Background Supermarkets are a major source of food for families, yet greater understanding of how product placement strategies influence the healthfulness of food choice is needed. We assessed the effect of improving the availability and positioning of fruit and vegetables in supermarkets, and removing confectionery from checkouts, on household-level purchases and store-level sales.

Methods This study was a natural experiment with a prospective matched cluster design set in a discount supermarket chain in England. The intervention had two components: new fresh fruit and vegetable sections at store entrances (replacing smaller displays at the back), and removal of confectionery from checkouts (replacing with healthier items, such as sugar-free gum, water and non-food items) and end-of-aisle opposite checkouts. Women customers aged 18 to 45 years, with a store loyalty card, who regularly shopped at three intervention and three matched control stores in England were recruited. Data from purchases using loyalty cards and for store sales were collected for the three months before and six months after refurbishment. Individual purchasing data were analysed using a difference-in-difference method, while store sales data were analysed using controlled interrupted time series by store pair, with differences synthesised using random-effects meta-analysis.

Results A total of 107 women provided household-level purchasing data. The proportion purchasing fresh fruit and vegetables per week rose in intervention stores at three months (0.2%) compared to a drop in control stores (-3.0%) (P=0.22), and at six months (1.7% vs -3.5%, P=0.05). The proportion purchasing healthier checkout items rose more in intervention stores compared to a drop in control stores at both three (1.0% vs -1.8%, P=0.04) and six (0.6% vs -1.6%, P=0.13) months. There were no differences in the purchases of frozen fruit and vegetables or confectionery. Increases in store-level sales of fresh fruit and vegetables were greater in intervention stores than predicted at three months (1.71SDs (95%CI 0.45, 2.96), P=0.01) and six months (2.42SDs (0.22, 4.62), P=0.03). Decreases in sales of confectionery were greater in intervention stores than predicted at three months (-1.05SDs (-1.98, -0.12), P=0.03) and six months (-1.37SDs (-2.95, 0.22), P=0.09). Sales of frozen fruit and vegetables and healthier checkout items showed no differences.

Conclusion This study provides some evidence to suggest that healthier product placement in supermarkets improves the healthfulness of both household-level purchasing and store-level sales. Improving fruit and vegetable placement should be considered by government alongside current plans to limit prominent placement of unhealthy foods.