likely to doctoral-level registered dietitian. For drates, sugar, fiber, vitamin A, vitamin contain fruits (54% vs 67%), vegeta- checklist reliability testing, the re- C, calcium, and iron) using the Mann– bles (17% vs 61%), juice with no sugar searchers conducted observations of 5 Whitney–Wilcoxon test. Multiple (10% vs 22%), and milk (20% vs 96%) premeasured sample packed lunches, comparison and post hoc comparison than NSLP meals. They also contained Journal of Nutrition Education and Behavior Volume 46, Number 6, 2014 Farris et al 623

Table 1. Comparisons of Nutrients Between School and Packed Lunches Among Pre-Kindergarten and Kindergarten Students in 3 Schools

School Meals (n ¼ 752) Packed Meals (n ¼ 562) P (Bonferroni Adjusted US Department of Significance ¼ .004)a Agriculture Nutrient Interquartile Interquartile Goals: K Through Fifth Mean Median Range Mean Median Range P Value** Grade and Pre-K Energy, kcal 512 520 106 608 588 240 < .001 550–650 kcal (K through fifth grade) 517 kcal (pre-K)b Carbohydrate, g 66 67 25 88 85 38 < .001 – Protein, g 26 27 8 18 17 9 < .001 9 g (K through fifth grade)c 7 g (pre-K)c Fat, g 13 13 5 21 20 14 < .001 < 19.5 g (K through fifth grade)d < 17 g (pre-K)d Saturated fat, g 3.6 3 4 6.8 6.5 6 < .001 < 6.5 ge < 5.7 g (pre-K)e Sodium, mg 1,021 1,038 364 883 818 482 < .001 < 1,230 mgf Fiber, g 7.2 7 4.5 4.8 4.5 4 < .001 – Sugar, g 35 35 12 45 43 37 < .001 – Vitamin A, IU 3,856 1,350 4,800 1,375 502 715 < .001 1,000 IUc Vitamin C, mg 24 17 22 43 29 59 < .001 15 mgc Calcium, mg 503 460 209 305 297 285 < .001 267 mgc Iron, mg 2.9 2.5 1.5 3.4 3 2 < .001 3.3 mgc K indicates kindergarten. aBased on Mann-Whitney-Wilcoxon test; bMinimum requirement for pre-K children served meals as part of the National School Lunch Program; cOne third of the Recommended Daily Allowance for age/grade group; dTotal fat not to exceed 30% of total calories over a school week (calculation based on mean calorie range); eSaturated fat not to exceed 10% of total calories over a school week (calculation based on meal calorie range); fSodium guidelines not required until 2014–2015 school year. more savory such as chips and nutritional profile for packed lunches intake among young children because crackers (57% vs 5%) and sugar- was of less nutritional quality than of the higher prevalence of sweetened beverages (40% vs 0%). NSLP lunches.2,10 School lunches items, savory items, and Some of the most commonly brought were more likely to meet nutrition sugar-sweetened beverages. These food items in packed lunches were pea- standards than packed lunches, findings mirror previous studies in nut butter (n ¼ 155), especially for fat and saturated fat, which NSLP participants consumed single-serving chips (n ¼ 145), single- which exceeded recommended levels fewer sugar-sweetened beverages serving yogurt items, including yogurt in packed lunches. The authors and energy-dense, solid foods than cups and yogurt tubes (n ¼ 117), Capri expected to find more differences in nonparticipants.8,20 This trend may Sun pouches (n ¼ 112), and the nutrition profile of packed lead to increased caloric intake Lunchables (prepackaged food and lunches among schools because the and ultimately higher body mass drink combinations) (n ¼ 102). Dessert various free and reduced lunch and childhood obesity.21,22 While items were classified as grain-based eligibility levels potentially indicated consumption of energy-dense, low- , chocolate bars, dairy-based various socioeconomic levels. The nutrient foods continues to rise, fruit desserts, gummies, and candy. Of the elementary school with the lowest and vegetable intakes in children are packed lunches, 61% contained a des- free and reduced lunch eligibility below recommended levels and tend sert item and 17% contained > 1des- level (33.3%) had the best nutrition to decrease with age.23,24 In this sert item. The NSLP did not provide profile compared with both other study the NSLP provided increased dessert items to students (Table 3). schools. This potentially indicates exposure to fruits and vegetables that the nutritional value of packed compared with packed lunches, lunches differs significantly among which is critical for acceptance, DISCUSSION schools based on eligibility for especially in young children.15,18 free and reduced school lunch. A Sodium content was higher in To the authors' knowledge, this is the wider variation in eligibility may school lunch than packed lunch first study to compare packed lunches be warranted to observe greater even though packed lunches provided to NSLP lunches over a 5-day period differences by level. many processed foods in the form of among pre-K and K, and after the im- The findings also suggest that prepackaged ready-to-eat items, pri- plementation of new NSLP nutrition packed lunches may contribute to marily owing to the use of processed standards. Like previous studies, the higher solid fat and added sugar food items in preparing NSLP lunches, 624 Farris et al Journal of Nutrition Education and Behavior Volume 46, Number 6, 2014

