Table 2 Summary of school meal nudge studies (n = 29) characteristics and outcomes

Adams et al., 2005 [23]

Cross-Sectional Study

Study Length: 1 day

Location: California

Age Group: Grades 1–5

n = NR, 4 schools

Quality Rating: Moderate

− Compared schools with existing salad bars to schools without salad bars

(Duration NR)

Selection:

Direct Weighing Individual

(1 day)

⊘ No significant difference in selection of fruits and vegetables at schools with and without salad bars (NS)

Consumption:

Direct Weighing Individual

(1 day)

⊘ No significant difference in consumption of fruits and vegetables at schools with and without salad bars (NS)

Adams et al., 2016 [29]

Cross-Sectional Study

Study Length: 1 day

Location: Arizona

Age Group: Middle School

n = 533 students, 6 schools

Quality Rating: Moderate

− Changing placement of salad bar to inside serving line, instead of outside the serving line

(Duration NR)

Selection:

Direct Weighing Individual

(6 days)

↑ Students self-served 5.38 times (95% CI: 4.04–7.17) more fresh fruits and vegetables by weight when the salad bar was inside the lunch line (141.2 g) compared to outside (26.6 g)

Consumption:

Direct Weighing Individual

(6 days)

↑ Students were 4.82 times (95% CI: 3.40–6.81) more likely to consume more fresh fruits and vegetables when the salad bar was inside the lunch line (74.1 g) compared to outside (16.6 g)

Waste:

Direct Weighing Individual

(6 days)

↑ Students wasted more fruits and vegetables when the salad bar was inside the lunch line (43%, 29.9 g) compared to outside (12%, 11.7 g) (p = NR)

Cohen et al., 2015 [30]

Cluster Randomized Trial

Study Length: 10 months

Location: Massachusetts

Age Group: Grades 1–8

n = 1587 students, 10 schools

Quality Rating: Moderate

− Using attractive bowls or baskets

− Signage and images promoting fruits and vegetables

− Changing fruit and vegetable placement

(All components 4 months)

Selection:

Visual Observation

(6 days)

↑ Fruit selection was greater at intervention schools (54.3%) than control schools (51.1%) (OR: 1.45, 95% CI: 1.13–1.87)

↑ Vegetable selection was greater at intervention schools (33.7%) than control schools (50.6%) (OR: 1.91, 95% CI: 1.46–2.50)

⊘ No significant difference in milk selection between intervention and control schools (NS)

Consumption:

Direct Weighing Individual

(6 days)

⊘ No significant difference in fruit consumption between intervention and control schools (NS)

⊘ No significant difference in vegetable consumption between intervention and control schools (NS)

⊘ No significant difference in milk consumption between intervention and control schools (NS)

Conklin et al., 2005 [31]

Non-Randomized Controlled Trial

Study Length: 12 weeks

Location: Pennsylvania

Age Group: Grades 9–12

n = NR, 6 schools

Quality Rating: Moderate

− Nutrition Facts labeling at point of sale for entrées

(6 weeks)

Selection:

Sales Data

(12 weeks)

↑ Decrease in the number of servings selected for pepperoni pizza, cheeseburgers, bacon cheeseburgers, and chicken dishes with > 20 g of fat per serving (p < .05)

↑ Increase in the number of servings selected for cheese pizza, hamburgers, and veggie burgers (p < .05)

Delaney et al., 2017 [32]

Cluster Randomized Trial

Study Length: 4 months

Location: Australia

Age Group: Grades K – 6

n = 2714 students, 10 schools

Quality Rating: Moderate

− Increasing the number of food choices/variety

− Creative, descriptive names to market food items (non-student developed)

− Online pre-orders

− Stop Light Labeling

− Positive prompting and reinforcement for healthy choices

(All components 2 months)

Selection:

Sales Data

(4 months)

↓ Fewer calories in selected lunches among intervention (1104 kJ) compared to control (1565 kJ) participants at post-intervention (p < .001)

