at www.jneb.org

Podcast available online Research Article

Testing a Beverage and Fruit/Vegetable EducationIntervention in a University Dining Hall

Mary J. Scourboutakos, PhD1; Catherine L. Mah, MD, PhD2,3; Sarah A. Murphy1;Frank N. Mazza1; Nathanael Barrett4; Bill McFadden, BComm4; Mary R. L'Abb�e, PhD1

1DepartmCanada2Divisiosity of N3Dalla L4BurwasConflict owith thiAddressFaculty oonto, Olabbe@u�2017 PBehaviohttp://dx

Journal

ABSTRACT

Objective: To test the effect of a nutrition intervention that included education and 2 labeling componentson students’ food choices.Design: Repeat cross-sectional study taking place on 6 dinner occasions before and 6 afterward.Setting: The study was conducted during dinner meals in a buffet-style dining hall in a university campusresidence, where students paid a set price and consumed all they cared to eat.Participants: University students (n¼ 368 to 510) visited the cafeteria on each of the data collection dates.Intervention: Fruit and vegetable consumptionwere encouraged; sugar-sweetened beverage consumptionwas discouraged using physical activity calorie equivalent labeling.Main Outcome Measures: Beverage choices and vegetable/fruit bar visits.Analysis: Logistic regression was used to compare the proportion of student who selected each beverage,fruit, or vegetable before and after the intervention, while controlling for menu and gender as covariates.Results: There was a significant decrease in the proportion of students selecting a sugar-sweetenedbeverage before vs after the intervention (49% vs 41%, respectively; P ¼ .004) and an increase in studentschoosing water (43% vs 54%, respectively; P< .001). There was a significant increase in students who tookfruit after the intervention (36%; P< .001) vs before (30%). The number of students visiting the vegetablebar significantly increased from 60% to 72% (P < .001).Conclusions: This interventionmay be a way to encourage healthy dietary choices in campus dining halls.Key Words: cafeteria, fruit, vegetable, sugar-sweetened beverages, university students, physical activitycalorie equivalent label (J Nutr Educ Behav. 2017;49:457-465.)

Accepted February 8, 2017. Published online March 28, 2017.

INTRODUCTION

The diets of university students havebeen shown to be low in fruits and veg-etables1-3 andhigh incalories, sugar, fat,and sodium,3-5 anddonotmeet nutrientrecommendations.6,7 Young adults/adolescents on average consume 230 calfrom sugar-sweetened beverages eachday.8 Furthermore, the transition from

ent of Nutritional Sciences, Faculty of

n of Community Health and Humanitiewfoundland, Newfoundland, Canadaana School of Public Health, Universith Dining Hall, Victoria College, Univf Interest Disclosure: The authors’ conflis article on www.jneb.org.for correspondence: Mary R. L’Abb�e,f Medicine, University of Toronto, FitN M5S 3E2, Canada; Phone: (416) 97toronto.caublished by Elsevier, Inc. on behalf or..doi.org/10.1016/j.jneb.2017.02.003

of Nutrition Education and Behav

high school to auniversityhas been iden-tifiedasacriticalperiod forweightgain.9-15

In recent years, there has been ashift toward buffet-style dining hallsin universities to give students greaterflexibility and more food choices.16

In these unstructured eating environ-ments, the traditional cafeteria lineup,which often dictates a choice from eachfood group, has been replaced with a

Medicine, University of Toronto, Ontario,

es, Faculty of Medicine, Memorial Univer-

y of Toronto, Ontario, Canadaersity of Toronto, Ontario, Canadact of interest disclosures can be found online

PhD, Department of Nutritional Sciences,zGerald Bldg, 150 College St, Rm 315, Tor-8-7235; Fax: (416) 971-2366; E-mail: mary.

f the Society for Nutrition Education and

ior � Volume 49, Number 6, 2017

food court–style setup that gives stu-dents freedom and flexibility in theirfood choices. Such settings overridemany of the barriers (such as cost,lack of access, and convenience) thatprevent individuals from consumingfruits and vegetables.17 However, stu-dents cite the readily available abun-dance of food in their campus dininghalls as a major cause of weight gain.16

Universitystudentshave littleknowl-edge about nutrition18 and are oftencompletely unaware of the calorie con-tentoftheirbeverages.19Efforts toeducateuniversity students about nutrition inclassroom20-24 and online settings25-27

have produced significant changes indietary habits. However, it is unrealisticto expect all students to participate ina nutrition program or course. Thus ithas been suggested that educationalprograms that aim to encourage healthyfood choices should take place in settingswhere food selection actually occurs,such as cafeterias or dining halls.28

Educational interventions have thepotential to influence food choices in

457

Table 1. Calorie Content of Beverages at Cafeteria and Estimated Minutes ofJogging Using Physical Activity Calorie Equivalent Labeling

BeverageCategory Beverage

Calories/8-ozCup

Estimated Minof Jogging toBurn Caloriesa

Water Water 0 0Diet soft drinks Coke Zero 0

Diet Coke 0Coffee and teab Coffee and tea 0

Soft drinks Ginger ale 87 16Sprite 95

Orange pop 100Coke 104

Juice Orange juice (from concentrate) 95 16Lemonade (from concentrate) 95

Cranberry cocktail 104Peach juice 109

Apple juice (from concentrate) 114

Flavoredcoffee/hotchocolate

Swiss mocha cappuccino 149 25

French vanilla coffee 162Hot chocolate 162

458 Scourboutakos et al Journal of Nutrition Education and Behavior � Volume 49, Number 6, 2017

cafeteria settings.29-34 In particular,interventions that direct consumers toa preferred choice without requiringcomparisons or interpretations maybe beneficial because they constitutean environmental nudge toward apreferred choice.29,30,35,36

