Digital Photography as a Tool to Measure School Cafeteria Consumption

RE S E A R C H AR T I C L E

Digital Photography as a Tool to MeasureSchool Cafeteria Consumption

MARK SWANSON, PhDABSTRACT

BACKGROUND: Assessing actual consumption of school cafeteria meals presents

challenges, given recall problems of children, the cost of direct observation, and the

time constraints in the school cafeteria setting. This study assesses the use of digital

photography as a technique to measure what elementary-aged students select and

actually consume from school cafeteria meals.

METHODS: Before and after still digital photographs were taken of labeled trays for

every lunch served to elementary students over 4 lunch periods. Two analysts visually

estimated the amount of each item consumed from every tray, and those estimates

were compared to evaluate interrater reliability.

RESULTS: Collection of photographic data was rapid and did not disrupt the busy

elementary cafeteria setting. Analysts’ estimates of consumption levels of meal com-

ponents (main and side dishes) were within 10% of each other in 92% of the cases.

Only 0.2% of items could not be analyzed due to children playing with food or other-

wise obstructed photographs.

CONCLUSIONS: Digital photography offers researchers and school food service per-

sonnel a highly accurate and cost-effective tool to measure actual consumption of

school cafeteria meals. Data collected through this method can be evaluated by sim-

ple counts of servings of produce or other food groups or by more detailed analyses

of nutritional composition.

Keywords: nutrition and diet; school food services; methods and materials of instruction.

Citation: Swanson M. Digital photography as a tool to measure school cafeteria

consumption. J Sch Health. 2008; 78: 432-437.

Assistant Professor, ([email protected]), Department of Health Behavior, College of Public Health, University of Kentucky, 121 Washington Ave, Lexington, KY40536-0003.

Address correspondence to: Mark Swanson, Assistant Professor, ([email protected]), Department of Health Behavior, College of Public Health, University ofKentucky, 121 Washington Ave, Lexington, KY 40536-0003.

The author wishes to thank Dr. Mark Dignan and the Center for Prevention Research at the University of Kentucky for financial support of this research.

432 d Journal of School Health d August 2008, Vol. 78, No. 8 d ª 2008, American School Health Association

One of the most critical issues affecting the health

of schoolchildren is dietary intake. Because of its

potential to affect significant numbers of children, the

school food environment has received increasing

attention in recent years as an important venue in

which to center efforts to combat poor nutrition. As

Kolbe1 suggested ina recent commentary in this journal,

school food services should be considered a component

ofmodern school health programs. The extent of weight

problems facing the nation’s children is sobering. Data

from the 1999-2000 National Health and Nutrition

Examination Survey indicate that over 15% of elemen-

tary school–aged children (aged 6-11) are overweight

(BMI for age � 95th percentile), and over 30% of the

same age group have a BMI at the 85th percentile or

higher. Entry into elementary school seems to be the

age at which weight problems escalate dramatically;

the prevalence of overweight or at risk for overweight

in children aged2-5 is 20.6%, jumping to 30.3% for ages

6-11, and increasing slightly to 30.4% for ages 12-19.2

In FY2006, over 30 million children participated in

the National School Lunch program.3 Recognizing the

great impact of the school food environment, a major

goal of Healthy People 2010 is to improve the quality of

foods eaten in school by students.4 School meals repre-

sent an important portion of the calories consumed by

children,5 significantly more than the school vending

machines and snack foods which have received a great

deal of attention in recent years.6 School food service

programshaveworked to improve the nutritional qual-

ity and appeal of meals offered to students in recent

years, particularly working to encourage greater con-

sumption of fruits and vegetables.7,8 US Department of

Agriculture standards are centered on what schools

must offer students, and some researchers have focused

exclusively on what students have chosen,9 yet as

Georgiou and colleagues point out, it is important to

distinguish between what is offered to students, what

students take, and what students consume.10 Without

accurate methods of measuring school cafeteria food

consumption,we remain stymied inour ability to assess

potential improvements, evaluate interventions, and

determine long-term progress in dietary change.

