Story 2000.

Methods	Trial name: 5‐a‐Day Power Plus. Study design: Cluster‐randomised controlled trial. Intervention duration: Unclear (beginning in the fourth grade (1994‐1995) and throughout the fifth‐grade school year (1995‐1996). Length of follow‐up from baseline: approximately 12 months. Differences in baseline characteristics: Similar – schools were matched pairs that were then randomised. Unit of allocation: Schools. Unit of analysis: Cafeteria observations.
Participants	School type: Elementary schools. Region: St. Paul, Minnesota, USA. Demographic/socioeconomic characteristics: Of the 1750 fourth‐grade students enrolled at baseline, 1.3% were Native American, 6.4% were Hispanic, 19.1% were African American, 25.2% were Asian American (largely Hmong), and 48% were white. Approximately 60% of the students were eligible for free or reduced‐price school meals. Inclusion/exclusion criteria: Not reported. Number of schools allocated: 20 inner‐city public elementary schools. Numbers by trial group: 10 intervention, 10 control. Recruitment: Recruitment strategy not reported. Recruitment rate: Not reported.
Interventions	Number of experimental conditions: 2 (1 intervention, 1 control). Policies, practices or programs targeted by the intervention: The intervention consisted of a parent, industry, curricula and classroom component. Only the food service and curricula components were subject to an implementation strategy. Only the food service component had an implementation outcome assessed between experimental groups and consequently, only food service implementation strategies were extracted. Regarding the food service intervention, the four food service intervention strategies were 1) point‐of‐purchase promotion of fruit and vegetable using characters and messages from the classroom curricula, 2) increasing the appeal of fruit and vegetable by enhancing their attractiveness, 3) increasing the variety and choice of fruit and vegetable served, and 4) offering an additional fruit choice on days when baked or frozen desserts were served. These four strategies involved eight guidelines on how to offer appealing fruit and vegetable choices and four promotion guidelines. Implementation strategies: EPOC: Educational meetings: ‐ Centralised training sessions were held for food service staff from the intervention schools. It was held during a regularly scheduled school day and was conducted by the 5‐a‐Day Power Plus staff. Food service staff attended the teacher training for 2 hours and also attended 2‐hour training after school each of the 2 intervention years. Food service staff members were paid for attending the training members were paid for attending the training. EPOC: Other: ‐ A local producer provided some fruit and vegetable for use in classroom taste testing, home snack packs, and to expand choice in school lunch. They also provided a 30‐minute presentation on fruit and vegetable to each of the fifth‐grade intervention classrooms. Service staff members were paid for attending the training. Theoretical underpinning: 5‐a‐Day Power Plus intervention program was guided by social cognitive theory and social learning theory. Description of control: Not reported but assume usual practice.
Outcomes	Outcome relating to the implementation of school service policies, practices or programs: ‐ Mean number of fruit and vegetable choices available 4th grade ‐ Mean number of fruit and vegetable choices available 5th grade ‐ Mean % of eight guidelines on how to offer appealing fruit and vegetable met 4th grade ‐ Mean % of eight guidelines on how to offer appealing fruit and vegetable met 5th grade ‐ Mean % of four fruit and vegetable promotions met 4th grade ‐ Mean % of four fruit and vegetable promotions met 5th grade Data collection method: An observation‐based process evaluation method was also developed to assess the food service intervention implementation. Direct observations were conducted in each of the 10 intervention and 10 control schools on a monthly basis using trained observers and standardised protocols and instruments. Validity of measures: Not reported however, considered an objective measure of implementation. Outcome relating to staff knowledge, skills or attitudes: Not reported. Outcome relating to cost: Not reported. Outcome relating to adverse consequences: Not reported. Outcome relating to child diet, physical activity or weight status: Student dietary intake. Data collection method: 24‐hour quantified food record and recall and student lunchroom observations. Validity of measures used: Not stated although observations considered an objective measure.
Notes	Research funding: Supported by Grant R01CA59805 from the National Institute of Health. Conflicts of interest: Not reported.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Schools were matched in pairs and randomly allocated to experimental group. The random sequence generation procedure is not described.
Allocation concealment (selection bias)	Unclear risk	There is no information provided about allocation concealment and therefore it is unclear if allocation was concealed.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Outcome group: Participants (teachers and cooks) were aware that they were being asked to implement an intervention. There was no blinding to group allocation and this is likely to influence performance.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Outcome group: Trained evaluation staff visited each school to conduct observations using standardised protocols and measures. However, there is no information provided about whether these personnel were blinded to group allocation and teacher self‐reported measures were completed.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Outcome group: There is no report of any schools dropping out of the study.
Selective reporting (reporting bias)	Unclear risk	There is no study protocol therefore it is unclear if there was selective outcome reporting.
Recruitment to cluster	Low risk	Individuals within each randomised cluster participated.
Baseline imbalance	Low risk	Schools were randomly allocated to condition and so risk of baseline imbalance is low.
Loss of cluster	Low risk	There is no report of any schools dropping out of the study.
Incorrect analysis	High risk	There was no reporting of statistical techniques apart from simple t‐tests.
Contamination	Low risk	Schools were randomly allocated to condition and so the risk of contamination is low.
Compatibility with individually randomised RCTs	Unclear risk	Unable to determine if a herd effect exists.