This study investigated the exposure to energy-dense snack foods (chips, chocolates and confectionery) and soft drinks in a sample of supermarkets across selected cities in eight developed countries. The shelf length of snack foods and the presence of these foods at checkouts and in end-of-aisle displays were assessed with noticeable variations detected between countries. UK supermarkets had the greatest aisle length devoted to chips, chocolate and confectionery, while soft drink aisle length was greatest in Australia. The proportion of both checkouts and end-of-aisle displays containing snack food was also highest in the Australian supermarkets sampled. In every country other than the Netherlands, snack foods were present at over 70% of checkouts.
From the results of this cross-sectional study, we are not able to discern whether variation is a result of differences in demand between countries, or whether supply is driving demand. It is likely that both are contributing. Demand may be driven strongly by cultural norms and traditional diet preferences. Influences on the supply side of the equation may include specific commercial arrangements between retailers and the food industry/suppliers, climatic differences between countries that influence the availability and price of supermarket items, agricultural policies such as the subsidisation of high-fructose corn-syrup production (for use as a sweetener in soft drinks) by the US government , import tariffs and trade agreements. The position of supermarkets in the larger food shopping environment may also explain some of the variation observed here. For example, it is likely the role of supermarkets in supplying fresh fruit and vegetables to consumers varies between nations based on the prominence and use of other retailers such as greengrocers or markets. Such variations may have contributed to the differences in the ratio of snack foods to fruits and vegetables that we observed.
Some of the findings reported here can be compared with the results of previous studies from individual countries. A 2006 study from Melbourne, Australia reported that 99% of checkouts displayed snack foods or soft drinks  which is consistent with our own findings from that city. Prior US research [46, 50] using a more inclusive definition of snack foods (included such items as nuts, cookies, doughnuts) found greater shelf lengths dedicated to snack food items than we report here. Because of the differences in definitions, the results from this study and our own study are not directly comparable.
Whilst the broader dietary and health implications of supermarkets have been discussed elsewhere [29, 30], the link between snack food availability and both purchasing behaviour and health indicators warrants consideration. Although research in this area is in its infancy, a couple of studies have been published. One prior Australian study did not find any link between snack food shelf length and purchasing ; however that study of only nine supermarkets was underpowered to detect a significant effect. In the US, a positive correlation between snack food shelf space and BMI was observed  however the effect size was small. Many policy and program interventions are already aiming to change food environments even though robust and consistent evidence is not yet available and a consensus on how change should best be achieved has not yet been reached . Additional studies linking within-store environments with purchasing habits, diets and obesity are therefore sorely required to support the obvious desire to improve our food environment.
While ecological data of the type reported from this study cannot infer causality, links between availability and national consumption patterns are of interest. In Dutch supermarkets, greater shelf space was allocated to soft drinks in comparison with the other snack foods assessed. This result correlates with the findings from the ENERGY study (which examines health in children across seven European countries) in which extremely high soft drink consumption amongst Dutch children was reported . Furthermore, soft drink consumption in the Netherlands has increased by 74% between 1980 and 2009 . In the North American supermarkets audited, shelf space devoted to soft drinks and potato chips was greater than for confectionery and chocolate in comparison with the other countries assessed. Within the US, soft drinks (soda) were reported to be the top dietary source of added sugars [57, 58] whilst potato chips were the top dietary source of oils . Amongst U.S. children both soft drinks and potato chips make substantial contributions to overall energy intake . The Swedish population has traditionally had a preference for sugar confectionery . Of interest however, are changes between 1980 and 2010 in the consumption of different snack foods and drinks. In that period, chocolate and confectionery (combined) increased by 53% per capita in Sweden whilst the consumption of soft drinks (including flavoured carbonated water) over this period increased three-fold and consumption of potato chips increased four-fold . These changes in consumption are reflected in the current snack food profile in Swedish supermarkets in which shelf space of confectionery and chocolate is no higher than for chips and soft drinks (noting that our measure does not include carbonated water). Although it is of interest to examine correlations between availability and consumption patterns, the lack of comparable national dietary indicators limits our ability to explore this in more detail.
Whilst limiting snack food exposure in other settings such as schools and workplaces has been a focus of some public health campaigns [61, 62], the supermarket environment is increasingly recognised as a potential intervention point . In addition to facilitating comparisons between countries, the results of this study also allow the assessment of the local food environment in each country. Such national food environment data is necessary to support and justify campaigns such as those calling for the removal of confectionery items from supermarket checkouts [61, 63, 64]. Efforts to improve diet and reduce obesity and other chronic diseases will be more successful when supported by strategies such as these that aim to create healthier environments. In the Netherlands, the Albert Heijn supermarket chain has the largest market share (34%)  and remains a profitable supermarket retailer despite not having high levels of snack food displays (relative to other supermarkets in this study). In reaction to a report by the Dutch Consumers’ Federation  on which supermarkets make the healthy choice the easy choice, Albert Heijn announced an initiative to remove all snack foods from checkouts. This action is reflected in the low proportion of Dutch checkouts with snack food displays observed in this study. That example suggests that such initiatives may involve a relatively low cost for supermarkets, can allow them to promote their brand as a healthy choice, and may be important in changing cultural norms around snacking behaviours. Other forms of advocacy to encourage healthier supermarkets are required  and these may include policy-level approaches initiated by governments.
The major strength of this study is the within-store assessment of energy-dense snack foods and soft drinks conducted in a sample of supermarkets across cities in multiple countries using a standardized measurement tool. Multiple aspects of the within-store food environment were captured, including static displays of shelf space and dynamic displays at ends-of-aisles and checkouts. Previously, within-store assessments of supermarket snack foods have been rare, limited in their ability to compare findings between countries and did not include the multiple aspects of the supermarket snack food environment assessed here.
We acknowledge that our definition of snack food was limited to four food categories. Other types of energy-dense snack foods not captured here are also available in supermarkets and in certain contexts these products may also be important snack foods (e.g. cookies and ice-cream in North America). While no universal definition of snack foods exists [51, 52], the definition used in this study was appropriate for a cross-country comparison as each of the snack foods examined were commonly available in all countries. The sampling strategy should also be considered when interpreting the findings. In Melbourne, Amsterdam and Montreal, data collection was undertaken according to area level disadvantage. Auditors in other countries sampled from a representative range of urban areas with no pre-specified stratification according to area-level disadvantage. All collaborators were instructed to sample from the major supermarket retailers in their setting. Although this meant that a greater variety of chains were sampled in some countries than others, we would expect that the diversity of store types in a country is a valid reflection of the choices available to consumers. The fourth limitation relates to the relatively small number of supermarkets examined in two of the countries (New Zealand and the United Kingdom). The findings from these two countries in particular should be treated with some caution as they may not accurately represent snack food availability in those settings. Despite adjusting for store size in our analysis, we did not have any indicator of the presence of non-food items present and it is possible that in some larger stores, a greater proportion of the store was allocated to such products. We did, however include an assessment of the ratio of snack foods to fruits and vegetables which is an effective indicator of the priority given to snack food relative to other items. Finally, other factors that may be important determinants of snack food purchasing (e.g. price, in-store promotions, variety, island bin displays) were not included in this study because of difficulties in comparing such features between countries.