Researchers have used a variety of techniques to

study the consumption patterns of children. One of

the most commonly used measures is the 24-hour

recall.11,12 In their review of 11 studies using food re-

calls with children, McPherson et al13 found a wide

range in mean energy intake estimates from that ob-

tained from validation standards. They also emphasize

that recall is particularly difficult to use with younger

children. Several studies14,15 combined 24-hour diet

recallswith direct observation of cafeteria consumption

of a small sample of students. Other commonly used

methods include food records,16,17 surveys of student

eating habits,18,19 and food frequency surveys.20-23

Each of these methods poses challenges, particularly

whenusedwithyoung children;McPherson et al13 pro-

vide a detailed review of the validity and reliability of

the most commonly used techniques. In confined set-

tings such as school cafeterias, direct observation of

meals consumed offers an excellent alternative to

methods which rely on recall or record keeping and is

often used as the validation standard for studies of

school-aged children. However, observation techni-

ques have traditionally required the use of many

well-trained observers, and the labor requirements

have severely limited the number of observations that

can be made. Another common method of measuring

consumption in cafeteria settings is plate waste studies.

Although highly accurate, plate waste studies for large

samples also tend to be labor intensive and impractical,

requiring either the use of a large number of trained

observers or limiting the study to a relatively small

sample.24 The lack of valid, reliable, and practicalmeth-

ods to assess school lunch intake continues to challenge

researchers and food service administrators.

This pilot study examined the use of digital photog-

raphy to measure the consumption of lunches among

elementary students in the school cafeteria setting.

While some research has been conducted on the use of

digital photography in nutrition research among older

populations, relatively little is know about its applica-

tion to lunches of elementary students. Williamson

et al25,26 demonstrated that digital photography is a

highly accurate means of estimating cafeteria con-

sumption among college students, but the method has

not yet been tested among a younger population. The

college cafeteria study conducted by Williamson et al

validated this researchmethod, finding a .92 correlation

between visual assessment using digital photography

and food intake weight.26 This study sought to deter-

mine if potential problems unique to the school envi-

ronment, such as playing with food, would invalidate

the photography technique among children. Addition-

ally, this study utilized still digital cameras, which are

more readily available and affordable to elementary

and secondary school staff and parents than the digital

video cameras used by Williamson et al. Thus, the goal

of this study was to determine whether digital cameras

represented a valid, reliable, and practical dietary

assessment method for elementary school cafeteria

research.

In this study, digital photography of lunch trays

before and after student consumption allowed for

visual estimation of every meal served on each study

day by multiple observers. Because the data were pre-

served in a photograph, a third observer could resolve

any significant disagreements between the initial 2

observers. Additionally, since visual estimations were

made in the laboratory, rather than in the cafeteria,

observers could take as much time as they needed to

carefully consider their assessment of the amounts

taken and consumed by each student.

Journal of School Health d August 2008, Vol. 78, No. 8 d ª 2008, American School Health Association d 433

METHODS

SubjectsFour lunch periods at 2 different elementary schools

in a rural Kentucky school district were studied for this

research. Data were collected on the lunches served

and consumed by 100% of first- through fifth-grade

students at both elementary schools. Over 80% of the

children in this district qualify for free/reduced price

meals; because this rate is so high, the district offers

free meals to all students. As a result, participation in

the meal program is high; according to district data,

over 95% of elementary students typically eat the

school meal. On each of the 4 days on which we

conducted this research, every student eating in the

cafeteria participated in the meal program, creating

ideal conditions for testing the photographic method

of assessing cafeteria consumption. The numbers of

students eating varied because some classes were

absent from school for field trips on some data collec-

tion days. Meals offered to students on the data collec-

tion days are presented in Table 1. All data were

collected in the spring of 2006.

This research focused on lunches, rather than stu-

dents. At no time were individual students associated

with particular lunches, and no personal identifying

information about students was collected. No photo-

graphs of students were taken as part of this research.

The research was approved by the Institutional

Review Board of the University of Kentucky and the

Superintendent and Food Service Director of the

county school system.

