Taylor 2018.

Study characteristics
Methods	Trial name: Shaping Healthy Choices Program (SHCP) Study design: cluster‐RCT for implementation outcome Intervention duration: 9 months Length of follow‐up from baseline: 9 months autumn 2012 to spring 2013 Differences in baseline characteristics: characteristics of schools and participating students were similar across the 2 school sites, although a greater proportion of control students had 1 or both parents with a bachelor's degree or above (Chi2 = 6.496, P < 0.05). Unit of allocation: cluster (school) Unit of analysis: ‐ Implementation outcomes: cluster (school) ‐ Behavioural/health outcomes: individual (child)
Participants	School type: 2 elementary schools Region: 2 school districts in central and Northern California, USA Demographics/socioeconomic characteristics: ethnic diversity index: control 76, intervention 64. FARMS eligibility: control 29.6, intervention 34.2; greater proportion of control students had 1 or both parents with a bachelor's degree or above (Chi2 = 6.496, P < 0.05). Inclusion/exclusion criteria Inclusion Schools were selected based on moderate eligibility rates for FARMS (30–49%), presence of ≥ 4 fourth‐grade classrooms, absence of a salad bar and absence of an instructional school garden. Exclusion:not reported Number of schools allocated Schools:2 Students:111 children with pretest data and 112 with post‐test data. Numbers by trial group n (controls baseline) = 39 students, 1 school n (controls follow‐up) = 39 students, 1 school n (interventions baseline) = 72 students, 1 school n (interventions follow‐up) = 72 students, 1 school (73 for vegetable outcome) Recruitment Schools:2 school districts in central and Northern California were recruited to participate in this pilot study evaluating the SHCP. Schools were selected based on moderate eligibility rates for FARMS (30–49%), presence of ≥ 4 fourth‐grade classrooms, absence of a salad bar and absence of an instructional school garden. Additional details on inclusion criteria and sample size determination are published elsewhere. Students:of the 303 students enrolled at the 2 schools, 294 fourth‐grade students (133 control, 161 intervention) consented into the study. Of these students, pretest data were available for 168 and 172 students for fruits and vegetables, respectively. Students with both pretest and post‐test data were included in the final sample, with data available for 111 and 112 students for fruits and vegetables, respectively. Recruitment rate Schools:not reported Students:not reported
Interventions	Number of experimental conditions: 2 (1 intervention, 1 control) Policies, practices or programmes targeted by the intervention: healthy eating Implementation strategies EPOC: educational outreach visits ‐ Food‐service staff provided an in‐class demonstration of salad bar use prior to its cafeteria launch marketed regionally grown produce in the cafeteria as a complement to messaging in parent newsletters. EPOC: local consensus processes ‐ School site‐specific wellness committees. EPOC: educational materials ‐ District's nutrition services director was provided a list of foods grown in the garden and used in cooking demonstrations, so purchases could mirror SHCP activities when possible; 9 take‐home activities and family newsletters. EPOC: external funding ‐ A salad bar was installed in the intervention school cafeteria. The school district was provided USD3000 to increase procurement of regionally grown produce for use in the National School Lunch Program (NSLP). EPOC: other ‐ Health fair Theoretical underpinning ‐ Of the evidence‐based intervention/policy/practice or programme: social‐ecological model and Social Cognitive Theory. ‐ Of the implementation strategy: none reported. Description of control: control schools received a delayed intervention during the 2013–2014 school year.
