TABLE 3.
Summary of studies evaluating farm to school programs (n = 7) or farm to school–related activities (n = 14)1
| First author (reference no.); year; population; location | QA score | Study design2 | Farm to school–related objective | Intervention components (duration) | Outcomes (measurement method) and when measured | Results | Farm to school implications |
|---|---|---|---|---|---|---|---|
| Bontrager Yoder (24); 2015; 1877 3rd- to 5th-grade students in 11 schools; Wisconsin | Fair | Repeated cross-sectional study | To examine whether Farm to School can reduce FV waste over time | Local food procurement, Smarter Lunchrooms (this was a cross-sectional, secondary data analysis. Food items were categorized as local or nonlocal and placement) | FV consumption (digital photography quarter plate waste method) measured up to 5 times over 4 y | Where identified, locally sourced items were wasted more than conventionally sourced (+0.1 cups, P < 0.0001). Increasing prior farm to school years decreased waste (−0.02 cups, P < 0.0001) | Differences in local vs. nonlocal food packaging may influence consumption rates. Schools with more prior years of F2SP wasted less FV than schools with fewer years of F2SP |
| Bontrager Yoder (25); 2014; 1117 3rd- to 5th-grade students; Wisconsin | Weak | Noncontrolled trial | To assess the effectiveness of F2SPs in increasing students’ FV consumption, knowledge, and attitudes | Nutrition education; local food promotion; local food procurement; taste tests; cooking demonstrations; V farm field trips (specific durations not provided) | Knowledge and attitudes (60-item survey), FV consumption (FFQ and lunch tray digital photography observation) collected at the start and end of the 2010–2011 academic year | Significant increases overall in knowledge (+1%), FV selection at lunch (+6%), and FV willingness to try scores (+1%, all P < 0.001); no change in overall dietary patterns | Farm to school programming improves mediators of FV consumption, although it does not lead to increases in overall consumption |
| Bristow (26); 2017; 1183 high school students; urban school district, St. Louis, Missouri | Weak | Noncontrolled trial | To determine if a locally grown sweet potato taste test and promotions will increase the selection of sweet potatoes | Taste testing (1 d); local food procurement (ongoing); local foods promotion (2 d during a 2-wk period) | Sweet potato selection (sales data) was measured on the same day that taste tests and promotions occurred | The number of servings of sweet potato selected increased from 0 at baseline to 4 at day of taste test to 24 servings 1 wk posttaste test (P < 0.05) | Promotions of local vegetables by taste testing are associated with increased consumption |
| Evans (18); 2012; 246 6–7th-grade students from 5 middle schools; Texas | Fair | Nonrandomized control trial | To test the influence of multicomponent programming on students’ knowledge, attitude, preference for, and consumption of FV and also the effects of individual components on FV consumption | Integrated curriculum and nutrition education (4 lessons over 5 mo); school gardens (ongoing); farmer visits (1–2 over 5 mo); field trips (once); taste tests (3 over 5 mo); local food promotion (once/wk for 5 mo); local food procurement (once/wk for 5 mo) | FV motivation, self-efficacy, preference, and knowledge (survey); FV consumption (FFQ); pretest baseline measured in January 2009, and posttest measured in May 2009 | Students exposed to 2 or more components decreased their preference for unhealthy foods (–1.05, P < 0.01), improved self-efficacy (1.97, P < 0.01), increased FV knowledge (0.57, P < 0.01), and increased FV servings/d consumed (0.96, P = 0.01), but there were no differences between intervention and comparison groups in FV motivation (0.64, P = 0.26) or preference (0.14, P = 0.24) | The treatment effect for exposure to only 1 component was limited to improvement in knowledge, but exposure to ≥2 also provided improvements in FV consumption and other nutrition attitudes |