Young people have said that shock-ing educational messages are neededto reduce their consumption of sugar-sweetened beverages.19 Physical activ-ity calorie equivalent (PACE) labelingillustrates the calorie content of a foodaccordingtotheminutesofphysical ac-tivity required to burn the calories inthat food.Health practitioners also sug-gested that introducing activity-equivalentcalorie labelingonfoodproducts is impor-tanttohelpindividualschangebehavior.37

To date, only 4 studies tested the ef-fectof PACElabelingonconsumer foodchoices; 3 of those studies used Web-based methodologies.38-41 The onlynon-Web-based study, a randomizedstudy in a metabolic kitchen and grad-uate student residence, found thatPACE labels led to a 139-cal decreasein calories ordered compared with thecontrol group (no labels); however,the researchers found no difference be-tween PACE and calorie labels.41

The current study had 2 objectives.The first was to the test the effect ofPACE labeling, combinedwithmessagingencouraging students to drink waterwhen they were thirsty, on students'beverage choices in a university cafete-ria. The second objective was to testthe effect on student's food choices ofmessaging that encouraged them tofill half their plate with fruits and veg-etables alongside a banner giving stu-dents 100 reasons to eat fruits andvegetables. The researchers hypothe-sized that the intervention would leadto fewer students choosing sugar-sweetened beverages and more visitsto the fruit/vegetable bar.

Chocolate milk Chocolate milk 170 27

Plain milkc 2% milk 123 Not available

aCalculated minutes of jogging were based on a general jogging Metabolic Equiv-alent of Task value of 7 and themean weight of a Canadian adult (75.6 kg), accord-ing to the Canadian Health Measures Survey (2009–2011).42; bMilk and sugar thatmay have been added to coffee and tea were not accounted for because these ad-ditions were available at a separate station and could be assessed by the datarecorder; cRegular (plain) milk was not labeled in the intervention as a means toavoid discouraging or encouraging milk consumption.Note: Physical activity calorie equivalent labeling illustrates the calorie content of afood according to the minutes of physical activity required to burn the calories inthat food. Preliminary research demonstrated that such labeling can deter energyconsumption.38,39

METHODSStudy Setting

This was a repeat cross-sectional designto measure the effects of an interven-tion implemented in Burwash DiningHall, a buffet-style cafeteria ina campusresidence, supplied by Gordon FoodServices, at the University of Toronto.The study protocol was approved bythe research ethics board at theUniver-sity of Toronto.

At Burwash Dining Hall, the major-ity of students were enrolled in a mealplan administered by the university;therefore students simply swiped theircard and could consume however muchthey chose, with no external factorssuch as price or convenience to influ-ence their decisions. Each dinner cost$12.00. Students eating in the dining-hall lived in residence, were generallyaged 18–23 years, and came from a widerange of academic programs. The pop-ulationwas 69% female and 31%maleand ethno-culturally diverse.

At each dinner the dining halloffered a number of entr�ees (such aschicken, lasagna, or fajitas), includingat least 1 vegetarian option or halal op-tion (such as Szechwan tofu or spinachsauce with rotini), a daily soup (such asmulligatawny, minestrone, black bean,

or cream of leek), a salad bar (includingvarious kinds of lettuce, fresh vegeta-bles, and dressings), and a selection offruits (apples, oranges, and bananas) andbreads, alongwithmultiple side dishes,beverages, and desserts. Foods servedduring the study collection dates aredescribed in SupplementaryMaterial 1.

The beverage cups in the dininghall held 8 fl oz ounces (when filledapproximately 0.5 cm below thebrim) and were transparent. Studentswere free to take as many cups of asmany beverages as they wished. The19 different beverage options availableat the cafeteria are listed in Table 1. Icewas available but was not accountedfor in these analyses. As for fruit, freshapples, bananas, and oranges wereavailable daily. Pears were occasionallyavailable. A fresh vegetable bar offering

Journal of Nutrition Education and Behavior � Volume 49, Number 6, 2017 Scourboutakos et al 459

various types of lettuce and salad top-pings (cucumbers, peppers, carrots,etc) as well as cooked vegetable optionsat an adjacent hot food bar were avail-able each day.

The Intervention

The study tookplace between September,2014 and April, 2015. The pre phaselasted fromOctober 1 to January 31 andthe post phase lasted from February 1(when the intervention was imple-mented) to April 30. Foodservice in-ventory data were also collected eachmonth fromOctober toApril anddirectobservational data were collected on 6menuoccasionsatdinnerduringtheprephase andon6matched-menuoccasionsduring the postintervention phase.

Beverage Education (PACELabeling)

The first part of the intervention was abeverage education campaign thatencouraged students to drink waterwhen they were thirsty (via a 3 � 2.5-ftposter) (Figure 1E) and used labelingthat illustrated the number of minutesof jogging it took to burn calories inthe different beverage options offered

Figure 1. Images of the beverage, fruit. andof the vegetable bar (8.5 � 11 in); (B) large bacafeteria (2 � 20 ft); (C) Harvard’s Healthy Eaequivalent labeling illustrating the number of m(E) View of the beverage station including sig

in the dining hall (via a 1 � 6-ft poster)(Figure 1D).