↓ Less saturated fat in selected lunches among intervention (2.87 g) compared to control (4.84 g) participants at post-intervention (p < .001)

↓ Less sodium in selected lunches among intervention (400 mg) compared to control (527 mg) participants at post-intervention (p < .001)

⊘ No significant difference in sugar content of selected lunches between intervention and control at post-intervention (NS)

Elbel et al., 2015 [33]

Non-Randomized Controlled Trial

Study Length: 10 months

Location: New York

Age Group: Grades K – 12

n = 19,000 students, 19 schools

Quality Rating: Moderate

− Introduction of water jet

(Approximately 10 months)

Selection:

Visual Observation

(4 days)

↑ Three months after implementation, there was a larger increase in water selection among intervention (+ 23%) schools compared to control schools (+ 2%) (p < .001)

↑ Ten months after implementation, there was a larger increase in water selection among intervention (+ 22%) schools compared to control schools (− 2%) (p < .001)

Elsbernd et al., 2016 [34]

Crossover Study Design

Study Length: 12 weeks

Location: Minnesota

Age Group: Grades K – 5

n = 800 students, 4 schools

Quality Rating: Moderate

− Verbal prompts for healthy item selection and/or consumption

− Serving vegetables first in isolation

(All components 3 days)

Selection:

Sales Data

(5 days)

↑ Selection of peppers was greater on the intervention days (65%) when vegetables were served first in isolation compared to control days (8%) (p < .0001)

Consumption:

Visual Observation

(5 days)

↑ Consumption of peppers was greater on the intervention days (4.1 g) when vegetables were served first in isolation compared to control days (mean 1.4 g) (p < .0001)

Waste:

Visual Observation

(5 days)

↑ Waste of peppers was greater on the intervention days (53–64%) when vegetables were served first in isolation compared to control days (8–38%) (p = NR)

Ensaff et al., 2015 [35]

Non-Randomized Controlled Trial

Study Length: 2 academic years

Location: United Kingdom

Age Group: Middle & High School

n = 2200 students, 2 schools

Quality Rating: Strong

− Changing food placement

− Making plant-based foods more convenient (disposable pots/trays used to serve meals, prefilled pots/trays)

− Promotional material (smiley stickers on packaging and posters, end of shelf label, encouraging posters)

(All components 6 weeks)

Selection:

Sales Data

(378 days)

↑ Students were 2.5 times more likely (OR: 2.49, 95% CI: 2.03–3.06) to select the designated items (whole fruit, fruit salad, vegetarian specials, and sandwiches containing salad) during the intervention period than at baseline (p < .001)

↑ Students were 3 times as likely to choose a fruit, vegetable, or salad item during the intervention relative to baseline (OR: 3.04, 95% CI: 2.50–3.69) (p < .001)

Goto et al., 2013 [36]

Cluster Randomized Trial

Study Length: 1 month (approx.)

Location: California

Age Group: Grades 1–6

n = 677 students, 3 schools

Quality Rating: Moderate

− Treatment 1: Increased white milk quantity available in the milk cooler compared to chocolate milk

− Treatment 2: Students must request chocolate milk

(Both conditions 1 week)

Selection:

Visual Observation

(10 days)

↑ At treatment 1 school (increased milk quantity), selection of white milk increased by 18% between pre- (30%, 74 students) to post-intervention (48%, 118 students) (p < .001)

⊘ No significant change in selection of white milk in treatment 2 school (ask for chocolate) (NS)

Consumption:

Direct Weighing Individual

(30 days)

⊘ No significant change in white milk consumption at either treatment school (NS)

Greene et al., 2017 [37]

Cluster Randomized Trial

Study Length: 9 weeks

Location: New York

Age Group: Grades 5–8

n = 2108 students, 10 schools

Quality Rating: Moderate

− Fruit placed first on lunch line

− Served two fruit options instead of one

− Slicing fruit

− Using attractive bowls for whole fruits

− Creative, descriptive names to market fruit items (non-student developed)

− Fruit “factoids” (promotions)

(All components 6 weeks)

Selection:

