Emerson 2017.

Methods	Design: ITS Timing: Prospective Allocation to group: N/A (ITS without a control group) Number of clusters or sites: 3 schools Number of individuals: 960 (of which approximately 60% participated in the school lunch programme, i.e. the intervention) Length of intervention: 20 months (October 2014 to May 2016), with 56 weeks of data collection
Participants	General description of participants: Elementary school children in Norwood, USA Age: Children (elementary school students, grade k to 6; no age details reported) Inclusion criteria: N/R (all children attending the participating schools who had lunch in the cafeteria were included) Exclusion criteria: N/R Recruitment: N/R Weight status at baseline: N/R SSB consumption at baseline: At baseline, 82% of children chose chocolate milk for lunch, and 10% plain milk Equity considerations: 71% of students were white, 13% black, and 10% Hispanic; 47% were girls, and 73% were low‐income, defined as less than 130% of the poverty level. Quote: "[T]he Norwood City School District (…) has a diverse and economically broad student body"
Interventions	Intervention: Small prizes for the selection of plain, fat‐free milk instead of chocolate milk. (Quote: "Green 'smiley‐faced' emoticons were placed by preferred food items (fruits, vegetables, plain fat free milk and entrée with whole grain) and signs were posted explaining the PP [Power Plate, i.e. a meal consisting of plain fat free milk, an entrée with whole grain, fresh fruit and fresh vegetables or a salad] on the first day of the intervention (…). Small prizes were given students selecting the PP on Tuesdays and Thursdays once the intervention began. On the first day of the intervention, a small prize, such as a bracelet, was distributed to students who selected the PP") Behavioural co‐intervention: None reported Control: No intervention
Outcomes	Measures of SSB intake: Rate of children choosing chocolate milk, assessed continuously throughout the study period with cash register data Measures of intake of alternatives to SSB: Rate of children choosing plain fat‐free milk, assessed continuously throughout the study period with cash register data Anthropometric measures: N/R Adverse outcomes: The study does not report how data on adverse outcomes were collected, and if adverse outcomes were observed or not Other outcomes: None included in this review
Context and implementation	Setting: Elementary schools Sector: Education Country: USA Year(s) when implemented: 2015 ‐ 2017 Mode of implementation: Pilot trial by researchers Level of implementation: Setting‐based intervention
Declarations	COI: "The authors declare no conflict of interest" Funding: "The project described was supported by the National Center for Advancing Translational Sciences of the National Institutes of Health, under Award Number 1UL1TR001425‐01" Trial registration: N/R Protocol availability: N/R
Notes	Note that figure 2 of the study’s primary publication (Emerson 2017) shows only data for the first part of the study. According to the study’s text, the intervention lasted considerably longer than suggested by figure 2, namely until May 2016, with 56 weeks of data collection. Quote: “The PPP was resumed for all three schools in September 2015 during the second academic year of the program and continued through the end of May2016. Thus the intervention lasted for 56 weeks with a 3‐week gap at the beginning of the 2nd academic year during weeks 23 to 26. Cafeteria cash register receipt data was collected for one month prior to the PPP intervention and then throughout the entire intervention period.”
Risk of bias
Bias	Authors' judgement	Support for judgement
Incomplete outcome data (attrition bias) All outcomes	Low risk	Missing outcome data are unlikely to bias results ‐ the routinely‐collected sales data can be assumed to be close to complete
Blinding (performance and detection bias) Objective outcomes	Low risk	Participants were not blinded, but outcomes are objective, and outcome data were collected independently of the intervention and study
Selective reporting (reporting bias)	Unclear risk	For chocolate and plain milk selection, only numerical data are reported; temporal trends can therefore be assessed only for Power Plate selection, which is only an indirect measure for chocolate and plain milk selection. Apart from this, there is no indication that outcomes were reported selectively. Outcome data on all food and beverage items targeted by the intervention, and mentioned in the Methods section, are reported
Independence of the intervention from other changes (performance bias)	Low risk	A number of other factors may have influenced the outcomes, but it is unlikely that they explain the existence and the direction of the effect. Quote: "Events are annotated [in the graph showing white milk selection over time] such as baseline period, when the PP was initiated and events that may have influenced the effectiveness of the program such as cafeteria staff changes at the Sharpsburg School or when the school dietary interns who helped run program left for summer break"
Pre‐specification of the intervention effect (detection bias)	High risk	The point of analysis is the point of intervention; the results reported in the study are, however, not based on a classical ITS analysis
Intervention effects on data collection (detection bias)	Low risk	Data were collected routinely and independently of the intervention and the study. It is unlikely that the intervention affected data collection. Quote: "Purchase data obtained from cash register receipts were supplied by the Food Services Department of Norwood City School District"
Other bias	High risk	While the graphical presentation of results strengthens our confidence that effects were due to the intervention, and cannot be explained with underlying temporal trends, these were not taken into account in analyses. In 1 of the 3 intervention schools, chocolate milk purchases seem to have reverted to baseline levels towards the end of the study period. The authors conduct a statistical process management (SPM) analysis, which differs from classical ITS analyses. Seasonality may have influenced outcomes, but it is unlikely that it explains the existence and direction of the effect. Quote: "Even though the dates of introduction are staggered, the effect is similar at each school suggesting that time of year is not a factor in PP selection"