specifically in entree items. To allow

a food manufacturers time to reformu- late products and schools more time .004) to build student acceptance of lower- ¼ sodium meals, the US Department of .121 .028 .249 .001 .001 .043 .203 .001 .001 .001 .033 .001 Agriculture are expected to transition < < < < < < new sodium standards in the NSLP in 2014–2015.7 Although school

(Bonferroni Adjusted lunch sodium levels are higher than Significance

P they are in packed lunched in this study, the NSLP was meeting 2014– 2015 school year recommended levels. 8.5 3.5 2 5.5 19 37 23 58 Unlike in previous studies, NSLP 679 244 497 260 Range school lunches were lower than the

Interquartile recommended level for iron and calo- ries, which could be the result of food selection by pre-K and K students.1,11 4 3 6.5 258; 46%) Many schools offer 2-3 fruit and 20 72 17 48 47 ¼ 590 vegetable choices per meal along (n with multiple options for a main

b b entree; under the new guidelines c c School 3: 52.7% Eligibility b b 4.4 3.1 6.8 students are only required to take a 21 17 91 46 47 899 838 288 258 minimum of 3 meal components 614 1,382 502 (food groups) at lunch, and only 1 of those food groups must be a fruit or vegetable. This study compares the 3 2 7 5 NSLP foods selected by students, not 47 11 28 64 276 372 603 309 all food items on a lunch menu. Range Mean Median Therefore, this finding does not Interquartile reflect poor adherence of schools in meeting the nutritional guidelines, but rather which food groups or 94; 17%) 4 2.5 7 99 22 54 59 18 items are chosen (or not chosen) by ¼ 649

(n NSLP participants. This study was consistent with previ- b b ous studies that found vitamin C to be c c c c School 2: 46.6% Eligibility 3.7 3.0 7.6 11,13 24 18 higher in packed lunches. Both 54 52 887 847 295 285

676 100 packed and NSLP lunches met 1,080 500 recommended levels; however, the higher amount of vitamin C in packed lunches most likely results from fi 4.5 2 6 forti ed sugar-sweetened beverages. 36 22 57 20 10 235 513 710 309 Several factors may limit the gener- Range Mean Median

Means followed by the same letter are not significantly different at the .004 level (Bonferroni corrected significance), using alizability of this study. The sample Interquartile

b,c was restricted to young elementary students in a rural area in Virginia and may not be applicable to urban, 5 3.5 6 210, 37%) 76 19 36 15 17 ¼

559 ethnically diverse, or older students.