Visual Observation

(Average of 5 baseline and 4 intervention days per school)

↑ Increase in average fruit selection from pre- (.59 servings) to post-intervention (.80 servings) (p < .001)

↑ Increase in average vegetable selection from pre- (.67 servings) to post-intervention (.98 servings) (p < .001)

↑ Increase in average white milk selection from pre- (.10 servings) to post-intervention (.14 servings) (p < .001)

Consumption:

Visual Observation

(Average of 5 baseline and 4 intervention days per school)

↑Increase in average fruit consumption from pre- (.73 servings) to post-intervention (.83 servings) (p < .001)

↑ Increase in average vegetable consumption from pre- (.57 servings) to post-intervention (.86 servings) (p < .001)

⊘ No change in average white milk consumption from pre- to post-intervention (NS)

Hakim et al., 2013 [38]

Non-Controlled Trial

Study Length: 2 weeks (approx.)

Location: Kansas

Age Group: Grades K – 8

n = 2064 students, 1 school

Quality Rating: Weak

− Increasing the number of fruit and vegetable choices and variety offered each day

(1 month)

Consumption:

Direct Weighing Individual

(20 days)

↑ Fruit consumption increased between pre-intervention (40%) and post-intervention (67%) (p < .01)

↑ Vegetable consumption increased between pre-intervention (23%) and post-intervention (41%) (p < .01)

Hanks et al., 2012 [39]

Before-After Study

Study Length: 3 months

Location: New York

Age Group: High School

n = 362 students, 1 school

Quality Rating: Weak

− Created a convenience line in which only options considered healthy (sub sandwich bar, salad bar, vegetables, whole fruit, fruit parfait, flavored milks) were offered

(2 months)

Selection:

Self-Reported Survey with no validity or reliability information provided

(4 days)

↑ Increase in selection of healthy foods (salad bar, vegetables, fruit, fruit parfaits, and sub sandwiches) from pre- (mean .66 items) to post-intervention (mean .79 items) (p < .001)

↑ Selection of flavored milk increased from pre- (mean 0.74 items) to post-intervention (mean 0.85 items) (p < .001)

⊘ No significant change in plain milk selection (NS)

Consumption:

Direct Weighing Individual

(4 days)

↓ Consumption of less healthy foods decreased by 27.9% (p < .001)

Hanks et al., 2013 [40]

Before-After Study

Study Length: 4 months

Location: New York

Age Group: Grades 7–12

n = NR, 2 schools

Quality Rating: Weak

− Using attractive bowls or baskets

− Changing food placement

− Creative, descriptive names to market food items (non-student developed)

− Verbal prompts to promote healthy items

− Creating healthy convenience line with only submarine sandwiches, fruit and vegetable sides

(All components 2 months)

Selection:

Sales Data

(12 days)

↑ Students selecting fruits increased from pre- (47.3%) to post-intervention (53.7%) (p = .012)

↑ Students selecting vegetables increased from pre- (35.8%) to post-intervention (44.0%) (p < .001)

Hanks et al., 2016 [41]

Cluster Randomized Trial

Study Length: 6 weeks

Location: New York

Age Group: Elementary School

n = 1133 students, 10 schools

Quality Rating: Moderate

− Treatment 1: Branded vegetable characters featured on vinyl promotional banners

− Treatment 2: Television promotional segments

− Treatment 3: Branded vegetable characters featured on vinyl promotional banners and television promotional segments

(All conditions 4 weeks)

Selection:

Sales Data

(6 weeks)

↑ Increase in average daily vegetable and salad servings selected from pre- (60 servings) to post-intervention (185 servings) in treatment 3 schools (p = .028)

⊘ No significant changes from pre- to post-intervention in vegetable and salad servings selected in treatment 1 and treatment 2 schools (NS)

Hunsberger et al., 2014 [42]

Non-Controlled Trial

Study Length: 2 months

Location: Oregon

Age Group: Grades 6–8

n = 531 students, 1 school

Quality Rating: Weak

− Calorie labels at point of purchase

(1 month)