| Jones (17); 2015; 15 elementary and 3 middle schools with ∼528 students per school; South Carolina | Fair | Nonrandomized control trial | To examine the impact of the first year of an F2SP on children's FV consumption | Local food procurement; school gardens; farm field trips; cooking demonstrations; promoting local foods (treatment duration is 1 school year but intensities are variable and unspecified. Exception: local food procurement—at least 2 fruits/vegetables per month) | FV consumption (digital photography Comstock plate waste method) data for treatment and control measured at the end of the school year, near or during the “South Carolina Grown Day” | 4.5% more F2SP children tried V (P < 0.05); 3.0% fewer F2SP children tried F (P < 0.05). F2SP children consumed 0.11 additional servings of V (P < 0.05) but 0.07 fewer servings of F were consumed (P < 0.05) | Children in schools with an F2SP were more likely to try and consume V, but this was negatively related to trying and consuming F |
| Moss (27); 2013; 65 3rd-grade students in 1 school; rural Illinois | Weak | Nonrandomized control trial | To test the effects of a farm field trip plus nutrition education on children's nutrition knowledge and FV consumption compared with nutrition education alone | Farm field trip (once in week 4), nutrition education (2 lectures of 30 min in weeks 2 and 3. Study duration: 4 wk) | Nutrition knowledge and FV consumption (22-item survey); pretest survey measured at week 1 and posttest survey at week 4 | No relation was found between self-reported F and V consumption and participation on the farm field trip (statistics for nonsignificant findings not reported). Significant differences were found concerning knowledge of fiber (χ2 = 11.6, P < 0.001), vitamins and minerals (χ2 = 4.4, P < 0.05) among the nutrition only and nutrition + farm field trip group compared with the control | There is insufficient evidence that a farm field trip provides significant influence on FV consumption compared with nutrition education alone |
| Smith (28); 2012; 2130 6th-grade students; Michigan schools | Weak | Noncontrolled trial | To describe a multicomponent F2SP aiming to improve student FV consumption and knowledge | School gardens; cooking classes; farmer visits; nutrition education; local food procurement; taste tests; train food service staff; physical activity promotion (treatment duration is not provided; thus, no information is available about time lapse between baseline and postintervention) | FV consumption (not reported). Emphasis of study is on collaboration; thus, the timing of postintervention measurement is not reported | 44.6% of students who did not consume at least 5 FV per day at baseline improved their consumption by at least 1 FV after intervention (P < 0.001) | Multicomponent F2SP may promote consumption of FV |
| Bates (29); 2015; 7th- through 12th-grade students; 2 schools in Provo, Utah | Strong | Crossover study | Examine the impact of serving fruit smoothies at school breakfast on daily fruit consumption | New product on menu (on the menu 13 out of 28 d of during-treatment data collection) | Selection and consumption (plate waste via quarter waste visual observation method) collected for 2–4 wk prior to the intervention and 4 wk during the intervention (68 d of data collection) | Students consuming ≥1 serving of whole fruit increased from baseline (4.3% to 45.1%; P < 0.01). Consumption returned to baseline levels when smoothies not offered | Students are more likely to consume smoothies than other fruit options, but this does not promote long-term behavior change |
| Blom-Hoffman (30); 2004; 91 kindergarten and 1st-grade students; underresourced urban school; northeastern United States | Strong | Cluster randomized trial | Examine the impact of a multicomponent nutrition program on nutrition knowledge and school lunch vegetable consumption | Nutrition education; taste tests; parent engagement; food coaches (classroom/knowledge/home component included a 5-wk intervention) | Nutrition knowledge (7-item survey) measured as pretest, posttest, and follow-up, vegetable consumption (visual observation using Comstock scale measured across 3 lunch periods during pretest, posttest, and follow-up) | Nutrition knowledge significantly higher in experimental group compared with control group (P = 0.0001), and the experimental group's knowledge scores were maintained at 1 mo later. There was no significant change in vegetable consumption (mean: 3.46% at pretest and 3.21% at posttest) | Multicomponent programs increase students’ nutrition knowledge despite variable intervention integrity, but they will not have an impact on vegetable consumption |
