The health benefits and cost-effectiveness of complete healthy vending

Vending machines contribute to growing levels of obesity. They typically contain energy dense, high fat snacks and attempts at persuading consumers to switch to healthier snacks sold within the same machine have had limited success. This study explored the health benefits and cost effectiveness of the complete replacement of regular snacks with healthy items. Two vending machines were manipulated in a 6-month trial, with a healthy and regular range of products alternated between the two machines every fortnight. Healthy vending resulted in a 61% drop in calories sold relative to regular vending, significant with time and product range as random factors. There was no evidence of compensatory behaviour from nearby shop sales nor in multi-item purchases from vending machines. The impact on profit was less clear. Sales dropped by 30% during healthy vending but variability across product range meant that the change was not significant. Overall our results demonstrate that complete healthy vending can be introduced in hospitals without a catastrophic loss in sales nor compensatory behaviours that offset the public health gains of consuming healthier products.

Introduction

Vending machines are machines of convenience. Offering 24-hour access to food and drink in public spaces such as schools, hospitals and workspaces, they are many people’s first choice when searching for a quick pick-me-up. It is therefore unsurprising that vending machines are predominantly stocked with quick-fix foods that are high in energy, but low in nutritional content. This has led some to describe vending machines as obesogenic food environments [1]. While the financial benefit of vending machines is undeniable, so too is their contribution to the obesity epidemic currently being evidenced across the globe [2]. With vending machines so widely dispersed across all manner of settings, addressing their contribution to the composition of the food environment is important for reducing obesity levels. In this study, we investigate one simple approach: the complete replacement of standard products typically considered unhealthy (i.e., high calorie, high fat products, such as chocolate bars) with healthier options. (i.e. products low in salt and added fat/sugar).

Previous research has predominantly focused on environments in which healthy and regular products are both available and consumers have the choice of which to purchase. The goal has been to find ways of increasing the proportion of healthier choices made by consumers. We divide this literature into two general approaches. The first is to make healthier choices more appealing using labelling or promotional material. Two studies have shown some success in this respect. French et al. [3] applied health labels and health-promoting signage to 55 vending machines over 12 months. Low-fat sales accounted for 14.3% of all sales in the no label condition, but 15.4% with labels and signs. The increase was significant across vending location. In a school-based randomised control trial, Kocken et al. [4] increased the availability of healthier options and used hand signals to label products as either being favourable (thumbs up– 99 calories or less), moderately unfavourable (index finger– 100–170 calories) and unfavourable (thumbs down– 171 or more calories). The intervention led to moderately unfavourable products accounting for 26.6% more of all sales, and unfavourable products accounting for 27.6% less of all sales. Sales of favourable products were unchanged, with less than 2% of all sales coming from such items. On the whole however, promotional messages appear to have limited success. A review of the use of promotional materials and labels providing health information in vending machines [5], found that effects were either absent or minimal across the eight studies considered. It may well be that individuals do not read labels, and if they do, the majority may not find their messages to be sufficient motivation for changing food choice [6].

The second approach is to alter the relative costs of the choices, either by taxing regular products or by subsidising healthy products. Subsidising healthier products has a robust positive impact, with greater reductions in price leading to higher proportions of healthy sales [5]. For example, French et al. [3], described above, applied price reductions of 10%, 25% and 50% on low-fat products, and observed significant increases in sales of healthy products of 1.0%, 4.2%, and 10.1% respectively, and with no loss in revenue. Taxing unhealthy options has been tested far less extensively but existing evidence shows similar benefits to subsidising healthy options. A 25% tax on unhealthy items was tested in a 14-month trial conducted across three vending locations [7]. In addition to the tax, the researchers tested a 25% discount on healthy items, a 25 second delay in the vending time of unhealthy options and a combination of time delays and pricing interventions. All interventions successfully increased the proportion of healthy choices made and the most effective was the 25% price tax on unhealthy products, which increased the proportion of healthy choices made within the machine by 13.6%.