Consumption:

Direct Weighing Aggregate

(Number of days NR)

↓ Decrease in calories consumed between pre- (668 Kcal) and post-intervention (621 Kcal) (p = .0040)

↓ Decrease in fat consumed between pre- (23.1 g) and post-intervention (21.1 g) (p = .0025)

Johnson et al., 2017 [43]

Cross-Sectional Study

Study Length: NR

Location: Louisiana

Age Group: Grades 7–12

n = 718 students, 21 schools

Quality Rating: Weak

− Compared schools with existing salad bars to schools without salad bars

(Duration NR)

Consumption:

Self-Administered 24-Hour Dietary Recall with no validity or reliability information provided

(Number of days NR)

↑ Students in salad bar schools had significantly higher median energy consumption during lunch (452 kcal) than students in non-salad bar schools (395 kcal) (p = .0136)

↓ Less fruit was consumed in salad bar schools (.06 cups) than non-salad bar schools (0.25 cups) (p = .0061)

⊘ No significant differences in vegetable consumption (NS)

Kenney et al., 2015 [44]

Cluster Randomized Trial

Study Length: 2 months

Location: Massachusetts

Age Group: Elementary, Middle, and High School

n = NR, 10 schools

Quality Rating: Moderate

− Promotional signage highlighting water source locations and promoting consumption

(16–23 days)

− Cup dispensers located next to cafeteria water fountains

(5–22 days)

Selection:

Visual Observation

(95 days)

↑ Larger increase from pre- to post-intervention in students who take free water during lunch period among intervention (+ 7.3%) compared to control (− 2.0%) (p < .001)

Consumption:

Visual Observation

(95 days)

↑ Larger increase from pre- to post-intervention in water consumption among intervention (+ 0.53 oz) compared to control (− 0.06 oz) (p < .001)

Larson et al., 2018 [45]

Cluster Randomized Trial

Study Length: 3 academic years

Location: Minnesota

Age Group: Grades 9–12

n = 364 students, 8 schools

Quality Rating: Strong

− Implementing a grab-and-go breakfast cart before school

(1 year intervention)

Participation:

Administrative Data

(2 years baseline and 1 year intervention data)

↑ Participation in school breakfast program increased from 13.0% in the first baseline year (T1) to 22.6% in the intervention year (T3) (p = 0.03)

Miller et al., 2015 [46]

Crossover Study Design

Study Length: 4 months (approx.)

Location: Minnesota

Age Group: Grades K – 5

n = 758 students, 1 school

Quality Rating: Moderate

− Provided increased portions sizes for orange wedges, baby carrots, greens beans, and applesauce as the default portion size

(Duration NR)

Selection:

Direct Weighing Individual

(3 days)

↑ Selection of oranges was higher on intervention days (51%) compared to control days (35%) (p < .0001)

↓ Selection of applesauce was lower on intervention days (36%) compared to control days (46%) (p < .0001)

⊘ No significant difference in green bean selection (NS)

Consumption:

Direct Weighing Individual

(3 days)

↑ Consumption of applesauce was greater on intervention days (119 g) compared to control (77 g) (p < .001)

↑ Consumption of orange slices was greater on intervention days (37 g) compared to control (22 g) (p < .001)

↑ Consumption of carrots was greater on intervention days (33 g) compared to control (20 g) (p = .02)

⊘ No significant difference in green bean consumption

Waste:

Direct Weighing Individual

(3 days)

↑ Waste of fruits and vegetables was 11–125% greater on intervention days compared to control days (p = NR)

Miller et al., 2016 [47]

Non-Randomized Controlled Trial

Study Length: 4 weeks

Location: Florida

Age Group: Grades 4–7

n = 169 students, 1 combined elementary and middle school

Quality Rating: Moderate

− Treatment 1: Pre-orders

− Treatment 2: Pre-orders with behavioral nudge (message alert if the student did not select a complete meal)

(All conditions 2 weeks)

Selection:

Video Recorded

Observation

(6 weeks)