Outcomes	Outcome relating to the implementation of school policies, practices or programmes: fruit and vegetable availability within the school canteen. Data collection method Based on produce expenditures and variety for use in the schools' NSLP. The extent of regional produce procurement was documented by measuring regional and non‐regional procurement dollars spent in the intervention and control school cafeterias on fresh fruits and vegetables. For this school district, regional produce was defined using the nutrition services director's definition, which was approximately 7 hours of driving time from farm to canteen. Procurement records, including school inventory and delivery records as well as district invoices and year‐end summaries, provided the current level of produce procurement and types of produce purchased. Validity of measures used: not reported Outcome relating to cost: availability: the extent of regional produce procurement was documented by measuring regional and non‐regional procurement dollars spent in the intervention and control school cafeterias on fresh fruits and vegetables. Procurement records, including school inventory and delivery records as well as district invoices and year‐end summaries, provided the current level of produce procurement and types of produce purchased. Each month's expenditures were divided by the number of meals served in that period to arrive at mean dollars spent per participating student per day. Outcome relating to adverse consequences: fruit and vegetable waste at lunchtime was assessed via lunchtime intake method. Outcome relating to child diet, PA or weight status: child lunchtime intake of fruit and vegetables (selection not included). Data collection method: lunchtime intake of fruits and vegetables was assessed among students choosing meals provided through the NSLP. Selection, consumption and plate waste were determined for all fruits excluding juice and all vegetables offered as side dishes. Vegetables integrated in entrée dishes (e.g. burritos, lasagne) were excluded, because the intent of this evaluation was to focus on changes in dietary behaviour that may indicate that vegetables are chosen actively, as opposed to consuming them passively in mixed dishes. Because there are well‐known limitations of self‐reported dietary intake, such as depending on one's capacity to accurately estimate portion sizes, 16 dietary behaviours were assessed using digital imaging. Adopting protocols described in previous studies, digital images of students’ lunch trays were collected at the beginning and end of the meal with the students' study ID to facilitate matching of the selection and plate waste images. Validity of measures used: this method had demonstrated reliability and validity for measurement of fruit and vegetable consumption among elementary school students.
Notes	Research funding: Division of Agriculture and Natural Resources, University of California (11‐1018), and US Department of Agriculture (2011‐38420‐20082, 221082). Conflicts of interest: none reported.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No details on randomisation process.
Allocation concealment (selection bias)	High risk	Allocation could not have been concealed. Intervention school would have determined their status as a salad bar was installed.
Blinding of participants and personnel (performance bias) Implementation outcome	High risk	Outcome group: procurement records. High: schools were not blinded to group allocation. The nature of the study precluded blinding. Outcome group: fruit and vegetable outcome data and food wastage (adverse effects). High: children unlikely to know allocation, researchers blind to condition. Trained research analysts, blinded to treatment assignment, estimated quantities of fruits and vegetables selected, consumed and wasted in cups.
Blinding of outcome assessment (detection bias) Implementation outcome	Low risk	Outcome group: objective data (procurement records) likely to be low.
Incomplete outcome data (attrition bias) Implementation outcome	Unclear risk	Outcome group: procurement records. Unclear: incomplete data unclear. Assumed available for both schools. Outcome group: fruit and vegetable outcome data. High: 111 and 112/294 (about 38%) consenting students had both pretest and post‐test data.
Selective reporting (reporting bias)	Unclear risk	Unclear – no trial registry or protocol available.
Other bias	High risk	Appeared to be at risk of contamination and other biases.
Recruitment to cluster	Unclear risk	Unclear if students knew group allocation when consenting to the study.
Baseline imbalance	High risk	Characteristics of schools and participating students were similar across the 2 school sites, although a greater proportion of control students had 1 or both parents with a bachelor's degree or above (Chi2 = 6.496, P < 0.05).
Loss of cluster	Low risk	Presume no loss of clusters as there were only 2 schools. 303 students enrolled at the 2 schools, 294 fourth‐grade students consented into the study. Of these students, pretest data were available for 168 and 172 students for fruit and vegetables, respectively. Students with both pretest and post‐test data were included in the final sample, with data available for 111 and 112 students for fruit and vegetables, respectively.
Incorrect analysis	High risk	Clustering not accounted for in analysis. 14 classrooms across 2 schools, analyses did not account for clustering because a small proportion of variability was attributed to clustering at the classroom level.
Compatibility with individually randomised RCTs	Unclear risk	No statement regarding this.
Overall risk of bias assessment	High risk	Most domains were at high risk of bias.