| Bogart (31); 2014; 2997 7th-grade students in Los Angeles Unified School District; Los Angeles, California | Strong | Cluster randomized Trial | To test the impact of a multicomponent intervention, including healthy food promotion and sliced/snack-size FV, on NSLP participation and FV selection | Smarter Lunchrooms; taste tests; physical education; nutrition promotions; food coaches (January 2009–June 2012; 5-wk interventions in spring semesters) | Lunch participation (administrative data), FV selection (sales data). Administration and sales data were collected for intervention duration | 15.3% increase in F selection (P = 0.006), 10.4% increase in lunches served (P < 0.001); 2.2% increase, but not statistically significant, in V selection (P > 0.1) | Multicomponent programming may influence students’ meal participation and F selection decisions, but a more intensive intervention may be required to shape V decisions |
| Cohen (32); 2015; 968 students, 3rd through 8th grade; 2 urban low-income school districts in Massachusetts | Strong | Cluster randomized trial | Test the association of a chef intervention and Smarter Lunchrooms intervention on school food selection and consumption | Chef-enhanced meals (7 mo of total exposure); train food service staff (no specified duration); Smarter Lunchrooms (4 mo; started 3 mo in the Chef intervention for treated schools) | Selection and consumption of school meal components (individual food weights) data were collected in 6 d per school: 2 randomly selected nonconsecutive days for the baseline and 2 d in each of the 2 postintervention data collection periods | At 3 mo, V selection increased in chef vs. control schools (OR: 1.75; 95% CI: 1.36, 2.24). At 7 mo, V selection increased in the chef (OR: 2.54; 95% CI: 1.83, 3.54), Smarter Lunchrooms (OR: 1.91; 95% CI: 1.46, 2.50), and chef plus Smarter Lunchrooms (OR: 7.38; 95% CI: 5.26, 10.35) schools compared with the controls. Smarter lunchrooms alone had no effect on V consumption (OR: –10.7%; 95% CI: –23.8%, 2.4%), but consumption increased in chef (OR: 24.5%; 95% CI: 10.0%, 39.0%) and chef plus Smarter Lunchroom schools. At 3 mo, F selection did not increase in a statistically significant way in chef vs. control schools (OR: 1.46; 95% CI: 0.67, 3.21). Long-term F selection increased in the chef (OR: 3.08; 95% CI: 2.23, 4.25), Smarter Lunchroom, (OR: 1.45; 95% CI: 1.13, 1.87), and chef plus Smarter Lunchroom (OR: 3.10; 95% CI: 2.26, 4.25) schools compared with controls. Among students selecting F, consumption was higher in chef schools (OR: 0.17 cups/d; 95% CI: 0.03, 0.30), whereas the Smarter Lunchroom had no effect (OR: –0.00 cups/d; 95% CI: –0.13, 0.11) | Children may require repeated, long-term exposure to new foods in order to influence their selection and consumption behaviors. Chef interventions alone or in combination with Smarter Lunchrooms may improve selection and consumption of school meals |
| Folta (33); 2006; 880 elementary school students; Boston metropolitan area | Strong | Cluster randomized trial | Test the effectiveness of promotional messages on selection of legumes at school lunch | Nutrition promotion (30- to 60-s message promoting bean dishes played once a day for 1 wk over the public address system) | Lunch entrée selection (plate waste via visual observation)—every day a bean dish was served | There were no overall significant differences between intervention and control schools (OR: 0.93; 95% CI: 0.81, 1.07; P = 0.31). Students who received daily messaging were 2.47 times more likely (95% CI: 1.74, 3.53; P < 0.001) to select a bean dish compared with students in the control school | Messages alone were not enough to significantly influence behavior. However, the overall message dose may not have been strong enough to change behavior |