↑ Increases in selection of vegetables from pre- to post-intervention were significant for treatment 1 (pre-orders only, + 17.6%, p < .001) and treatment 2 (pre-order with behavioral nudge, + 24.6%, p < .001), but not control (+ 5.4%, p = NS)

↑ Increases in selection of fruits from pre- to post-intervention were significant for treatment 1 (pre-orders only, + 23.0%, p < .001) and treatment 2 schools (pre-order with behavioral nudge, + 37.8%, p < .001) but not control (+ 4.8%, p = NS)

↑ Increases in selection of low-fat milk from pre- to post-intervention were significant for treatment 1 (+ 7.3%, p = .011), treatment 2 (+ 12.2%, p < .001), and control (+ 20.7%, p = .011)

Moreno-Black et al., 2017 [48]

Repeated Cross-Sectional (Longitudinal)

Study Length: 10 months (approx.)

Location: Oregon

Age Group: Grades 1–5

n = 3000 students, 7 schools

Quality Rating: Moderate

− Fruit and vegetable placement on salad bar vs hot line

(Spring and Fall semester)

Selection:

Digital Photography

(Number of days NR)

↑ Students were more likely to select fruit if when it was placed on the hot lunch line (p < .05)

↑ Students were more likely to select any vegetable when fruit was placed on the hot lunch line (p < .001)

↑ Students were more likely to select at least one healthy vegetable item when both fruit (p < .001) and vegetables (p < .05) were placed on the hot lunch line

Morizet et al., 2012 [49]

Non-Randomized Controlled Trial

Study Length: 2 days

Location: France

Age Group: Elementary and Middle School

n = 227 students, 3 schools

Quality Rating: Weak

− Treatment 1: Basic descriptive label (i.e., new carrot recipe)

− Treatment 2: Model-related label (i.e., new carrot recipe, Special Mix for Super Heroes)

− Control Group: No labels

(All conditions 2 days)

Selection:

Visual Observation

(6 days)

↑ Selection of new vegetable dish was greater for treatment 1 (p = .012) and treatment 2 (p = .002) compared to control

⊘ No significant difference between treatment 1 and treatment 2 in selection of familiar and new dishes (NS)

Redden et al., 2015 [50]

Crossover Study Design

Study Length: 3 months

Location: USA

Age Group: Grades K – 5

n = 755 students, 1 school

Quality Rating: Strong

− Offering youth vegetables in a small cup prior to moving through the lunch line, instead of only serving vegetables through the lunch line

(Field Study: 1 control day, 1 intervention day, 3 months apart)

(Longitudinal: 1 control/baseline day, 3 intervention days, 1 control day, over 3 month period)

Selection:

Visual Observation

(Field Study: 2 days, Longitudinal: 5 days)

⊘ Field Study: Likelihood of selecting carrots from lunch line did not differ between treatment and control days (NS)

↓ Longitudinal: Percentage of students selecting broccoli from lunch line was higher on the baseline control day (13.8%) than the first treatment day (4.1%, p < .0001), the second (9.4%, p < .01), and the third treatment day (5.3%, p < .0001)

↓ Longitudinal: Percentage of students selecting broccoli from lunch line higher on final control day (8.5%) than first (4.1%, p < .01) and third (5.3%, p < .01), but not the second treatment day (9.4%, NS)

Consumption:

Direct Weighing Individual

(Field Study: 2 days, Longitudinal: 5 days)

↑ Field Study: Students consumed 10.3 g more carrots total on treatment day (12.7 g) than control day (2.4 g) (p < .0001)

↑ Longitudinal: Students consumed more broccoli on the first treatment day (3.99 g), second treatment day (4.06 g) and the third treatment day (2.10 g) than both baseline control (0.84 g) and final control (0.90 g) days (p < .0001 for all treatment-control comparisons)

Schwartz et al., 2007 [51]

Cluster Randomized Trial

Study Length: 2 months (approx.)