ProcedureFour research assistants worked with the principal

investigator on each day of data collection. Two as-

sistants were responsible for taking pictures of the

‘‘before’’ trays, while the other 2 assistants worked

together to take the ‘‘after’’ pictures. Depending on

the meals served, research assistants occasionally

needed to make some adjustments to the ‘‘after’’

trays to make certain that all items were visible in

the photograph. For example, orange peelings some-

times needed to be separated from uneaten portions

of oranges on 1 of the days during this particular pro-

ject. A reference sample tray of food was taken from

the cafeteria each day, each item on the tray was

weighed, and the reference tray was photographed.

This weight was the standard used when estimating

the amounts offered to and consumed by students.

Prior to the beginning of the lunch period, pre-

printed unique numbers on cardstock were attached

to disposable lunch trays. Students had their trays

filled as usual, but upon leaving the serving line, they

placed their trays on a table where pictures were

taken of each tray using Canon PowerShot SD550

(Lake Success, NY) digital cameras fixed on tripods.

The cameras were set 16 inches above the trays at

approximately a 45° angle. Students then carried their

trays to the tables and ate their meals. Once the

research team became experienced in this procedure,

each tray took less than 5 seconds to situate and pho-

tograph. This is an important consideration, given the

large number of children who need to move through

a cafeteria line quickly in a school lunchroom.

After finishing their meals, students were in-

structed by teachers and cafeteria monitors to leave

their trays and all waste on the table. Research staff

then arranged items on the trays to ensure that all

items would be visible in photographs, and additional

SD550 cameras were used to take pictures of these

trays and any accompanying waste materials, such as

apple cores and orange peels. A total of 4 digital cam-

eras were used in this project (2 for before and 2 for

after pictures); the number of cameras required will

vary with the size of the student population being

studied. Next, the tray numbers were affixed to the

beverage containers, which were then set aside for

physical measurement of leftover quantities after the

end of the lunch period. Finally, research staff dis-

posed of the trays, clearing the tables for the next

group of students.

Data AnalysisAnalysis of the photographic data began with 2 an-

alysts comparing each before and after picture to visu-

ally estimate the percentage of each item consumed,

to the nearest 10% increment, by each student. Many

times, visual clues from the ‘‘after’’ picture were used

to more accurately estimate the amount of food actu-

ally taken by the student when students returned to

the serving line to purchase additional servings of

some items.

All data were entered into Microsoft Excel (Red-

mond, WA), and the estimate of percentage consumed

for each item by analyst 1 was subtracted from the

Table 1. Content of Meals Offered

School 1, Day 1 School 1, Day 2 School 2, Day 1 School 2, Day 2

Entree Tortilla chips, ground beef Chicken nuggets Chicken patty PizzaProduce Shredded lettuce, tomato salsa, apple Green beans, apple, orange Green beans Corn, grapes, lettuce saladSide dishes Cheese, refried beans Mashed potatoes, roll Mashed potatoes, gravy, roll Salad dressingDesserts None None Pudding Chocolate cake


estimate by analyst 2. Any differences in observations

greater than 50% were reevaluated by the principal

investigator. Generally, these large discrepancies rep-

resented data entry errors. The 2 closest of these 3 in-

dependent estimates were retained for analysis. Then,

the 2 estimates were averaged, resulting in an average

consumption estimate for each item on each tray.

To examine the reliability of this method of data

collection, the ratings of the 2 analysts of the photo-

graphs were compared for each food item. Missing

data were also assessed, including before or after tray

photographs which were not taken or which were not

clear enough to be interpretable and photographs in

which the quantities of individual food items con-

sumed could not be discerned.

RESULTS

A total of 859 lunches were served over the 4 days

of the study period. For 33 of these lunches (3.8%),

either the before consumption (n = 7) or after con-

sumption (n = 26) pictures were missing, resulting in

a final sample size of 826 lunches. The missing

‘‘before’’ pictures were those not taken when data

collection threatened to slow the cafeteria serving

line. The missing ‘‘after’’ pictures were due to stu-

dents disposing of lunches in their usual manner

before a member of the research team was able to take

a photograph.