Substantial health gains can be made by manipulating price and other factors within mixed vending machines [3, 5, 7]. Nonetheless, more impressive changes in health behaviour might be obtained by removing regular products entirely and leaving only healthy products. If people have no choice about choosing healthy products, they will surely become healthier. Prior studies adopting the complete replacement approach are few in number but those that exist suggest some difficulties. The first is that consumers may go elsewhere to search for unhealthy products. While vending machines follow certain healthy eating restrictions, other vendors do not. If consumers are frustrated or dissatisfied with the offering in the machine, they may search for alternatives at other nearby outlets. While vending would be 100% healthy, behaviour would not necessarily be so. This problem is illustrated by studies examining the effect of removing sugared soft drinks from vending machines in schools. Taber, Chriqui, Frank and Chaloupka [8] measured the association between vending access to soft drinks and consumption in 10,000 schools across the United States. They found that overall consumption of soft drinks was higher in children where vending of soft drinks was prohibited. Taber and colleagues [8] suggested that without a whole-environment change to soft drink supply, children were able to obtain drinks elsewhere, where they overcompensated for the removal of soft drinks from the school vending machines. In the same vein, evidence from another school study shows that while soft drink consumption on school premises decreased where vending machines did not sell soft drinks, overall consumption (including off premises) did not differ across groups [9]. This highlights the need to evaluate the effectiveness of vending interventions not only based on what happens inside the machine, but also on what happens outside of it (c.f. [3, 7, 10]).

A second concern for interventions that completely remove unhealthy products is that consumers may buy more healthy products to replace the perceived loss of satiation that occurs from not consuming unhealthy products (e.g. buying two healthy bars to replace one unhealthy bar). There is no evidence for this compensatory behaviour but nor are there any vending studies that have tracked the behaviour of individual consumers in real-choice scenarios. Data showing an increase in the sales of healthy products (e.g. with price reductions in French et al. [3]) are consistent with fewer individuals making more multi-item purchases. Furthermore, the advent of credit card readers in vending machines removes some of the impediments to multi-item purchasing (e.g. less need for coinage, less time needed per transactions), and so multi-item purchases might be more frequent now than in the past. If multi-item purchasing is more common in machines that stock only healthy items, behaviour may not necessarily be healthier.

A final concern with complete healthy vending is that the loss in revenue may be unsustainable. Balancing cost and public health benefit is a common concern for health interventions [11, 12]. Catering departments may be reliant on vending machine revenue, and any action that would jeopardize these profits may be approached with reluctance. Previous vending studies have not observed a drop in profit during healthy vending interventions (e.g. [3, 4, 7]). However, the interpretation and relevance of these studies with respect to completely healthy vending is complicated. First, previous studies have all used partial healthy vending. Particularly determined consumers have always had the option of choosing their regular products if they wished. As noted, if regular products are not available, consumers may switch to purchasing elsewhere or to purchasing nothing at all. Complete healthy vending would suffer a drop in revenue from these consumers whereas partial healthy vending would not. Second, some partial healthy vending interventions have included salient price discounts on healthy products [3, 4, 7]. The discount has the effect of increasing sales volume by creating extra demand, thereby offsetting any drop in demand for the regular product (and increasing the total number of calories consumed). In complete healthy vending, any discount applied to healthy products would not be salient because there would be no non-discounted products with which to compare it. Discounting healthy products within complete healthy vending may not be a viable option for balancing revenue. Finally, the product range used across studies varies considerably and random effects of product range have not been included in statistical analyses. While these studies have demonstrated that their interventions are both health and cost effective within their product ranges, it is difficult to know whether effects on sales volume (or any other dependent measure) would replicate in environments with different product ranges.

In summary, complete replacement of unhealthy products may have the potential for large health improvements. Nonetheless, it may result in problematic compensatory behaviours, and implications on revenue are unknown. In this study we report a controlled experiment to test the health and cost implications of total healthy vending. Sales data was collected from vending machines located in a hospital environment. Stock was varied fortnightly between a regular (unhealthy) range and a healthy range over the course of six months. To assess healthy behaviour, we measured calorific content of products sold, and to measure sales, we analysed sales volume and cost/profit. The analysis for both used linear mixed models with time and product range as random effects. Potential compensatory behaviours were also explored. To establish whether consumers would search elsewhere for regular snacks during healthy vending periods, we analysed sales data from the nearby convenience shop, with the hypothesis that unhealthy sales would increase at the shop during healthy vending. Finally, we used credit card data and on-site observations to test the hypothesis that individuals would compensate for the lack of unhealthy options available at the machine by making more multi-item purchases during healthy vending periods than regular periods.