Location: Connecticut

Age Group: Grades 1–4

n = 323 students, 2 schools

Quality Rating: Moderate

− Verbal prompts promoting healthy items

(Duration NR)

Selection:

Visual Observation

(2 days)

↑ Students were more likely (OR: 1.9, 95% CI: 1.1–3.3) to select a serving of fruit in the intervention school (48%) compared to the control school (32%)

Consumption:

Visual Observation

(2 days)

↑ Students who selected fruit were more likely (OR: 2.3, 95% CI: 1.3–4.2) to consume fruit in the intervention school (87%) compared to the control school (65%)

Siegel et al., 2015 [52]

Non-Controlled Trial

Study Length: 4 months

Location: Ohio

Age Group: Grades K – 6

n = 297 students, 1 school

Quality Rating: Weak

− Smiley face signs placed next to encouraged items

(7 weeks)

− Intermittent verbal reminders

(Intermittently over 7 weeks)

Selection:

Sales Data

(Number of days NR)

↑ Increased plain white fat free milk selection from 7.4 to 17.9% of students (p < .001)

↓ Decrease in low-fat chocolate milk selection from 86.5 to 77.1% of students (p < .001)

⊘ No significant change in overall milk purchases (NS)

⊘ No significant change in fruit purchases (NS)

Swanson et al., 2009 [53]

Non-Controlled Trial

Study Length: NR

Location: Kentucky

Age Group: Grades K – 4

n = 491 students, 1 school

Quality Rating: Weak

− Slicing apples and oranges before serving

(1 day intervention condition, 1 day control condition)

Selection:

Digital Photography

(2 days)

↑ Selection of sliced oranges (16.2, 95% CI: 13.0–19.7) was greater than selection of whole oranges (5.5, 95% CI: 3.6–7.9)

⊘ No significant difference in apple selection (NS)

Consumption:

Digital Photography

(2 days)

↑ More students consumed entire serving of sliced oranges (10.2, 95% CI: 7.6–13.3) than whole oranges (2.3, 95% CI: 1.1–4.0)

⊘ No significant difference in apple consumption (NS)

Thompson et al., 2017 [54]

Non-Controlled Trial

Study Length: 8 months

Location: Minnesota

Age Group: Grades 1–4

n = 1861 students, 2 schools

Quality Rating: Weak

− Slicing apples

− Using attractive bowls or baskets

− Changing fruit and vegetable placement

− Creative, descriptive names to market food items (non-student developed)

− Menu displays with serving size suggestions

(All components 4 months)

Selection:

Direct Weighing Individual

(4 days)

↑ Percent of students selecting a serving of fruit increased from 95.5 to 98.1% (p = .02)

⊘ No significant change in vegetable selection (NS)

Consumption:

Direct Weighing Individual

(4 days)

⊘ No significant change in fruit consumption (NS)

⊘ No significant change in vegetable consumption (NS)

Wansink et al., 2013 [55]

Cluster Randomized Trial

Study Length: 2 weeks (approx.)

Location: New York

Age Group: Middle School

n = 643 students, 6 schools

Quality Rating: Moderate

− Slicing apples

(Duration NR)

Selection:

Visual Observation

(4 days)

↑ Intervention schools with fruit slicers had a larger increase in average daily apple sales (10%) compared to control schools (6%) (p < .01)

Consumption:

Visual Observation

(4 days)

⊘ No significant changes in total apple consumption from pre- to post-intervention at treatment schools (NS)

Waste:

Visual Observation

(4 days)

⊘ No significant difference in total apple waste (by weight) from pre- to post-intervention at treatment schools (NS)

Zellner et al., 2016 [56]

Non-Controlled Trial

Study Length: 2 months

Location: Pennsylvania

Age Group: Grades 3–4

n = 25 students, 1 school

Quality Rating: Moderate

− Offering fruit at the end of the meal, instead of during the meal

(1 day intervention condition, 1 day control condition)

Consumption:

Visual Observation

(2 days)

↑ Larger percent of students ate at least some of the vegetable (kale salad) in intervention (100%) compared to control (40%) (p = .0017)