This idea was adapted from an info-graphic (illustrating the number ofpackets of sugar in different beverages)produced by a public health nongov-ernmental organization–led campaignwas not previously evaluated.42 How-ever, in this study, instead of packetsof sugar, minutes of jogging, a formof PACE labeling,39,43 was usedcombined with a positive message todrink water when you are thirsty.

Theminutesof joggingon theposterwere based on the total calories in eachbeverage according to the manufac-turers' data provided by Gordon FoodServices.On theposter and in the studyanalysis, the 19 beverages were catego-rized into the following groups: juice,soft drinks, chocolate milk, flavoredcoffee/hot chocolate, diet soft drinks,coffee/tea, and plain water (Table 1).Regular (plain) milk was not labeled,to avoid discouraging or encouragingmilk consumption. Minutes of joggingfor each beverage category were aver-aged andestimatedusing theMetabolicEquivalent of Task equation (1 MET ¼kcal/[kilograms/hour]).44 The calcula-tion was based on a general joggingMET value of 7 and the mean weightof a Canadian adult, 75.6 kg, according

vegetable intervention: (A) Harvard’s Healthynner giving students 100 reasons to eat fruitsting Plate sign at the fruit station (4 � 5 ft); (Dinutes of jogging it takes to burn the caloriesn encouraging students to Drink water when

to the Canadian Health Measures Sur-vey (2009–2011).45

Because thewater came out of a tap,signs that saidDid you know thiswater isfiltered? (10 � 30 in) were installed inSeptember, 2014, before the studybegan. They were present throughoutthe pre and post phase to ensure thatfear of tap water did not deter waterconsumption during the study.

Fruit and Vegetable Education

Thispartof theinterventionwasanutri-tion promotion campaign promotingfruit and vegetable consumption. Itwas composed of posters hung at theentranceof thedininghall (not shown),at the fruit station (4 � 5 ft in size)(Figure 1C), and posted on top of thevegetable bar (8.5 � 11 in) (Figure 1A).In addition, a banner giving students100 reasons to eat fruits and vegetableswas hung in the entranceway to theserving area (2 � 20 ft) (Figure 1B). Thepositionof themessageswithin thedin-inghallwas strategically selected tomakethe information attention-grabbingand maximize exposure to the inter-vention, so that they were immedi-ately visible to everyone entering thedining hall.29

Eating Plate infographic displayed on topand vegetables at the entranceway to the) Sign illustrating physical activity caloriein each beverage option offered (1 � 6 ft);you are thirsty (3 � 2.5 ft).

460 Scourboutakos et al Journal of Nutrition Education and Behavior � Volume 49, Number 6, 2017

To encourage students to eat morefruits and vegetables, text stating Fillhalf your plate with fruits and vegetablesand This is what a balanced meal lookslike was posted alongside a Harvard'sHealthy Eating Plate infographic.46

The content of the posters was basedon principles from the literature showingthat positive health messages (eat this)framed in terms of the benefits aremore effective than negative messages(don't eat that) when the target popu-lation is university students who eaton campus.47,48

Intervention Outcomes

Twoprimary outcomesweremeasured:beverage choices and fruit and vege-table choices. The intervention aimedto decrease consumption of sugar-sweetened beverages as well as increasefruit and vegetable consumption.

Data CollectionDirectlymeasured observational datacollection. In-person, direct observa-tional data collection took place duringdinner from4:30to7:30PMon6specificmenu occasions before the interventionand on the same 6 menu occasions af-ter it.Thedininghallhasa5-weekrotatingmenu cycle; therefore, data collectiondates (listed in Supplementary Material 2)were selected to ensure that data werecollected ondayswhen the same foodswere being served before and after theintervention (matched menus). DatawerecollectedonTuesdayandWednesdayevenings (because these were themosthighly attended weekdays) during 3 ofthe 5 menu cycles.

A 2-person data collection team,including the lead author (MJS) and aresearch assistant trained by MJS, wasstationed in the dining hall for theduration of dinner on the 12 datacollection dates. One surveyor was sta-tioned in the beverage area. Using aclipboard and a pencil, this surveyor re-corded the gender and beverage choiceof every studentwho selected abeverage,in addition to themeasured cupvolume(measured visually in quarter-cup in-tervals, eg, full cup, three-quarter cup,half-cup, one-quarter cup) of beveragetaken. This observer was conspicu-ously located beside the beverage sta-tion; thus, students were aware that astudywas takingplace.When studentsinquired about the study (<1% of stu-

dents inquired throughout the entirestudy), they were told that the studywas investigating food waste. Multiplevisits by the same student were re-corded as separate, independent visits.

The second data collector was sta-tioned near the fruit and vegetable barto record fruit and vegetable choices.Using a clipboard and pencil, this sur-veyor recorded the frequencyof studentstaking each whole fruit (eg, apples, ba-nanas, oranges, pears) aswell as the fre-quency of students taking vegetablesfrom the fresh or cooked vegetable sta-tion. The surveyor also recorded thegender of each student taking fruitsor vegetables. The denominator for fruitandvegetable analyseswas basedon thetotal number of students visiting thecafeteria at each dinner during whichdata were collected. This information wasprovidedbycafeteria staffwhoroutinelyrecorded the number of attendees.

