French 2004.

Study characteristics
Methods	Trial name: Trying Alternative Cafeteria Options in Schools (TACOS) Study design: cluster‐RCT Intervention duration: 2 years Length of follow‐up from baseline: 2 years Differences in baseline characteristics: similar Unit of allocation: schools Unit of analysis: schools and students
Participants	School type: secondary schools Region: St Paul Metropolitan, Minneapolis, USA Demographic/socioeconomic characteristics: schools were predominantly urban. On average, 14% of students were non‐white (range 3–77%), and 9% were eligible for free lunch (range 1–57%). Food services for 2 schools were run by food service management companies, 18 schools by the school district food services, 19 schools prepared meals on‐site and 20 schools participated in the NSLP. Inclusion criteria ‐ Presence of an à la carte area in the school cafeteria operated by the school food service. ‐ A food service director and principal willing to take part in the study for 2 school years. ‐ A willingness to be randomly assigned to intervention or control group. ‐ Computerised à la carte sales data. ‐ A willingness to share these data with researchers, allow a mail‐based administration of student evaluation surveys, and allow student groups to collaborate with research staff on the development and implementation of school‐wide promotional activities involving foods offered in the à la carte area. Number of schools allocated: 20 Numbers by trial group n (controls baseline) = 10 n (controls follow‐up) = 10 n (interventions baseline) = 10 n (interventions follow‐up) = 10 Recruitment: not reported Recruitment rate: 80% Of the 25 eligible secondary schools invited to participate, 5 declined, primarily because of the respective food service directors' concern about the additional food service staff burden related to compliance with research protocols. To avoid contamination caused by schools sharing the same food service director, only 1 school per district was included in the study. For student survey mean response rates for the 3 surveys were 75%, 75%, and 77%, and did not differ significantly between intervention and control schools.
Interventions	Number of experimental conditions: 2 (1 intervention, 1 control) Policies, practices or programmes targeted by the intervention Intervention consisted of increasing the availability of lower‐fat foods in cafeteria à la carte areas and implementing school‐wide, student‐based promotions of these lower‐fat foods. The goal was to increase lower‐fat à la carte food availability by 30% relative to baseline. The ultimate goal was to have 50% of products be lower fat. Implementation strategies EPOC: local consensus processes ‐ Quarterly meetings between research and food service staff were held to review progress towards goals. EPOC: tailored intervention ‐ Development of tailored lists of higher‐ and lower‐fat foods for schools. EPOC: education meetings ‐ TACOS staff worked with the student groups and their faculty advisors to train the students for specific promotional activities and to act as liaisons between students and the food service staff. EPOC: pay for performance ‐ Student groups were offered financial incentives for completing each promotion (from USD100 to USD300, depending on the complexity of the promotion). Theoretical underpinning: not reported Description of control: no intervention control
Outcomes	Outcome relating to the implementation of school service policies, practices or programmes ‐ Students had seen posters in school about cafeteria food choices ‐ Students had heard any messages over public address system, in school ‐ Students had heard about any contests or events at school about cafeteria food choices ‐ Students had taken part in any taste tests, food samplings, or contests in the school cafeteria ‐ Percentage low‐fat à la carte foods Data collection method: to monitor the extent to which intervention schools implemented their lower‐fat à la carte food availability goals, trained research staff visited each intervention school every 3 weeks to record all foods offered at lunchtime in the à la carte areas. In addition, complete à la carte inventories in intervention and control schools were conducted by trained research staff at baseline and after the second intervention year. Student exposure to the TACOS intervention activities was assessed with a series of 4 questions on a mailed student survey. Validity of measures used: not reported. Both self‐reports and objective measures were used. Outcome relating to cost: not reported Outcome relating to adverse consequences: school food service revenue was reported. Data collection method: data on school food service revenues were collected at the end of each school semester from a same point‐of‐sales software program including revenues from student reimbursable lunches, student à la carte foods, total à la carte foods and total school food service revenues. Validity of measures used: not reported Outcome relating to child diet, PA or weight status: nutritional food choices 2 outcome measures assessing nutrition intake: 1. percentage of lower‐fat à la carte foods sold and 2. students' self‐reported food choices. Data collection method: 1. students' self‐reported food choices: student food choices were measured via a mailed survey to a random sample of 75 students per school, according to the Dillman method (Dillman 1978), during the autumn of 2000, the spring of 2001, and the spring of 2002. Surveys comprised 48 questions related to students’ food choices. 2. sales data were collected on a weekly basis in electronic format from school food service staff in each of the 20 schools. Validity of measures used: sales data: objective
Notes	Research funding: supported by the National Institutes of Health (grant R18 HL61305). Conflicts of interest: not reported
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Schools were randomly allocated to experimental group. Random sequence generation procedure was not described.
Allocation concealment (selection bias)	Unclear risk	There was no information provided about allocation concealment and, therefore, it was unclear if allocation was concealed.
Blinding of participants and personnel (performance bias) Implementation outcome	High risk	Outcome group: student data and observations: no mention that students or TACOS staff were blinded. Outcome: adverse impact. Low: school revenue objective data.
Blinding of outcome assessment (detection bias) Implementation outcome	High risk	Outcome group: student data and observations: no mention that students or TACOS staff were blinded. Outcome: adverse impact. Low: school revenue objective data.
Incomplete outcome data (attrition bias) Implementation outcome	Low risk	Outcome group: all 20 secondary schools participated.
Selective reporting (reporting bias)	Unclear risk	There was no study protocol, therefore, it was unclear if there was selective outcome reporting.
Other bias	Low risk	Did not appear to be at risk of contamination.
Recruitment to cluster	Low risk	Schools were randomised to condition and those within each school participated.
Baseline imbalance	Low risk	Schools were randomised to condition.
Loss of cluster	Low risk	There were no losses of clusters.
Incorrect analysis	Unclear risk	Graphs of data by school showed that the first 3 weeks and last 3 weeks of each 40‐week academic year demonstrated excessive variation attributable to start‐up and termination process. Therefore, data from these weeks were excluded from the analysis. Unclear how the full analysis was performed.
Compatibility with individually randomised RCTs	Unclear risk	Unable to determine if a herd effect existed.
Overall risk of bias assessment	Unclear risk	Most domains were at unclear risk of bias.