| Foster (34); 2008; 1349 4th- through 6th-grade students from 10 schools; mid-Atlantic region | Strong | Cluster randomized trial | To examine the impact of a multicomponent intervention on the prevention of overweight and obesity | Integrated curriculum (2 y); nutrition education (50 h of nutrition education per student/school year); nutrition promotions; parent education (various frequency: home and school association meetings; report card nights; weekly nutrition workshops); school nutrition policies (all food served was changed in the intervention schools for the 2 school years) | BMI (z scores), dietary intake (FFQ), collected at baseline (spring semester) and then after 2 y (spring semester) | Students in the intervention were 35% less likely to be overweight after 2 y (OR: 0.65; 95% CI: 0.54, 0.79; P < 0.01). There was no significant change in obesity prevalence (OR: 1.09; 95% CI: 0.85, 1.40; P = 0.48) or in dietary intake, expressed as total energy in kJ/d (OR: –104.27; 95% CI: –234.29, –25.73; P = 0.12); total fat in g/d (OR: –3.78; 95% CI: –8.59, 1.02; P = 0.12); F and V, n per day (OR: –0.04; 95% CI: –0.37, 0.30; P = 0.82) | A multicomponent program may decrease the prevalence of overweight children |
| Hendy (35); 2005; 346 1st-, 2nd-, and 4th-grade students; rural eastern Pennsylvania | Strong | Cluster randomized trial | Test effectiveness of a multicomponent intervention on FV preference and consumption | Smarter Lunchrooms; incentives for FV consumption; peer modeling (12 meals observed 3 d/wk; total intervention length not disclosed by authors) | Student FV consumption (plate waste via visual observation collected during intervention), FV preference (hedonic scoring interviews conducted 2 wk and 7 mo postintervention) | Significant association between reinforcement condition (F or V) and associated item consumption, but increases in V consumption were not sustained. Fruit preference ratings increased (P < 0.02) and V preference ratings also increased but did not reach significance (P < 0.07). FV preference returned to baseline 7 mo after intervention completion | Improving child choice and providing incentives may have an initial impact on food preferences and consumption, but these gains may not be sustained without additional intervention |
| Hoffman (36)3; 2010, 297 kindergarteners and 1st graders; urban northeastern United States | Strong | Cluster randomized trial | Examine the association of intervention components on student consumption of FV during school lunch over 2 y | Nutrition promotions; food coaches (2 y) | FV consumption (individual food weights), FV preference and knowledge (20-item and 9-item surveys); measurements collected each spring during the intervention | Significant increase in consumption of FV at end of year 1 (F = 18 g, P < 0.001; V = 7 g, P < 0.01) and of F at end of year 2 (F = 7 g, P < 0.001); no difference for V at end of year 2; no significant differences in FV preferences or BMI | Initial gains in FV consumption but no long-term gain in V consumption |
| Hoffman (37)3; 2011; 297 kindergarteners and 1st graders; urban northeastern United States | Strong | Cluster randomized trial | Examine the impact of a multicomponent nutrition promotion intervention on student consumption, preference, and knowledge of FV, and on BMI over 3.5 y | Nutrition promotions; food coaches (2.5 y) | FV consumption (individual food weights), FV preference and knowledge (20-item and 9-item surveys), and BMI; measurements collected 5 times across 3.5 y (preintervention and each spring thereafter) | Student FV knowledge was 0.5 points (P < 0.05) higher than controls 3.5 y after intervention. No long-term impact on consumption, preferences, or BMI | Initial gains in FV consumption are difficult to maintain over the long term, but the impact on knowledge is more feasibly sustained |
| Perry (38); 2004; 1820 1st- and 3rd-grade students, 2 school districts; Minneapolis and Saint Paul, Minnesota | Strong | Cluster randomized trial | To determine if a cafeteria-based intervention would increase FV consumption | Nutrition promotion; Smarter Lunchrooms; taste tests; train food service staff (2 consecutive school years) | FV consumption (visual observation); measurements collected at the end of the intervention (after 2 y) | Significantly higher intakes of total consumed FV servings (0.14, P = 0.03) and fruits (0.17, P < 0.01); not statistically different intakes for fruit juice (–0.01, P = 0.77) or vegetables (–0.06, P = 0.32) | Cafeteria promotions are effective in creating modest FV consumption increases, but they may be more effective in concert with other interventions |