A description of direct observa-tional data collection dates includedin the final analysis can be found inSupplementary Material 2. Directobservational data collection for fruiton February 11 was excluded fromthe final analysis because a Valentine'sDay dessert buffet confounded stu-dents' fruit selections on that date. Inaddition, fruits and vegetables selectedduring the first 2 data collection dateswere collected using a different meth-odology and thus were excluded forconsistency.Rateof fruit disappearance(ie, number of fruits present at the startvs the end)was theoriginal data collec-tionmethod; however, it yielded inac-curate results and was thus changed todirectly measured observation.

Cafeteria inventory data collection.Inventory data, representing long-term secular trends in the cafeteria'ssupply purchasing, were collected as asecondary data source. The inventorydata analysis was originally intendedto verify that the data collected viadirect observation were not the resultof social desirability bias. Inventorydata were provided by dining hall staffvia their suppliers on a monthly basis.Inventory data represented consump-tion during breakfast, lunch, and din-ner for the entire study duration. Thenumber of cases of beverages orderedeach month from September, 2014 toApril, 2015 was provided. The numberof cases of fruit ordered each month

during the study period was alsocollected. Inventory data for vegetableswere not available. Data for September,December, and April were excludedfrom inventory data analyses becausethese months were not full academicmonths and were consequently under-attended; hence ordering was atypical.

Analysis

The total number of students choosingeach beverage (eg, soft drinks, juice)was tabulated and calculated as a pro-portion of total visits to the beveragestation.Beveragevolumewasconvertedfrom the cup measure (full cup, three-quarters cup,halfcup, etc) toanapprox-imated volume (milliliters) and furtherconverted to calories for all caloric bev-erages. All analyses were conducted us-ing SAS. (version 9.3; SAS Institute,Inc., Cary, NC; 2011). The analysesused logistic regression for primary ana-lyses comparing the proportion of stu-dents selecting each beverage beforeand after the intervention, while con-trolling for covariates including genderand the menu served on each datacollection date. In secondary analyses,data for all sugar-sweetened beverages(juice, softdrinks,chocolatemilk,flavoredcoffee, andhot chocolate) takenbeforeand after the intervention were aggre-gated and compared. When interpret-ing significance, Bonferroni adjustmentswere used to control for multiple com-parisons. Average calories ormilliliters(fornoncaloricbeverages)of eachbeveragetaken by students before and after theinterventionwerecalculatedandcomparedusingANCOVA (proc glm) andMonte-Carlo simulationof theexactP (becausedata were not normally distributed)while controlling for gender and menu.The proportion of students visiting thesalad bar and the proportion takingfruit before vs after the interventionwere compared using logistic regres-sion while controlling for menu andgender as a covariate, and adjustingfor multiple comparisons using theBonferroni method.

Inventory data for fruit were nottransformed. Inventory data for bever-ageswereconvertedintolitersofbeverage.The number of liters of each individualbeverage ordered eachmonth from sup-pliers was adjusted to a weekly averageto account for missed weeks such asreading week, and final exams week

Journal of Nutrition Education and Behavior � Volume 49, Number 6, 2017 Scourboutakos et al 461

when cafeteria attendance was lowerthan usual. Proc correlation was usedto test agreement between the inven-tory and directly measured observa-tional data.

RESULTS

A range of 368–510 students visitedthe dining hall for each dinner mealon the nights whendatawere collected(Supplementary Material 2). Overall, atotal of 6,412 visits to the beverage sta-tion were recorded (3,232 pre and3,180 post), totaling 8,570 cups filledwith beverages (Table 2). In all, 56%of beverages was taken by femalesand 44% by males. A total of 3,668visits to the vegetable bar were re-corded, with 63% of visits made by fe-males and 36% bymales. A total of 954visits to the fruit stationwere recorded,with 56% of visits made by femalesand 44% by males.

Inventory data for average liters ofeach beverage and number of casesof fruit ordered each month wereincluded for the months of October,November, and January in the pre

Table 2. Sample Characteristics at BaselinStation and Frequency of Bevera

Sample Characteristics

Total recorded visits to beverage stationa

Visits preintervention

Visits postintervention

Total cups filled with each beverage

Proportion of cups filled with each beveragWaterJuiceSoft drinksChocolate milkCoffee/teaMilkDiet soft drinksFlavored coffee/hot chocolate

Total recorded visits to vegetable bara

Total recorded visits to fruit stationb

Frequency of each fruit takenApplesBananasOrangesPears

aBased on 12 data collection dates (6 pbBased on 9 data collection dates (4 prein

phase as well as February and Marchin the post phase of the intervention.

Beverages (Direct ObservationalCollection)

At baseline, the most popular beveragechoices among students were water(40% of cups were filled with water),juice (18%), soft drinks (14%), andchocolate milk (8%) (Table 2). Afterthe intervention was implemented,therewasa smalldecrease in thepropor-tionof students taking softdrinks, juice,chocolatemilk, flavored coffee, and hotchocolate; however, this trend was notsignificant. When all sugar-sweetenedbeverages were combined, there was asignificant decrease (P ¼ .004) in theproportionof students selectinga sugar-sweetened beverage before vs after theintervention (49% vs 41%, respectively)(Figure 2). The average amount of calo-ries or milliliters (for noncaloric drinks)of each beverage taken before vs afterthe intervention was not significantlydifferent (SupplementaryMaterial 3).

There was a significant increase(P < .001) in the proportion of stu-

e Including Visits to Beverage and Fruitge and Fruit Consumption

n

6,412

3,232

3,180

8,570

e at baseline1,640 (40%)739 (18%)591 (14%)321 (8%)305 (7%)267 (3%)138 (7%)105 (3%)

3668

954

393 (41%)292 (31%)194 (20%)75 (8%)

reintervention and 6 postintervention);tervention and 5 postintervention).

dents drinking water before vs afterthe intervention (43% vs 54%, respec-tively) (Figure 2).