Data ReliabilityTo assess the level of interrater reliability, it was

necessary to evaluate individual lunch items, rather

than entire meals. Because each tray had multiple

items on it, a total of 5394 entrees, fruits and vegeta-

bles, and side items was evaluated. The amount con-

sumed of each of these items was estimated, to the

nearest 10% increment, by 2 trained research assis-

tants. The 2 estimates of the amount consumed were

then compared, and 158 of these observations (2.9%)

had a difference of greater than 50%.

A comparison of the 2 raters’ estimates of the

amount of each item consumed demonstrated an

extremely high level of interrater reliability resulting

from this method of visual estimation. Of the 5394

total items, 92% (n = 4962) were rated within 10% of

each other by the 2 independent raters, 97% (n =5232) were rated within 20%, and 99% (n = 5340)

were rated within 30%. Table 2 describes the level of

agreement disaggregated by the type of food. The

greatest similarity between raters was found for pro-

duce items, which were commonly served in small

plastic cups, thus providing a clear reference guide for

making visual estimations. The lowest level of inter-

rater reliability was found for main entrees; most nota-

ble among these being the ground beef and tortilla

chips. Children typically mixed these items, making

visual estimation more difficult and less consistent.

A concern about using this method with elemen-

tary school–aged children was the potential propen-

sity of students to play with their food in a manner

that would make visual estimation of consumption

difficult or impossible. While such cases did occur in

our sample, they proved to be rare; only 13 of the

5394 items (0.2%) were coded by analysts as unable

to be interpreted.

An initially puzzling finding in the photographs

was the frequent absence of evidence of consumed

apples and oranges, in the form of apple cores and

orange peelings, in the ‘‘after’’ pictures. On 2 separate

days, whole apples were served, and on 1 of those

days, whole unpeeled oranges were also offered to

students. While a total of 153 apples and 42 oranges

were served to students, there was no evidence of 45

of the apples (29%) and 10 (24%) of the oranges in

the ‘‘after’’ photographs. Our interpretation of this

missing data, confirmed by conversations with cafete-

ria staff, was that students were taking whole fruit out

of the cafeteria for consumption at a later time. In

fact, cafeteria workers and teachers encouraged stu-

dents to take apples and oranges with them to eat on

the bus ride home after school. While this creates

obvious problems for analysis of consumption, it may

well increase overall produce consumption.

DISCUSSION

This study demonstrates the utility of digital pho-

tography in measuring the consumption levels of ele-

mentary students in school cafeterias. This method

offers researchers the ability to collect highly detailed,

low cost, and accurate data on consumption, while

imposing a minimal burden on cafeteria operations.

Unlike traditional methods of nutritional research,

this form of observational study does not rely upon

recall, an important consideration when dealing with

young children.

The method described here can potentially be uti-

lized both by academic researchers studying the

nutritional implications of cafeteria menu options

and by school food service personnel interested in

testing the effectiveness of strategies to increase the

consumption of particular components (eg, fruits

Table 2. Interrater Reliability by Food Type

Number ofServings

% Items Rated by 2 Independent Raters

Within 10% Within 20% Within 30%

Entrees 1067 87 95 99Produce 2020 94 98 99Side dishes 1772 91 97 99Dessert 535 91 97 99


and vegetables) of school meals. Using digital pho-

tography to study cafeteria meals is not particularly

complex or burdensome and could easily be carried

out by parent volunteers under the coordination of

a food service director, particularly with some guid-

ance from academic researchers. Many parents and

schools already have digital cameras that could be

used in these efforts. Such a team could explore such

issues as food preferences of students, the most effec-

tive means of presentation of fruits and vegetables to

promote student consumption, and the effects of

nutritional promotion and education efforts, among

other subjects.

