Prompting consumers to make healthier food choices in hospitals: a cluster randomised controlled trial

Background: Hospitals in the UK offer snacks for sale to patients, staff and visitors. Despite the NHS's health promoting role, and tightening of regulations around which foods can be sold in hospitals, many snacks purchased in this setting are unhealthy. The present project tests the effectiveness of theory-based point of purchase prompts (PPPs; a form of cognitive nudge) designed to make it cognitively easier for consumers to compare available products and choose healthier options. Methods: Hospital shops in Scotland (n = 30) were recruited into a cluster randomised controlled trial to test whether a PPP could reduce the average calorie, fat and/or sugar content of purchased snacks. Inclusion criteria stated that eligible sites; sold food; were located in a hospital; and were accessible to staff, patients and visitors. The PPP intervention was a theory-based sign (tailored to the available range in each location) designed to cognitively simplify healthier snack choices by facilitating cross-product comparison. Shops were randomised to display PPPs (intervention;n = 15) or not (control;n = 15) using block randomisation controlling for shop size. Data on all snacks purchased from participating shops were obtained from retailers for a 12-week baseline and 12-week follow-up period. Primary outcomes were the average calorie (kcals), fat(g) and sugar(g) content of snacks purchased each day. Secondary outcomes were the average customer spend per item purchased ( pound,p) and the total number of snacks purchased daily. Shop staff were not blinded to group assignment but data providers were. Data were analysed using mixed effects multi-level regression models. Results: Data from > 1 million snack purchases were analysed. Snacks purchased from intervention sites were on average significantly lower in calorie (gamma = - 1.84,p < .001) and sugar (gamma = - 0.18,p = .030) at follow up relative to baseline but only the reduction in calories was significantly different to control. Average spend per item also reduced significantly in intervention (but not control) sites (gamma = - 0.89,p < .001). The intervention had no effect on the fat content of snacks or the number of snacks sold. Conclusions: Simple, theory-based point of purchase prompts can produce small but statistically significant reductions in the energy content of snack purchases from hospital shops.