Altering the availability of healthier vs. less healthy items in UK hospital vending machines: A multiple treatment reversal design
Pechey R., Jenkins H., Cartwright E., Marteau TM.
© 2019 The Author(s). Background: Altering the availability of healthier or less-healthy products may increase healthier purchases, but evidence is currently limited. The current study aimed to investigate the impact of altering the absolute-and-relative availability of healthier and less-healthy products - i.e. simultaneously altering the number of options available and the proportion of healthier options - in hospital vending machines. Methods: An adapted multiple treatment reversal design was used, altering products available in ten vending machines serving snack foods and/or cold drinks in one English hospital. Machines were randomised to one of two sequences for the seven 4-week study periods: ABCADEA or ADEABCA. In Condition A (study periods 1, 4 and 7) the proportions of healthier products were standardised across all machines, so that 25% of all snack slots and 75% of drink slots were healthier. In Condition B, 20% of vending machine slots were emptied by removing less-healthy products. In Condition C, the empty slots created in Condition B were filled with healthier products. Conditions D and E were operationalised in the same way as B and C, except healthier products were removed in D, and then less-healthy products added in E. Sales data were obtained from machine restocking records. Separate linear mixed models were conducted to examine the impact of altering availability on energy purchased (kcal) from (i) snacks or (ii) drinks each week, with random effects for vending machine. Results: The energy purchased from drinks was reduced when the number of slots containing less-healthy drinks was decreased, compared to standardised levels (- 52.6%; 95%CI: - 69.3,-26.9). Findings were inconclusive for energy purchased from snacks when less-healthy snack slots were reduced (- 17.2%; 95%CI: - 47.4,30.5). Results for altering the number of slots for healthier drinks or snacks were similarly inconclusive, with no statistically significant impact on energy purchased. Conclusions: Reducing the availability of less-healthy drinks could reduce the energy purchased from drinks in vending machines. Further studies are needed to establish whether any effects might be smaller for snacks, or found with higher baseline proportions of healthier options.