The use of this method is limited to school-served

meals and would not work well with meals brought

by students from home because of the difficulty in ob-

taining ‘‘before’’ pictures and the lack of standardized

serving sizes. Additionally, certain types of items

served in school meals are not amenable to measure-

ment via visual estimation through photographs. In

this research, analysts found it impossible to accu-

rately estimate consumption of condiments served in

individual packets, including ketchup, salad dressing,

dipping sauce for chicken nuggets, and spread for

rolls. As a result, such items had to be excluded from

the analysis of these meals. Occasional a la carte items

such as bagged chips, although not appearing in this

sample, would also be difficult to evaluate without

removing any leftover amounts from the bag prior to

photography. It is important to plan prior to data col-

lection how such items will be accurately photo-

graphed in the ‘‘after’’ pictures.

Analysis of the data obtained through this method

will be determined by the goals of the particular re-

search project. Simple counts of produce servings con-

sumed can be conducted rapidly and inexpensively,

while more complex nutritional analyses of the data

collected through digital photography, for example,

could be conducted using such computerized analysis

programs as the University of Minnesota’s Nutrition

Data System for Research.27 And while this pilot study

did not connect personal identifiers to lunch trays,

it would not be difficult to design research that links

individual student characteristics (age, gender, and

BMI) to consumption patterns.

There are several limitations to this study to be

considered. First, the study did not use weighed por-

tions of food to test the validity of visual assessment.

While Williamson et al26 have provided validation of

the digital photographic method using comparisons to

actual weights of food waste, inclusion of this step

would have further validated the method for use in

the elementary school setting. Additionally, the reli-

ability of visual estimation varies according to types of

food; researchers utilizing this method should pay spe-

cial attention to reliability concerns depending on the

specific foods being served.

Improving school consumption habits is 1 important

strategy in ongoing efforts to improve school health in

the United States. Rather than maintaining the tradi-

tional focus of school food service on meal offerings or

menus, this project developed and tested a method that

assessed actual consumption. An improved and more

accurate understanding of student consumption will

give researchers and administrators the information

needed to design healthier meals that students will not

just place on their trays but will actually eat. Digital

photography offers researchers and school food service

administrators an important tool to collect this needed

information in a resource- and time-effective manner.

REFERENCES

1. Kolbe LJ. A framework for school health programs in the 21st

century. J Sch Health. 2005;75:226-228.

2. Ogden CL, Flegal KM, Carroll MD, Johnson CL. Prevalence and

trends in overweight among US children and adolescents, 1999-

2000. JAMA. 2002;288:1728-1732.

3. Food andNutrition Service. National school lunchmonthly data.

Available at: http://www.fns.usda.gov/pd/slsummar.htm. Ac-

cessed April 7, 2007.

4. US Department of Health and Human Services. Healthy People

2010. Washington, DC: US Government Printing Office; 2000.

5. French SA, StoryM, Fulkerson JA, Hannan P. An environmental

intervention to promote lower-fat food choices in secondary

schools: outcomes of the TACOS study. Am J Public Health.

2004;94:1507-1512.

6. Adair LS, PopkinBM.Are child eatingpatternsbeing transformed

globally? Obesity Res. 2005;13:1281-1299.

7. French SA. Public health strategies for dietary change: schools

and workplaces. J Nutr. 2005;135:910-912.

8. Wechsler H, Devereaux RS, Davis M, Collins J. Using the school

environment to promote physical activity and healthy eating.

Prev Med. 2000;31:S121-S137.

9. Folta SC, Goldberg JP, Economos C, Bell R, Landers S, Hyatt R.

Assessing the use of school public address systems to deliver

nutrition messages to children: shape up Somerville—audio ad-

ventures. J Sch Health. 2006;76:459-464.

10. Georgiou C, Martin L, Long R. What third graders select

and eat from school luncheswhen they have choices. J Child Nutr

& Manag [serial online]. 2005;29(2). Available from: http://docs.

schoolnutrition.org/newsroom/jcnm/05fall/georgiou.index.asp.

Accessed June 2, 2008.

11. Luepker RV, Perry CL, McKinlay SM, et al. Outcomes of a field

trial to improve children’s dietary patterns and physical activity.