Fruits and Vegetables (DirectObservational Collection)

There was a significant difference(P< .001) in the proportion of studentstaking fruit before vs after the interven-tion (30%vs36%, respectively) (Figure 3).Theproportionof students visiting thevegetable bar significantly increasedfrom 60% to 72% (P < .001).

Inventory Data Results

Inventory data for beverages showedsome trends toward decreases (amongchocolate milk, apple juice, orangejuice and flavored coffees); however,equally large increases were observedamong other beverages, such as cran-berry cocktail and lemonade (Table 3).Inventory data for the number of fruitcases ordered each month showed nodiscernable trend (Figure 4). With theexception of milk, inventory data forjuices, chocolate milk, flavored coffee/hot chocolate, tea/coffee, apples, oranges,bananas, andpearswere not correlatedwith direct observational data collec-tion findings.

DISCUSSION

This studyshowedthatanutritioninter-vention that included education and 2labeling components (PACE labeling aswell as messaging to fill half your platewith fruits and vegetables) could resultin a modest decrease in the number ofstudentschoosingsugar-sweetenedbev-erages and a modest increase in thosechoosing water and visiting the fruitand vegetable bar.

At the start of this research, therewere only 4 investigations of PACE la-beling; 3 of the 4 were conducted usingonline surveys.38-40This study investigatedPACE labelling in a real world setting.Unlike previous online surveys testingPACE labeling, which showed decreasesof 93–206 calories,39,43 this study foundthat students who continued to choosesugar-sweetened beverages after theinterventionwere on average choosingthe same number of calories as beforethe intervention. This suggests thatthis intervention decreased the numberof students choosing sugar-sweetened

Figure 2. Proportion of students who chose each beverage before and after the intervention (*P < .001), logistic regression andMonte-Carlo simulations of exact P, controlling for menu and sex.

462 Scourboutakos et al Journal of Nutrition Education and Behavior � Volume 49, Number 6, 2017

beverages but did not appear to changethe amount of calories the habitualdrinkers were choosing.

Unlike this study, which detected asimultaneous increase in theproportionof students drinking water, previousinterventions that discouraged sugar-sweetened beverage consumption, suchas the Dutch Obesity Intervention inteenagers, did not see a simultaneousincrease inwater.49 Furthermore,meta-analyses of beverage taxation foundthat consumers replaced sugar-sweetenedbeverages with fruit juice, whole milk,anddietsoftdrinks,ie,switchingbehavior.50

In this study, the researchers did not

Figure 3. Proportion of students who visitedand after the invention (*P < .001), logistic reexact P, controlling for menu and sex.

observe switching behavior becausethere were no increases in fruit juice,diet soft drinks, or milk during thepost phase. The supplementary state-ment thatwas included in the interven-tion, Drink water when you are thirsty,may have encouraged students specif-ically to replace sugar-sweetened bev-erages with water. However, this is notcertain. Further research is needed todisentangle the relative effects of label-ing, nutrition promotion messagessuch as the statement about drinkingwater, and other factors such as priceto discern whether 1 component ofthe intervention was mainly respon-

the fruit station and vegetable bar beforegression and Monte-Carlo simulations of

sible for switching effects, or whetherthere was a synergy by having multiplecomponents present.

Previous research inauniversitycon-venience store showed that simplytagging healthy options with a labelthat said Fuel your life healthy campuscan increase purchases of tagged itemsby 3.6%,30whichwas a smaller increasethan was observed in this study, inwhich thenumberof studentschoosingfruit increased by 6%, whereas waterincreasedby11%andvegetables increasedby 12%. The higher increase observedin this study could be explained byresearch showing that educationalstrategies that link knowledge aboutthe attributes of a food with self-relevant health consequences were morelikely toproducebehavior change.47,51

This was accomplished by combiningthe directional statement Fill half yourplate with fruits and vegetables with abanner highlighting 100 reasons to eatfruits and vegetables, and aligns withprevious research showing that providingstudents with information about thebenefits of a food may be superior tosimply highlighting healthy choices.51

The lack of congruency between theinventory data and the data collectedvia direct observation suggested thatresults should be interpreted withcaution. However, there are a numberof potential explanations for this lackof correlation. First, the data collectedvia direct observation only representedchoices made at dinner, whereas in-ventory data represented all choicesmade during breakfast, lunch, and din-ner. Second, there were a limited num-ber of data points in the inventory

Table 3. Inventory Data Illustrating Weekly Average Number of Liters of EachBeverage Taken Before and After Intervention

BeveragePrea Postb

% ChangeMean Liters/Wk Mean Liters/Wk

2% milk 313 303 �10

Chocolate milk 245 217 �9

Cranberry cocktail 85 128 15

Lemonade 104 123 12

Apple juice 184 110 �6

Orange juice 224 208 �9

Peach juice 120 151 13

Total juice 717 720 10

French vanilla cappuccino 26 19 �8

Swiss mocha cappuccino 23 5 �2

Hot chocolate 34 49 14

aPre data were based the weekly average over 12 weeks; bPost data were basedon the weekly average over 7 weeks. Reading week and 3 weeks in April wereexcluded owing to decreased attendance in the dining hall.Note: Inventory data for soft drinkswas not provided by the suppliers and thereforewas not included in this analysis.

