Hua 2017.

Methods	Design: Cluster‐RCT Timing: Prospective Allocation to group: Quote: "The researchers used a random number generator to randomly assign vending machines into one of eight different conditions in a 2×2×2 factorial design that manipulated availability of healthier products, pricing, and/or promotional signage" Number of clusters or sites: 28 beverage‐vending machines (in addition to 28 food‐vending machines, which do not contribute data to the results included in this review) Number of individuals: N/R (the analysis is based on anonymous transaction data not linked to individual participants) Length of intervention: 5 months
Participants	General description of participants: Students, staff and visitors of Yale University, USA, using beverage‐vending machines located in dormitories, libraries, administrative buildings, and department buildings with classrooms Age: All ages (no age details reported) Inclusion criteria: Quote: "Snack and beverage machines that were co‐located in a single location were included" (no further details provided) Exclusion criteria: N/R Recruitment: N/A Weight status at baseline: N/R SSB consumption at baseline: Not reported for individual participants, but the study reports that "[t]he best‐selling beverages preintervention were Diet Coke (Coca‐Cola Company) (20 oz and 12 oz), Coke (Coca‐Cola Company) (20 oz and 12 oz), and Coke Zero (Coca‐Cola Company) (20 oz)" Equity considerations: Not reported explicitly, but the study's primary report seems to imply that the intervention took place at Yale University in New Haven, Connecticut, implying a socio‐economically privileged sample
Interventions	Intervention: Vending machine redesign, including improved availability and lower prices for healthier beverages and promotional signs. (Quote: "Beverage guidelines were adapted from the New York City Agency Food Standards guidelines for vending machines. Healthier beverages included water and other beverages with 25 kcal/8 oz (eg, unsweetened iced tea or diet carbonated beverages). Based on New York City standards, these beverages made up at least 75% of available products, with the remaining 25% of products permitted to include regular, 8‐oz sodas placed in the bottom row. (…) Beverage machines randomized to the price intervention sold water for $1 (instead of the prestudy price of $1.50 to $2). (…) [B]everage machines randomized to the promotional sign intervention received one or two of two stickers, depending on condition (...): '$1 Water' or 'Rethink your drink'") Behavioural co‐intervention: None reported Control: No intervention
Outcomes	Measures of SSB intake: Change in best‐selling beverages, assessed continuously throughout the study period through routinely‐collected sales data Measures of intake of alternatives to SSB: Change in best‐selling beverages, assessed continuously throughout the study period through routinely‐collected sales data Anthropometric measures: N/R Adverse outcomes: Target group discontent in the case of 2 vending machines (no further details provided on how and when this outcome was assessed) Other outcomes: Revenue made in vending machines, and total number of products sold, assessed continuously throughout the study period through routinely‐collected sales data
Context and implementation	Setting: University campus Sector: Higher education Country: USA Year(s) when implemented: 2015 Mode of implementation: Pilot trial by researchers Level of implementation: Setting‐based intervention
Declarations	COI: "No potential conflict of interest was reported by the authors" Funding: "Funding/Support: None to report" Trial registration: N/R Protocol availability: N/R
Notes	None
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "The researchers used a random number generator to randomly assign vending machines into one of eight different conditions"
Allocation concealment (selection bias)	Low risk	Allocation was by vending machine, and allocation was done at the start of the study
Similarity of baseline outcome measurements (selection bias)	Unclear risk	Baseline outcome measurements differed substantially across vending machines, and no means for study groups are reported. It is unclear if this was taken into account in the analysis
Similarity of other baseline characteristics (selection and performance bias)	Unclear risk	All vending machines were served by the same provider, but apart from the outcome measurements no other baseline characteristics are reported
Incomplete outcome data (attrition bias) All outcomes	Low risk	Only 2 (out of 56) vending machines dropped out due to customer dissatisfaction, but were included in the intention‐to‐treat analyses. Sales data from the vending machines were probably close to complete. It seems unlikely that incomplete outcome data substantially influenced the results
Blinding (performance and detection bias) Objective outcomes	Low risk	Outcomes were objective (routinely‐collected sales data)
Contamination (performance bias)	High risk	All vending machines were located on the same university campus, and it is likely that vending‐machine customers, who were students, faculty and staff, did not always use the same vending machine, and were exposed to all 3 interventions in varying degrees. In particular with regard to the promotional signs and the re‐organisation of the beverage‐vending machines, this might have biased results towards null
Selective reporting (reporting bias)	High risk	Only outcomes for healthy beverages (but not for unhealthy beverages) are reported, even though it can be assumed that data on those were also routinely collected through the sales data system of the vending machine supplier
Other bias	Low risk	No other concerns