The Child and Adolescent Trial for Cardiovascular Health.

CATCH collaborative group. JAMA. 1996;275:768-776.

12. Reynolds KD, Baranowski T, Bishop DB, et al. Patterns in child

and adolescent consumption of fruit and vegetables: effects of

gender and ethnicity across four sites. J Am Coll Nutr. 1999;18:

248-254.

13. McPherson RS, Hoelscher DM, Alexander M, Scanlon KS,

Serdula MK. Dietary assessment methods among school-aged

children: validity and reliability. Prev Med. 2000;31:S11-S33.

14. Perry CL, Bishop DB, Taylor G, et al. Changing fruit and vege-

table consumption among children: the 5-a-Day Power Plus

program in St. Paul, Minnesota. Am J Public Health. 1998;88:

603-609.

15. Reynolds KD, Franklin FA, Binkley D, et al. Increasing the fruit

and vegetable consumption of fourth-graders: results from the

High 5 Project. Prev Med. 2000;30:309-319.


16. Cullen KW, Eagan J, Baranowski T, Owens E, de Moor C. Effect

of a la carte and snack bar foods at school on children’s lunch-

time intake of fruits and vegetables. J Am Diet Assoc. 2000;100:

1482-1486.

17. Resnicow K, Davis-Hearn M, Smith M, et al. Social-cognitive

predictors of fruit and vegetable intake in children.Health Psychol-

ogy. 1997;16:272-276.

18. Neumark-Sztainer D, French S, Hannan P, StoryM, Fulkerson J.

School lunch and snacking patterns among high school students:

associations with school food environment and policies. Int J Be-

havNutr Phys Act [serial online]. 2005;2:14.Available from: Inter-

national Society for Behavioral Nutrition and Physical Activity,

http://www.isbnpa.org/content/2/1/14. Accessed April 7, 2007.

19. Stevens J, Cornell CE, StoryM, et al. Development of a question-

naire to assess knowledge, attitudes, and behaviors in American

Indian children. Am J Clin Nutr. 1999;69:773S-781S.

20. Mangunkusumo RT, Duisterhout JS, de Graaff N, Maarsingh EJ,

de Koning HJ, Raat H. Internet versus paper mode of health and

health behavior questionnaires in elementary schools: asthma

and fruit as examples. J Sch Health. 2006;76:80-86.

21. Bere E, Klepp K-I. Changes in accessibility and preferences pre-

dict children’s future fruit and vegetable intake. Int J Behav Nutr

Phys Act [serial online]. 2005;2:15. Available from: International

Society for Behavioral Nutrition and Physical Activity, http://

www.isbnpa.org/content/2/1/15. Accessed April 7, 2007.

22. Neumark-SztainerD,WallM,PerryC, StoryM.Correlatesof fruit

and vegetable intake among adolescents: findings from Project

EAT. Prev Med. 2003;37:198-208.

23. Perry CL, Bishop DB, Taylor GL, et al. A randomized school trial

of environmental strategies to encourage fruit and vegetable

consumption among children. Health Educ Behav. 2004;31:

65-76.

24. Comstock E, St. Pierre R, Mackiernan Y. Measuring individual

plate waste in school lunches. Visual estimation and children’s

ratings vs. actual weighing of plate waste. J Am Diet Assoc. 1981;

79:290-296.

25. Williamson DA, Allen HR, Martin PD, Alfonso A, Gerald B, Hunt

A. Digital photography: a newmethod for estimating food intake

in cafeteria settings. Eat Weight Disord. 2004;9:24-28.

26. Williamson DA, Allen R, Martin PD, Alfonso AJ, Gerald B,

Hunt A. Comparison of digital photography to weighed and

visual estimation of portion sizes. J Am Diet Assoc. 2003;103:

1139-1145.

27. Nutrition Coordinating Center. Nutrition Data System for

Research. Minneapolis, Minn: Nutrition Coordinating Center,

University of Minnesota; 2006.


Documents

Digital Photography as a Tool to Measure School Cafeteria Consumption