Journal of Nutrition Education and Behavior � Volume 49, Number 6, 2017 Scourboutakos et al 463

data: September, December, and Aprilhad to be excluded because they wereincomplete months, which left only5 months (3 pre and 2 post). Third, in-ventory decisions at the cafeteria wereretrospective rather than prospective,ie, using past consumption as a predic-tion for future consumption. For example,if the dining hall ordered an excess ofa certain beverage in September, theywould order less in October. Theselimitations illustrate some of the prob-lems of using inventory data in aprogram

Figure 4. Inventory data trends in the cases

intervention setting, even though esti-mates of food availability in terms ofcommodity supply and disappearanceare a commonly used economic metricfor consumption, and in evaluation ofpolicy interventions.52 Essentially in-ventory data are an availabilitymeasureand are perhaps more appropriate as ameasure of environmental changerather than a proxy for consumption.For instance, in this study, the sharpdrop in oranges observed during Octoberrepresented a supply shortage, not a

of fruit ordered each month.

decrease in consumption. Inventory datawould also be more useful when usedfor longitudinal analysis in time series.

The researchers used inventory dataas a second data source to assess poten-tial social desirability bias in datacollected through direct observation.It remains unclear as to whether thisobjectivewasmet. Intheory, theobserveddecrease in sugar-sweetened beverageswasmore likely to be influenced by theeffect of social desirability bias becausethe surveyor was conspicuous. Mean-while, the observed increase in fruitsand vegetables was less likely to beinfluenced, because the recorder wasless conspicuously located.

Previous research demonstratedthat a majority of university studentsare interested in receivingnutrition in-formation.53 These results may not begeneralizable to other populations,because this study was conductedwith a population more likely to beinterested in the intervention. Food se-lection behavior is also more likely tobe influenced by labeling in situationswith a limited range of choices, similarto the beverage station at the dininghall, where labeling can contrast withthe options available and provide aclear cue as to the preferred choice.54

Price was not a factor in decision mak-ing in this dining hall, which alsolimited generalizability to other con-sumer food environments.

Other limitations included the useof food and beverages taken tomeasurechoice and the fact that this may nothavecorrelatedwithanobjectivemeasure

464 Scourboutakos et al Journal of Nutrition Education and Behavior � Volume 49, Number 6, 2017

of dietary intake. In addition, italthough whole fruit and vegetableitems were measured, this underesti-matedproduceconsumption,particularlytotal vegetables taken, because oftenvegetables were mixed in with thehot entr�ees and thus could not be ac-counted for via the data collectionmethodology employed.

Another limitation was the repeatcross-sectional design, which limitedthe types of inference that could bemade about the effects of the interven-tion. However, many of the con-founders in this study, including theorange shortages, special food days,and Valentine's dessert buffet, wouldstill be problematic, even with a morerigorous study design.

Seasonal variation, sometimes referredto as temporal variation, is an impor-tant characteristic of measuring foodsales and purchasing over the courseof a calendar year. A strength of thisstudy was that measurements weremade on occasions over a full calendaryear.However, because datawas collectedcross-sectionally and not longitudinallyover this period, the contribution oftemporal variation to variation in foodchoice behavior could not be analyzed.Thatbeingsaid,previous researchshowedthatuniversity students' cafeteria snackpurchases become less healthy witheach passing week of the semester55;the researchers did not observe this inthe current study. Another strength wasthat efforts were made to minimize theeffect of seasonal food choices. Forinstance, the intervention was startedin February to prevent January weightloss resolutions from affecting results.Itwasfoundthat foodandbeveragechoicesin January were not different fromthose in any other month. Further-more, the long duration of the studyillustrated that a washout effect wasnot observed 2 months after the inter-ventions were implemented.

IMPLICATIONS FORRESEARCH ANDPRACTICE

This study showed that an interventioncomposed of nutrition education and 2labeling components (focused on bev-erages and fruits and vegetables) maybe auseful strategy todecrease students'consumption of sugar-sweetened bev-erages and increase consumption of

water, fruits, and vegetables. Moreover,these results showed that specificallyPACElabelingmaybeawaytomotivatehealthier beverage choices inuniversitycafeteria settings. Future research isneeded to test this intervention inotherpopulations and in settings where costis a factor in food choice.

ACKNOWLEDGMENTS

Thisworkwas supportedbyaCanadianInstitutes of Health Research, VanierCanada Graduate Scholarship (MJS),Cancer Care Ontario/Canadian Insti-tutesofHealthResearchTrainingGrantinPopulation Intervention forChronicDisease Prevention: A Pan-CanadianProgram (grant no. 538932) (MJS);and a McHenry Unrestricted ResearchGrant from the University of Toronto(MRL). A sincere thank you goes outto the former president of Victoria Col-lege, Dr Paul Gooch, for allowing theauthors to conduct this study at his col-lege and for supporting this endeavor.In addition, thank you to all of the col-lege staff and students who kindlywelcomed the authors into their dininghall. Special thank you to AdrianaChomka,AndreaHung, andLaurenRe-nlund for assisting with data collectionand/or data entry; to the L'Abb�e labgirls, especially Dr JoAnne Arcand, forintellectual input and advice; and toDr Paul Corey for statistical guidance.

SUPPLEMENTARY DATA

Supplementary data related to thisarticle can be found online at http://dx.doi.org/10.1016/j.jneb.2017.02.003.