(n This study examined only foods and beverages served and/or offered to c c children over the school lunch period. c b b b School 1: 33.3% Eligibility 5.6 4.0 6.5 20 19

38 Lunches were analyzed using observa- 34 861 782 330 327 Mean Median 570 tion techniques and not weighted samples. Possibly the largest limita- tion of this study was that actual con- sumption of food items was not

Nutritional Profile of PackedEligibility Lunches of Pre-Kindergarten and Kindergarten Students From 3 Separate Elementary Schools by Free and Reduced School Lunch measured. Future studies should also consider gathering food waste data to compare food consumption across

Based on Mann–Whitney–Wilcoxon test; various schools and between packed Table 2. Energy, kcal a Tukey’s Honest Significant Difference. Carbohydrates, g 80 Protein, g Fat, g Saturated fat, g Sodium, mg Fiber, g Sugar, g Vitamin A, IUVitamin C, mg Calcium, mg 1,499Iron, mg 515 and school lunches.25 To protect the Journal of Nutrition Education and Behavior Volume 46, Number 6, 2014 Farris et al 625

7. Nutrition standards in the National Table 3. Comparison of Food Items Included in School Meals and Packed Lunches School Lunch and School of Pre-Kindergarten and Kindergarten Students Program, final rule. 77 Federal Register 17 (January 26, 2012), pp. 4088–4167. School Meals Packed Meals 8. Briefel RR, Wilson A, Gleason PM. (n ¼ 752) (n ¼ 562) Consumption of low-nutrient, energy- Food Item (% of students) (% of students) dense foods and beverages at school, $ 1 fruit item (not juice) 67 54* home, and other locations among school $ 1 vegetable/bean item (not juice) 61 17* lunch participants and nonparticipants. $ 1 dessert item (ie, grain-based desserts, 0 61** JAmDietAssoc. 2009;109:S79-S90. chocolate bars, dairy based desserts, 9. Hur I, Burgess-Champoux T, Reicks M. gummies, candy) Higher quality intake from school lunch > 1 dessert item 0 17** meals compared with bagged lunches. $ 1 savory snack item (ie, chips, crackers) 5 57* ICAN. 2011;3:70-75. No sugar-added juice 22 10* 10. Johnston CA, Moreno JP, El-Mubasher A, Sugar-added juice/soda 0 40** Woehler D. School lunches and lunches Milk (flavored and unflavored) 96 20* brought from home: a comparative anal- ysis. Child Obesity. 2012;8:364-368. < *Statistically significant at .001 level with chi-square test; **Unable to determine 11. Lund SM. Nutritional quality of elemen- statistical significance owing to 0% in category. tary school student lunches in a Minnesota suburban school district. [dissertation]. privacy of elementary students, the children, parents, and perhaps even Duluth, MN: University of Minnesota; research team collected no identifying school policy regarding packed 2008. information to track students over the lunches should be encouraged to pro- 12. Melnik TA, Rhoades SJ, Wales KR, 5-day study period, which would have mote healthier options within packed Cowell C, Wolfe WS. Food consump- allowed for student-level analyses. In lunches and/or encourage participa- tion patterns of elementary schoolchil- previous studies, the longest observa- tion and increase enrollment in NSLP. dren in New York City. J Am Diet tion period was 3 days. Five days is a Assoc. 1998;98:159-164. more appropriate timeframe because 13. Rainville AJ. Nutritional quality of the new nutrition standards provide REFERENCES reimbursable school lunches compared guidance over a 5-day period, and to lunches brought from home in there may be day-to-day variations in 1. US Department of Education. Back to elementary schools in two southeastern the nutritional quality of school and school statistics. http://nces.ed.gov/fast Michigan districts. J Child Nutr Manage- packed lunches that may not be de- facts/display.asp?id¼372. Accessed January ment. 2001;25:13-18. tected during a 3-day observation 15, 2014. 14. US Census Bureau. State and county period. 2. Johnson CM, Bednar C, Kwon J, quickfacts. http://quickfacts.census.gov/ Gustof A. Comparison of nutrient con- qfd/states/51/51121.html. Accessed Aug- tent and cost of home-packed lunches to ust 25, 2014. IMPLICATIONS FOR reimbursable school lunch nutrient 15. Birch L, Fisher JO. Development of RESEARCH AND standards and prices. J Child Nutr Man- behaviors among children and ad- PRACTICE agement. 2010;33:1-8. olescents. Pediatrics. 1998;101:539-549. 3. Ohri-Vachaspati PL, Turner L, 16. Gregory JE, Paxton SJ, Brozovic AM. Habits develop in early childhood and Chaloupka FJ. Fresh fruit and vegetable Maternal feeding practices predict fruit continue into adolescence and adult- program participation in elementary and vegetable consumption in young hood.16,17 Therefore, this is a critical schools in the United States and avail- children: results of a 12-month longitu- time to promote healthy eating. ability of fruits and vegetables in school dinal study. Appetite. 2011;57:167-172. Determining the many factors that lunch meals. J Acad Nutr Diet. 2012;112: 17. Singer MR, Moore LL, Garrahie EJ, influence the decision to participate 921-926. Ellison RC. The tracking of nutrient in the NSLP or bring a packed lunch 4. Gordon AR, Cohen R, intake in young children: the Framing- from home is vital to addressing the Crepinsek MK, Fox M, Hall J, ham Children’s Study. Am J Public poor quality of packed lunches. Zeidman E. The third school nutrition Health. 1995;85:1673-1677. Recently, Ohri-Vachaspati26 reported dietary assessment study: background 18. Skinner JD, Carruth BR, Bounds W, that participation in schools meals and study design. J Am Diet Assoc. Ziegler PJ. Children’s food preferences: was independently associated with 2009;109:S20-S30. a longitudinal analysis. J Am Diet Assoc. whether parents perceived the meals 5. US Department of Agriculture. Na- 2002;102:1638-1647. to be healthy. Additional research is tional School Lunch Program: total 19. Richter SL, Vandervet LM, Macaskill LA, needed to explore additional factors participation. http://www.fns.usda.gov/ Salvadori MI, Seabrook JA, Dwor- related to choosing packed over pd/01slfypart.htm. Accessed August 9, atzek PD. Accuracy and reliability of school lunches and/or decision mak- 2013. direct observations of home-packed ing regarding what is offered in 6. US Senate. 111th Congress. 3307, lunches in elementary schools by trained packed lunches. Furthermore, nutri- Healthy, Hunger-Free Kids Act of nutrition students. JAcadNutrDiet. 2012; tion education programs targeting 2010. 2010. 112:1603-1607. 626 Farris et al Journal of Nutrition Education and Behavior Volume 46, Number 6, 2014

