An evaluation of the impact of a restrictive retail food environment intervention in a rural community pharmacy setting

Figure 1 shows weekly CSD sales from all three food stores in Baddeck from January 1, 2013 to May 8, 2015. Pre-intervention, the pharmacy accounted for approximately 6 % of total CSD sales ($CAD) in Baddeck. The two peaks in the graph demonstrate two CSD sales peaks in the summer months (June, July and August) in 2013 and 2014. Table 1 shows descriptive statistics of weekly CSD sales from each store. For all stores, weekly CSD sales were significantly lower in non-peak weeks relative to summer peak weeks.

Table 2 shows the results from the two final ARIMA models. Baseline weekly average CSD sales were $3105.10. Summer peak was significantly associated with higher weekly CSD sales at the community level. Model 1 illustrates that after controlling for summer peak, weekly CSD sales were $353.60 (11.4 %) lower in the community as a whole in the post-intervention period (p = 0.39). Model 2 shows that after controlling for model-specified seasonality, weekly CSD sales were $663.10 (21.4 %) lower at the community level, post-intervention (p = 0.13).

Table 3 shows the results from two ARIMA models (using Store 1 data and then Store 2 data) illustrating potential switching behaviours. No significant increase in CSD sales was observed in either of the stores.

This study examined the impact of a restrictive RFE intervention on weekly CSD sales at the community level. The analysis contributes to the limited literature on impact of restrictive interventions. We found that after controlling for seasonal variations in CSD sales, a restrictive retailer-led intervention banning sales of CSD in a community pharmacy was associated with a non-significant decline in sales of CSD at the community level in Baddeck, Nova Scotia. Another key finding was that there was no evidence of “switching behaviour”, that is, consumers did not buy more CSD from Store 1 or Store 2 after the pharmacy stopped selling CSD. This finding is meaningful from a population health intervention standpoint in particular, as it indicates that the pharmacy, rather than being a purposeful consumer destination for CSD purchasing, may actually have acted mainly as a source of impulse CSD purchasing within the community nutrition environment, as has been suggested previously [22].

Evidence on the impact of RFE interventions on diets and health has typically been generated by evaluations of permissive rather than restrictive interventions [16, 19]. Recent literature suggests mixed results of permissive interventions on healthy food purchasing [16, 25] and/or dietary intake [16, 17, 26]. While permissive interventions increase access to and availability of healthful options, such interventions may not be optimally effective without compatible efforts to restrict the availability of less healthy choices—producing an overall net food environment shift. This study contributes to the literature on restrictive RFE interventions, and finds a non-statistically significant impact of a pharmacy intervention in a small rural community.

Further research is required to analyze the myriad retail settings in which food and beverages are now sold. Industry analyses suggest that revenues of large chain pharmacies are split evenly between front-of store sales (e.g., foods and beverages, household items) and prescriptions, although smaller and independent pharmacies still generate most of their revenue through prescription sales [27]. The relatively minor proportion of revenue derived from the sale of sweet beverages at the study pharmacy facilitated the pharmacist’s decision to remove them without risking significant declines in revenues. Pharmacies may have an improved capacity to accommodate restrictive interventions as part of their business model, and may therefore be ideal settings in which to test these interventions. To the extent that such interventions can be established as profitable—or at least not a major financial risk—it will increase the potential for scaling up implementation among other small, independent food retailers.

Community pharmacies are health care settings that have been identified as an important health promotion setting given their geographic accessibility, diverse customer base, and the high level of public confidence in pharmacists [28‐31]. Programs that engage pharmacies as health promotion settings typically focus on community pharmacists’ potential role as nutrition counsellors [29‐32]. However, the dual roles of many pharmacists, first as health care professionals and second as business owners, can sometimes conflict. For example, the literature suggests that consumers [30] and nutritionists [31] perceive pharmacists who sell infant formula and bottles as having a conflict of interest in promoting breast-feeding. Consumers also perceive financial conflicts of interests when pharmacists recommend specific weight loss products they sell [29, 32]. Indeed, for community pharmacists, “both patient care and viability of the business need to be maintained” [32]. Therefore, interventions to promote healthy diets in pharmacies must also be economically feasible.

Financial conflicts of interest related to products sold at pharmacies are not new. Tobacco sales in pharmacies have spurred much debate over pharmacists’ role as health care professionals, given that tobacco use continues to be the leading cause of preventable death in the world [33]. Nine of 10 Canadian provinces prohibit tobacco sales in pharmacies [34]. While sugar-sweetened beverages are not typically regarded as being as harmful as tobacco products, sugar-sweetened beverage consumption is harmful to health [1, 2]. When CVS/Caremark announced it would stop selling tobacco in 2014, it did so based on carefully considered costs, including the normative and social costs of selling tobacco, which were high [34]. Future research should continue to monitor public opinion of sugar-sweetened beverages as a way of understanding normative and social risks for pharmacies that sell sugar sweetened beverages.

Several limitations of this study should be noted. First, we were unable to examine consumption phenomena at the individual/household level such as substitution effects. That is, we were unable to determine whether consumers who purchased less CSD ultimately increased their purchase of other types of beverages, including more or less nutritious beverages. Second, this analysis was also unable to distinguish between sales of diet and regular CSD. It is therefore unclear what proportion of the decline in sales of CSD was due to a reduction in sales of diet CSD. Industry analysis suggests that diet soft drinks have historically accounted for a small proportion of beverage sales [35]. It is therefore likely that observed declines were largely attributable to reductions in sales of sugar-sweetened CSD. Third, there was no control community in this natural experiment. The secular decline cited in industry documentation has not been as large as the declines observed in the current study, with total CSD sales volume declining approximately 3 % over same time period in the study, since 2013 [35]. Further, most food environment intervention studies have used an uncontrolled pre-post design since the interventions are often outside of the researchers’ control [19]. Finally, the sales data we obtained from the retailers did not include quantity of volume of CSD. This data limitation means that we were not able to detect the actual volume of CSD purchased. Considering tiered pricing strategies (i.e., CSD in larger-volume bottles cost less per ounce than CSD in smaller-volume bottles), if consumers switched to purchasing larger CSD bottles, our data would not reflect this phenomenon.

Despite these limitations, this study was among the first to examine the impact of a restrictive RFE intervention on CSD sales at the community level. An important strength of this study is that objective weekly CSD sales data were obtained from all retail stores selling beverages in the study community over a 123 week period. The strength of the ARIMA model in addressing both autocorrelation in the data as well as seasonal variation in weekly CSD sales is also notable.

This is a promising area of inquiry that would benefit from research directions that engage complementary disciplinary expertise. For example, determining the level of public support for the pharmacy intervention would be a first step to understanding the feasibility and acceptability of this type of intervention. The financial impact of voluntary prohibitions on the sale of sugar-sweetened beverages in pharmacies should also be examined. Finally, future research should further examine the extent to which declines in the sales of CSD translate into reduced CSD consumption at the individual/household level.

RFE, ARIMA