The study will adopt a stepped wedge randomised controlled trial (SW-CRT) design involving 42 hot beverage vending machines in 27 buildings (clusters). Clusters will be randomly allocated to one of eight sequences across 24 weeks, which will dictate the time at which clusters implement the vending machine PACE labelling intervention (condition). Weeks one to four will constitute a baseline period, during which the data will be collected from all vending machines without PACE labelling in place. This will be followed by 20 weeks during which the PACE labelling intervention will be introduced and maintained at each building according to the SW-CRT design. Specifically, every two weeks, three clusters will move from no PACE labelling to the intervention condition.
All vending machines have the same seven hot beverage options (hot water, espresso, black coffee, flat white coffee, cappuccino, mocha, latte macchiato and hot chocolate). When vending machines are randomised to display the PACE labelling, they will display sticker labels showing the number of calories and minutes walking (averaged speed) to expend the calories of each beverage.
Data for the number and type of hot beverages transactions will be collected every week or two weeks from each vending machine. Photograph date stamped evidence of the number of transactions beverages will be collected.
The primary outcome will be averaged kilocalories (kcals) per transaction. We will evaluate the proportion of total drinks dispensed that are high calorie, and so each beverage dispensed will be dichotomised as being high-calorie or not high-calorie. It is possible people choose not to have a beverage at all in response to the intervention. A secondary aim is to compare total number of transactions of beverages between the PACE labelling and comparator condition.
Randomisation will be performed at the cluster (building) level. A total of 27 buildings will be randomised to one of eight sequence using covariate constrained randomisation (CCR) to ensure a balance of several cluster-level covariates. The covariates included in this constrained randomisation procedure will be: type of building, number of vending machines in each building/cluster and average baseline (weeks 1-4) mocha transactions per cluster (most caloric beverage) (defined as low, medium and high). Three days prior to the next batch of clusters transitioning to the intervention condition, the trial statistician will reveal the subsequent vending machines that have been randomised to display the PACE labelling in line with the study plan. The allocation sequence will not be known to researchers except for the trial statisticians.