| Reynolds (39); 2000; 1698 4th-grade students from 28 schools | Strong | Cluster randomized trial | To test the association of a multicomponent dietary intervention on student FV consumption and psychosocial measures | Nutrition education; taste tests; Smarter Lunchrooms; train food service staff; parent outreach (2 consecutive school years) | FV consumption (24-h recall, plate waste via visual observation); psychosocial measures such as nutrition knowledge and self-efficacy (67-item questionnaire); questionnaire administered at baseline; 1- and 2-y measurements collected at least 1 mo after termination of that year's intervention | Significant difference in consumption of FV servings between groups at follow-up 1 (3.96 vs. 2.28 servings; P < 0.0001) and follow-up 2 (3.2 vs. 2.21 servings; P < 0.0001), but these 24-h recall outcomes were not able to be replicated via visual observation. The intervention groups had significantly higher nutrition knowledge scores at follow-up 1 and 2; intervention group self-efficacy scores were higher at follow-up 1 only (0.240 vs. 0.014; P < 0.0004) | Multicomponent nutrition education is associated with increased consumption of FV over the long term, but challenges with lunch time constraints may have limited impacts to out of school meals |
| Wells (19); 2015; 3061 2nd-, 4th-, and 5th-grade students in 49 low-income schools throughout the United States | Strong | Cluster randomized trial | To examine the effects of a school garden intervention on the science knowledge of elementary schoolchildren | School gardens; integrated curriculum (2 consecutive school years) | Nutrition and plant science knowledge (7-item questionnaire); survey administered at baseline and at 3 points during the intervention | Intervention group showed greater increases in knowledge (an increase from 3.2 out of 7 correct answers to 3.84; P < 0.0001). Schools with higher garden intervention fidelity experienced the highest increases in science scores (increased by 0.78; P < 0.0001) | A robust school garden intervention combined with integrated curriculum may lead to modest increases in science knowledge among low-income students |
| Willi (40); 2012; 4363 6th-grade students; 42 middle schools throughout the United States | Strong | Cluster randomized trial | To test the association of a multicomponent intervention on cardiovascular disease risk in middle school students | Nutrition promotion; physical education; serving healthier options in cafeteria (2.5 consecutive school years) | Blood pressure, fasting blood lipids, BMI; measurements taken at baseline and end of study | No significant associations of blood lipids, BMI, or overall blood pressure; significant decrease in the occurrence of hypertension (from 13.8% to 8.3%) in non-Hispanic black males (P = 0.0073) and of hypertension (from 11.5% to 6.5%) in non-Hispanic white males (P = 0.0370) | A multicomponent program may not improve overweight/obesity or blood pressure overall, but it may reduce blood pressure in high-risk populations |
| Williams (41); 2016; 225 3rd- through 5th-grade students; New York City public schools | Strong | Cluster randomized trial | To test the impact of calorie labels and culturally appropriate nutrition promotions and nutrition education on snack purchases at school | Nutrition promotion; nutrition education (18 mo) | Snack selection (snack sales data); collected 1, 7, and 12 d after the intervention depending on the site | 20% decrease in total kilocalories per snack (P < .01) postintervention; increase in number of healthy food sales from 8% to 36% (P < 0.01) and decrease in unhealthy foods from 51% to 36% (P < 0.01) postintervention | Culturally relevant nutrition education paired with calorie labeling may improve snack choices of students at school |
1
Studies that explicitly investigated F2SP are presented first, followed by studies that investigated farm to school–related activities. F, fruit; F2SP, farm to school program; FV, fruit and vegetable; NSLP, National School Lunch Program; QA, quality assessment; V, vegetable.
2
Study designs were cataloged according to definitions developed by the Academy of Nutrition and Dietetics (23).
3
There are data overlap between these 2 studies.