smarter lunchrooms randomized control trial: results from ... advance/journals/jneb/20… · the...
TRANSCRIPT
Project Overview
The Smarter Lunchroom Movement utilizes principles of
behavioral economics to make small, low or no-cost
environmental changes in school lunchrooms to improve the
selection of healthful foods by students.
The current research project aims to build upon findings
from several successful pilot studies, and test these
environmental changes on a larger scale in several urban
and rural public middle school lunchrooms in New York.
• Year 2: Test single lunchroom changes
• Year 3: Test combination lunchroom changes
In Year 3, the objective was to determine the impact of
combining multiple Smarter Lunchroom interventions on the
selection and consumption of healthy food items in the
school lunchroom. This design was informed by the positive
results of single Smarter Lunchroom protocols in Year 2.
Year 3: Intervention Overview
• 11* middle schools from urban and rural districts in
Western New York participated during the 2014-2015
school year.
• Fall 2014: 5 schools (2 urban, 3 rural) received the
intervention, and 6 schools (3 urban, 3 rural) were control
schools.
• Spring 2015: control schools became treatment schools,
and treatment schools became maintenance schools.
• Maintenance schools were observed to see if they
continued implementing Smarter Lunchrooms protocols,
after the intervention period was completed.
• The intervention combined Smarter Lunchroom protocols
targeted at increasing the convenience, visibility, and
attractiveness of fruits, vegetables, and white milk.
• Combined intervention included (See Figure 1):
1. Promotion of fruits: locating fruit first & in multiple
locations, using large fruit bowl & attractive serving cups
for fruits, labeling fruit with creative names, and
displaying fruit factoid posters.
2. Promotion of vegetables: locating vegetables after
entrée & at multiple locations, using attractive serving
cups for vegetables, labeling vegetables with creative
names, and displaying vegetable factoid posters.
3. Promotion of low-fat white milk: labeling white milk,
displaying low-fat white milk in front of flavored milk, and
as 1/3 of visible beverages.
* One urban school was eliminated due to inability to
implement intervention.
Acknowledgements
This project was supported by Agriculture and
Food Research Initiative Grant no. 2012-68001-
19604 from the USDA National Institute of Food and
Agriculture, Childhood Obesity Prevention:
Integrated Research, Education, and Extension to
Prevent Childhood Obesity – A2101.
For more information, visit:
http://www.smarterlunchrooms.org
Smarter Lunchrooms Randomized Control Trial: Results from Year 3Katherine Greene MPH, 1 Gnel Gabrielyan Ph.D. 1 Adam Brumberg, 1
Jamie Dollahite Ph.D., 2 David Just Ph.D., 1 Brian Wansink Ph.D., 1
1 Cornell University, Charles H. Dyson School of Applied Economics and Management 2 Cornell University, Division of Nutritional Science
Year 3: Results
The results below are limited to schools from the Fall
2014 intervention group (n=5). To evaluate
sustainability, we compared Fall 2014 pre intervention
data with Spring 2015 post-intervention data.
• Fruit selection increased overall by 11% (p<0.05),
mainly due to a 33% (p<0.001) increase in urban
schools.
• Fruit consumption decreased overall by 6%
(p<0.001), primarily due to a 7% (p<0.001) decrease
in rural schools.
• Vegetable selection decreased overall by 27%
(p<0.001).
• Vegetable consumption decreased by 4% (p<0.05)
in all schools, mainly due to an 8% (p<0. 001) in rural
schools.
Methods
Quarter Waste Method of Visual Estimation: measures the
percent of each item that was wasted on a tray. Researchers
visited the cafeterias pre- and during-intervention to record
the selection and waste of students who purchased a school
lunch. Researchers observed whether 0%, 25%, 50%, 75%,
or 100% of each food item was left on the tray.
The pre-intervention was a period of 3 weeks, and the
intervention was a period of 6-weeks. Schools switched
conditions at the midpoint of the year.