20. Gleason P, Briefel R, Wilson A, and health correlates, NHANES III, 25. Byker CJ, Farris AR, Marcenelle M, Dodd AH. School meal program partici- 1988 to 1994. Arch Pediatr Adolesc Med. Davis GC, Serrano EL. Food waste in a pation and its association with dietary pat- 2003;157:789. school nutrition program after imple- terns and childhood obesity. Princeton, 23. Guenther PM, Dodd KW, Reedy J, mentation of new lunch program guide- NJ: US Dept of Agriculture, Economic Krebs-Smith SM. Most Americans eat lines. JNutrEducBehav. 2014 May 21. Research Service, 2009, Report No. 55. much less than recommended amounts http://dx.doi.org/10.1016/j.jneb.2014.03 21. Drewnowski A, Bellisle F. Liquid calo- of fruits and vegetables. J Am Diet As- .009 [Epub ahead of print]. ries, sugar, and body weight. Am J Clin soc. 2006;106:1371-1379. 26. Ohri-Vachaspati P. Parental perception Nutr. 2007;85:651-661. 24. Lorson BA, Melgar-Quinonez HR, of the nutritional quality of school 22. Kant AK. Reported consumption of Taylor CA. Correlates of fruit and meals and its association with students’ low-nutrient-density foods by Amer- vegetable intakes in US children. JAm school lunch participation. Appetite. ican children and adolescents: nutritional Diet Assoc. 2009;109:474-478. 2014;74:44-47.