REFERENCES

1. Richards A, Kattelmann KK, Ren C.Motivating 18- to 24-year-olds to increasetheir fruit and vegetable consumption. JAm Diet Assoc. 2006;106:1405-1411.

2. Chung S-J, Hoerr SL. Predictors of fruitand vegetable intakes in young adults bygender. Nutr Res. 2005;25:453-463.

3. Horacek TM, Betts NM. College stu-dents’ dietary intake and quality ac-cording to their Myers Briggs Typeindicator personality preferences. JNutr Educ. 1998;30:387-395.

4. Hampl JS, Betts NM. Comparisons ofdietary intake and sources of fat in low-and high-fat diets of 18- to 24-year-olds. J Am Diet Assoc. 1995;95:893-897.

5. Anding JD, Suminski RR, Boss L. Die-tary intake, body mass index, exercise,and alcohol: are collegewomen followingthe dietary guidelines for Americans? JAm Coll Health. 2001;49:167-171.

6. Burke JD, Reilly RA, Morrell JS,Lofgren IE. The University of NewHampshire’s Young Adult HealthRisk Screening Initiative. J AmDiet As-soc. 2009;109:1751-1758.

7. SchuetteLK,SongWO,HoerrSL.Quan-titative use of the Food Guide Pyramid toevaluate dietary intake of college students.J Am Diet Assoc. 1996;96:453-457.

8. Nielsen SJ, Popkin BM. Changes inbeverage intake between 1977 and2001. Am J Prev Med. 2004;27:205-210.

9. AndersonDA,Shapiro JR,Lundgren JD.The freshman year of college as a criticalperiod for weight gain: an initial evalua-tion. Eat Behav. 2003;4:363-367.

10. Wengreen HJ, Moncur C. Change indiet, physical activity, and body weightamong young-adults during the transi-tion from high school to college.Nutr J.2009;8:32.

11. Hoffman DJ, Policastro P, Quick V,Lee SK. Changes in body weight andfat mass of men and women in the firstyear of college: a study of the ‘‘freshman15.’’. J Am Coll Health. 2006;55:41-45.

12. Vella-Zarb RA, Elgar FJ. The ‘‘fresh-man 5’’: a meta-analysis of weightgain in the freshman year of college. JAm Coll Health. 2009;58:161-166.

13. Jung ME, Bray SR, Martin Ginis KA.Behavior change and the freshman 15:tracking physical activity and dietarypatterns in 1st-year university women.J Am Coll Health. 2008;56:523-530.

14. Gropper SS, Simmons KP, Gaines A,et al. The freshman 15—a closer look.J Am Coll Health. 2009;58:223-231.

15. Butler SM, Black DR, Blue CL,Gretebeck RJ. Change in diet, physicalactivity, and bodyweight in female col-lege freshman.Am J Health Behav. 2004;28:24-32.

16. Nelson MC, Kocos R, Lytle LA,Perry CL. Understanding the perceiveddeterminants of weight-related behav-iors in late adolescence: a qualitativeanalysis among college youth. J NutrEduc Behav. 2009;41:287-292.

17. Yeh MC, Ickes SB, Lowenstein LM,et al. Understanding barriers and facili-tators of fruit and vegetable consump-tion among a diverse multi-ethnicpopulation in the USA. Health PromotInt. 2008;23:42-51.

18. Matthews JI, Doerr L, Dworatzek PD.University students intend to eat better

Journal of Nutrition Education and Behavior � Volume 49, Number 6, 2017 Scourboutakos et al 465

but lack coping self-efficacy and knowl-edge of dietary recommendations. JNutr Educ Behav. 2016;48:12-19.e1.

19. Block JP, Gillman MW, Linakis SK,Goldman RE. ‘‘If it tastes good, I’mdrinking it’’: qualitative studyofbeverageconsumption among college students. JAdolesc Health. 2013;52:702-706.

20. HaE-J,Caine-BishN.Effect of nutritionintervention using a general nutritioncourse for promoting fruit and vegetableconsumption among college students. JNutr Educ Behav. 2009;41:103-109.

21. Ha EJ, Caine-Bish N, Holloman C,Lowry-Gordon K. Evaluation of effec-tiveness of class-based nutrition interven-tion on changes in soft drink and milkconsumption among young adults.Nutr J. 2009;8:50.

22. Ha EJ, Caine-Bish N. Interactive intro-ductory nutrition course focusing ondisease prevention increased whole-grain consumption by college students.J Nutr Educ Behav. 2011;43:263-267.

23. Matvienko O, Lewis DS, Schafer E. Acollegenutrition science course as an inter-vention to prevent weight gain in femalecollege freshmen. J Nutr Educ. 2001;33:95-101.

24. Emrich TE, Mazier MJ. Impact ofnutrition education on university stu-dents’ fat consumption. Can J Diet PractRes. 2009;70:187-192.

25. Gow RW, Trace SE, Mazzeo SE. Pre-venting weight gain in first year collegestudents: an online intervention to pre-vent the ‘‘freshman fifteen.’’. Eat Behav.2010;11:33-39.

26. Harvey-Berino J, Pope L, Gold BC,Leonard H, Belliveau C. Undergradand overweight: an online behavioralweight management program for col-lege students. J Nutr Educ Behav. 2012;44:604-608.

27. Franko DL, Cousineau TM, Trant M,et al. Motivation, self-efficacy, physicalactivity and nutrition in college stu-dents: randomized controlled trial ofan internet-based education program.Prev Med. 2008;47:369-377.

