The results from this investigation demonstrate the ubiquitous exposure to displays of chocolate, crisps, confectionery and soft drinks within Melbourne supermarkets. Such displays increase the frequency in a shopping trip that supermarket customers are exposed to snacks foods within Melbourne supermarkets and are designed to increase impulse-driven purchases [12-14]. Within the studied supermarkets, we found chocolate was the most prominent snack food item on display, appearing at the majority of checkouts and most frequently in island bin displays. Interestingly, while chocolate was the most common snack food item at the back-of-aisle displays, soft drinks (both diet and regular) were more common at the front-of-aisle displays. It is likely that this is because chocolate is already offered at the checkouts located opposite to the front-of-aisle displays. Each of the displays investigated here are considered dynamic in nature and are essentially independent of static aisle shelf displays.

Within supermarkets, consumers make a number of unplanned purchasing decisions, some of which are impulsive [10,14]. Individuals who make unplanned and impulsive food choice are more likely to consume lower amounts of healthy foods such as fruits and vegetables [20] while those with an impulsive personality trait are more likely to overeat and be overweight [23]. Food retailers themselves recognise that snack foods such as confectionery are frequently bought impulsively [12]. Where hunger interacts with impulsiveness [21] or in-store marketing/promotions exist [24], the purchase of snack foods is likely to be higher. Consequently, the display of snack food items in supermarkets is not random and relies upon a profit driven approach [12,13]. By providing snack food at the checkouts and broadening the range of snack food displays in different parts of the stores, retailers maximise the opportunity to sell snack food items on impulse [12,13].

While previous research has demonstrated similarly high numbers of snack foods at supermarket checkouts [13,16], we have also reported other potential incidental exposures within the store. This is an important distinction as our results demonstrate that at least in those supermarkets examined, the combination of snack food available in end-of-aisle, checkout, and island bin displays, in addition to the regular position in the aisle shelves, means exposure to snack-food displays within supermarkets is almost unavoidable. What remains to be ascertained is how these exposures are likely to impact on food purchasing decisions and health. Shelf space dedicated to snack foods was reported to be unrelated to socioeconomic differences in snack food purchasing in Australia (although this study may have been underpowered to detect differences) [19] while another US study reported small positive correlations between shelf space with BMI [25]. However, in another study, a greater variety of snack food items in supermarkets was unrelated to snack food consumption [26], highlighting a need for further research.

The presence of supermarkets in a neighbourhood has been linked to healthier eating and a lower weight status [27,28] and they are also likely to provide local employment opportunities. Despite these potential community-level benefits, our findings raise important questions about the role of product availability and placement within supermarkets in promoting healthy eating behaviours. The level of supermarket exposure to energy-dense, nutrient-poor snack foods is at odds with what is required to prevent further escalation (or even reversal) of current high obesity rates. 'Parents Jury' campaigns in Australia [29] and the UK [30] have called for the removal of confectionery items from checkouts within supermarkets. In response, food retailers in the UK took one of three approaches. Some were proactive in removing confectionery from all checkouts; some offered specific confectionery-free checkouts; and some resisted all calls to remove confectionery from their checkouts [12]. Although the removal of confectionery from checkouts gives the impression of a win for public health advocates, the reality is that retailers often use other prime locations for snack food displays (including the end-of-aisles) and since the removal of snack foods from checkouts, some evidence suggests that sales in the form of multipacks actually increased [12].

By including supermarkets from least and most socioeconomically disadvantaged neighbourhoods, we were able to account for whether the displays were socioeconomically patterned however we found no evidence of this. One plausible explanation is that the nature of the displays presented in this study are more likely to be dictated by other market forces such as promotions by snack food companies and that any variation by area-level disadvantage is more likely to be observed in the regular aisle displays [34]. Further, supermarket chains are also likely to use much more sophisticated indicators of the characteristics of the local area and potential customers, thus applying one indicator of area-level socioeconomic disadvantage may not be adequate to determine whether stocking practices differ between areas.

This study is strengthened by the use of an audit tool that captured a more detailed display of snack food availability than that measured in previous studies. By measuring checkouts, end-of-aisle displays and island bin displays, we captured the deliberate placement of dynamic displays that are designed to increase impulse purchases of snack foods made by customers. Whilst we undertook a small amount of test-retest reliability audits, we recognise that our analysis relies on a single within-store observation. Our reliability tests suggested some small variation in the products being offered in the end-of-aisle and island bins but on closer examination these variations were from one snack food product to another (e.g. from chocolate to soft drink) rather than from a snack food product to a non-snack food product. Further, a previous study from the US that examined produce within-store suggested that stores have at least short-term stability and that a single observation is often an accurate reflection of the stores' usual stocking practices [31]. The snack foods included in our study are not the only energy-dense, nutrient poor foods available to customers within supermarkets. A universal definition of 'snack food' does not exist [32,33], and for the context of this research, we limited our definition to food and beverage types that are often consumed outside of the three main meals and would be considered energy-dense, high in sodium and low in micronutrients.