Year 4: Intervention Overview
Based on process evaluation findings of varying
fidelity, a new research objective was identified to
test a.) fewer protocol items, and b.) food service
ownership of intervention.
• 18 New York middle schools were recruited
during the Spring 2016 semester.
• Schools were allocated into 3 groups: selection,
matched/ assigned, and control
1. Selection: created their own intervention
from a list of protocols
2. Matched/ assigned: matched with selection
school based on practices, and assigned
same protocols.
3. Control: made no changes to existing set up.
• Intervention was completed May 2016.
School Location Group* White Black Hispanic OtherEconomic
Disadvantage**
Enrollment
(Grades 5-8)
1 Rural Control 42% 9% 46% 3% 77% 465
2 Rural Control 76% 5% 10% 9% 62% 509
3 Rural Control 94% 1% 3% 1% 54% 271
4 Rural Treatment 64% 3% 25% 8% 73% 501
5 Rural Treatment 84% 2% 10% 4% 39% 450
6 Rural Treatment 59% 4% 22% 15% 75% 505
7 Urban Control 2% 90% 5% 3% 80% 344
8 Urban Control 5% 13% 78% 4% 88% 337
9 Urban Control 13% 73% 11% 3% 85% 285
10 Urban Treatment 28% 47% 15% 9% 78% 297
11 Urban Treatment 14% 39% 20% 27% 91% 385
Figure 1: Intervention Protocol Diagram
Lunch Line Progression
!"#$%' ( ) * + -1 3-1?>25*3
@*>' & *5*A>*B*>1' 1., 3*7.1
/27 *21->5 1** .0* +(, -. ' 7 3-1?>25
Fruit is
displayed first
on the line.
Monkey Fuel Fantastic
Fruit
Weekly Fruit Factoid
Creative names
displayed on school lunch
menu/ daily menu board.
Broccoli
Bites
Weekly Vegetable Factoid
Large fruit bowl
is displayed at
eye-level
Creative name cards are
displayed next to fruit,
vegetable and white milk
options.
At least 2 fruit options and 2
vegetable options displayed
in 2 separate locations.
Vegetable
displayed after
the entrée.
Ice Cold White Milk
White milk makes up at
least 1/3 of all visible
beverages in each cooler
Laminated Fruit and
Vegetable Factoids
displayed in a visible
location.
X-Ray Vision
Carrots
& Hummus
Today’s Menu:
• Turkey Burger
• Wheat Bun
• Fantastic Fruit
• Monkey Fuel
• X-Ray Vision
Carrots & Hummus
• Broccoli Bites
• Ice Cold White Milk
Year 3 Demographic and Enrollment Data of School Sample
0
0.2
0.4
0.6
0.8
1
1.2
All Urban Rural All Urban Rural
Selection Consumption
Aver
age
Am
ou
nt
of
Fru
it I
tem
s S
elec
ted
or
Con
sum
ed
Fruit Selection & Consumption
Fall 2014 Intervention Schools
Fall Pre
Fall Post
Spring Pre
Spring Post
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
All Urban Rural All Urban Rural
Selection Consumption
Aver
age
Am
ou
nt
of
Veg
etab
le I
tem
s S
elec
ted
or
Con
sum
edVegetable Selection & Consumption
Fall 2014 Intervention Schools
Fall Pre
Fall Post
Spring Pre
Spring Post
Year 3: Discussion
It is possible that combination interventions may not
be as effective as single interventions (i.e. fruit or
vegetable only), which previously had demonstrated
positive results (Greene et al., 2016; Gabrielyan et
al. 2016). Additionally, the amount of protocol items
may have been too difficult for staff to implement.
Additional analysis, including diff-in-diff and group
mean t-tests will be performed to better understand
the impact over multiple groups and periods of time.
Lastly, process data will be incorporated into the
analysis, to determine impacts of fidelity.