28. Zifferblatt SM, Wilbur CS, Pinsky JL.A new direction for public health care:changing cafeteria eating habits. J AmDiet Assoc. 1980;76:15-20.

29. Buscher LA, Martin KA, Crocker S.Point-of-purchase messages framed interms of cost, convenience, taste, andenergy improve healthful snack selec-tion in a college foodservice setting. JAm Diet Assoc. 2001;101:909-913.

30. Freedman MR, Connors R. Point-of-purchase nutrition information influencesfood-purchasing behaviors of college stu-dents: a pilot study. JAmDiet Assoc. 2010;110:1222-1226.

31. Vanderlee L, Hammond D. Does nutri-tion information on menus impact foodchoice? Comparisons across two hospi-tal cafeterias. Public Health Nutr. 2014;17:1393-1402.

32. Vyth EL, Steenhuis IH, Heymans MW,Roodenburg AJ, Brug J, Seidell JC. In-fluence of placement of a nutrition logoon cafeteria menu items on lunchtimefood Choices at Dutch work sites. JAm Diet Assoc. 2011;111:131-136.

33. Turconi G, Bazzano R, Roggi C,Cena H. Helping consumers make amore consciousnutritional choice: accept-abilityofnutrition informationat a cafete-ria. Public Health Nutr. 2012;15:792-801.

34. Ries CP, Schoon DK. Evaluation of acafeteria-based education program for col-lege students. JNutr Educ. 1986;18:107-110.

35. Auchincloss AH, Young C, Davis AL,Wasson S, Chilton M, Karamanian V.Barriers and facilitators of consumer useof nutrition labels at sit-down restaurantchains. Public Health Nutr. 2013;16:1-8.

36. Cohen DA, Babey SH. Contextual in-fluences on eating behaviours: heuristicprocessing and dietary choices. ObesRev. 2012;13:766-779.

37. Cramer S. Food should be labelled withthe exercise needed to expend its calo-ries. BMJ. 2016;353:i1856.

38. Dowray S, Swartz JJ, Braxton D,Viera AJ. Potential effect of physical ac-tivity based menu labels on the caloriecontent of selected fast food meals.Appetite. 2013;62:173-181.

39. Antonelli R, Viera AJ. Potential effectof Physical Activity Calorie Equivalent(PACE) labeling on adult fast foodordering and exercise. PLoS One.2015;10:e0134289.

40. Viera AJ, Antonelli R. Potential effectof physical activity calorie equivalentlabeling on parent fast food decisions.Pediatrics. 2015;135:e376-e382.

41. James A, Adams-Huet B, Shah M.Menu labels displaying the kilocaloriecontent or the exercise equivalent: ef-fects on energy ordered and consumedin young adults. Am J Health Promot.2015;29:294-302.

42. Live Well Omaha—Make the HealthyChoice the Easy Choice. http://livewellomaha.org/. Accessed November 17,2015.

43. Platkin C, YehMC, Hirsch K, et al. Theeffect of menu labeling with calories andexercise equivalents on food selectionand consumption.BMCObes. 2014;1:21.

44. Compendium of physical activities.https://sites.google.com/site/compendiumofphysicalactivities/. Accessed November17, 2015.

45. Statistics Canada Health Statistics Divi-sion. Canadian Health Measures Survey:Cycle 2Data Tables 2009 to 2011. Ottawa,Canada: Health Canada; 2013.

46. Harvard School of Public Health.Healthy Eating Plate and Healthy EatingPyramid. http://www.hsph.harvard.edu/nutritionsource/healthy-eating-plate/. Ac-cessed November 18, 2015.

47. Rothman AJ, Salovey P. Shaping per-ceptions to motivate healthy behavior:the role of message framing. PsycholBull. 1997;121:3-19.

48. Wansink B, Pope L. When do gain-framed health messages work betterthan fear appeals?NutrRev. 2015;73:4-11.

49. Veitch J, Singh A, van StralenMM, vanMechelen W, Brug J, Chinapaw MJ.Reduction in sugar-sweetened bever-ages is not associated with more wateror diet drinks. Public Health Nutr.2011;14:1388-1393.

50. Cabrera Escobar MA, Veerman JL,Tollman SM, Bertram MY, Hofman KJ.Evidence that a tax on sugar sweetenedbeverages reduces the obesity rate: ameta-analysis. BMC Public Health. 2013;13:1072.

51. WansinkB,WestgrenRE,CheneyMM.Hierarchy of nutritional knowledge thatrelates to the consumption of a functionalfood.Nutrition. 2005;21:264-268.

52. US Department of Agriculture Eco-nomic Research Service. Food Avail-ability (per capita) Data System. https://www.ers.usda.gov/data-products/food-availability-per-capita-data-system/. Ac-cessed December 28, 2015.

53. BoehmS, Selves EJ,RaleighE,RonisD,Butler PM, Jacobs M. College students’perception of vulnerability/susceptibilityand desire for health information. PatientEduc Counsel. 1993;21:77-87.

54. Davis-Chervin D, Rogers T, Clark M.Influencing food selection with point-of-choice nutrition information. JNutr Educ. 1985;17:18-22.

55. Wansink B, Cao Y, Saini P, ShimizuM,Just DR. College cafeteria snack foodpurchases become less healthy witheach passing week of the semester. PublicHealth Nutr. 2013;16:1291-1295.

465.e1 Scourboutakos et al Journal of Nutrition Education and Behavior � Volume 49, Number 6, 2017

CONFLICT OF INTEREST

The authors have not stated any con-flicts of interest.