This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Research Protocols, is properly cited. The complete bibliographic information, a link to the original publication on http://www.researchprotocols.org, as well as this copyright and license information must be included.
Discretionary food and beverages (products high in saturated fat, added sugars, and salt) are detrimental to a healthy diet. Nevertheless, they provide 42% of total energy and account for 53% of food and beverage expenditure for remote living Aboriginal and Torres Strait Islander Australians, contributing to the excessive burden of chronic diseases experienced by this population group.
The aim of this study is to test an intervention to reduce sales of discretionary products, in collaboration with the Arnhem Land Progress Aboriginal Corporation (ALPA), which operates 25 stores in very remote Australia, by reducing their merchandising and substituting with core products in remote Australian communities.
We will use a community-level randomized controlled pragmatic trial design. Stores randomized to the intervention group will be supported by ALPA to reduce merchandising of 4 food categories (sugar, sugar-sweetened beverages, sweet biscuits, and confectionery) that together provide 64% of energy from discretionary foods and 87% of total free sugars in very remote community stores. The remaining stores (50% of total) will serve as controls and conduct business as usual. Electronic store sales data will be collected at baseline, 12-weeks intervention, and 24-weeks postintervention to objectively assess the primary outcome of percent change in purchases of free sugars (g/megajoule) and secondary business- and diet-related outcomes. Critical to ensuring translation to improved store policies and healthier diets in remote Indigenous Australia, we will conduct (1) an in-depth implementation evaluation to assess fidelity, (2) a customer intercept survey to investigate the relationship between customer characteristics and discretionary food purchasing, and (3) a qualitative study to identify policy supports for scale-up of health-enabling policy action in stores.
As of August 2018, 20 stores consented to participate and were randomized to receive the intervention or continue usual business. The 12-week strategy ended in December 2018. The 24-week postintervention follow-up will occur in May 2019. Trial results are expected for 2019.
Novel pragmatic research approaches are needed to inform policy for healthy retail food environments. This research will greatly advance our understanding of how the retail food environment can be used to improve population-level diet in the remote Australian Aboriginal and Torres Strait Islander context and retail settings globally.
Australian New Zealand Clinical Trials Registry ACTRN12618001588280; http://www.anzctr.org.au/Trial/Registration/TrialReview.aspx?id=375933 (Archived by WebCite at http://www.webcitation.org/76dbQEmwN)
DERR1-10.2196/12646
Aboriginal and Torres Strait Islander people residing in remote Australian communities bear a disproportionate burden of preventable chronic diseases [
The majority of food consumed in remote communities is purchased at the local community store [
We recently led the “stores healthy options project in remote indigenous communities (SHOP@RIC)” study, a large trial with 20 remote community stores to assess the impact of a price discount on purchasing [
Merchandising is the “activity of promoting the sale of goods, especially by their presentation in retail outlets,” and it includes activities such as display techniques, free samples, pricing, shelf talkers, and other point-of-sale methods [
Our research will increase the understanding of merchandising as a factor influencing dietary behavior in consumer food environments [
Our study objectives are to:
assess the impact of reducing discretionary product merchandising on customer purchasing and retail business performance in remote Indigenous communities,
identify characteristics of customers associated with discretionary product purchasing, and
analyze and characterize the policy supports needed to scale-up nutrition evidence uptake in retail stores in remote Indigenous Australia.
This study tests the hypotheses that, over a 12-week intervention (and at 24-weeks postintervention), a strategy designed to reduce the merchandising of target discretionary products will reduce grams of free sugars per megajoule (MJ) energy (ie, sugars added to products plus sugars naturally present in honey, syrups, and fruit juices) in foods and drinks purchased through the community store (our primary outcome measure), and it will have positive impacts on secondary outcome measures relating to business performance and diet. This paper will describe the detailed protocol for study aims 1 and 2, and it excludes the policy analysis, aim 3.
Aboriginal and Torres Strait Islander people represent 3.3% of the Australian population [
ALPA is Australia’s largest Indigenous corporation [
We will use a community-level pragmatic randomized parallel group, 2-arm, superiority trial with a 1:1 allocation ratio design and a baseline, 12-week intervention period and 24-week postintervention period to assess the effect of the intervention on customer purchasing and store business performance measured objectively through store sales data (
Study design schema.
Our intervention has been co-designed with ALPA, and it will reduce the merchandising of high sugar discretionary products and subsequent desirability of these products, while allowing for substitute merchandising of core foods. Our overarching aim is to reduce the purchasing of targeted discretionary items. Due to the unknown impact of the intervention on business outcomes including level of resources needed for full implementation, ALPA considered a 12-week intervention period as acceptable. We previously reported short-term (ie, less than 6 months) food price interventions to be effective when applied in stores and/or supermarkets [
The strategy is informed by a social ecological theory that poses that behavior is shaped by interaction between the individual and the environment [
The implementation intervention logic is based on the Behavior Change Wheel [
Ethical approval has been granted by the combined NT Department of Health and Menzies School of Health Research Human Research Ethics Committee (ref: HREC-2018-3048) and the Far North Queensland Human Research Ethics Committee (ref: HREC/18/QCH/23-1211).
This research will be conducted in partnership with ALPA who, at the time of designing the study, managed 25 stores in 24 communities. The ALPA stores comprise 3 corporation types, as these are as follows: company owned stores (6 in Queensland), stores managed on behalf of Aboriginal store owners (12 in NT and 1 in Queensland), and member stores where community residents are the shareholders (6 in NT). Overall, 12 of these stores operate in communities where there is more than 1 store. All stores managed by ALPA will be eligible. Using stores managed by a single store association will help ensure uniform and high-level implementation fidelity. In the 1 community where there are 2 ALPA stores, both stores will be allocated together to intervention or control during randomization.
The recruitment process will commence with the study being first presented and discussed with the ALPA and the Island and Cape boards. These boards are able to give consent for the ALPA and Island and Cape owned stores. On approval from the ALPA board, contact will then be made with each of the stores in NT with a management agreement with ALPA and a meeting arranged with the respective store boards. A study story will be used to facilitate a face-to-face discussion with store board directors. This will be facilitated by ALPA personnel alone or together with a member or members of the research team. Local authority groups in each of the communities will be informed of the study via a letter describing the study purpose and indicating that the community store board will be invited to participate.
Consenting stores will be allocated to intervention or control using random number ranking, Stata version 15, StataCorp LLC.
Blinding. Blinding of store allocation is not possible for store managers, ALPA personnel responsible for intervention implementation, research staff administering the intervention, or customers and other community members. This will not impact the data collected for the primary and secondary outcomes, as the store sales data used to measure intervention effect are objective electronic data. In addition, objective measures in the form of photographs will be used to assess implementation fidelity.
Through our co-design approach with ALPA, the intervention (
A traffic light system based on the NT School Nutrition and Healthy Eating Guidelines [
The ALPA General Manager of Retail Services and ALPA Nutritionist, who are both study investigators, will use a store task list, and they will use a drinks fridge planogram (physical layout diagram) developed by ALPA with input of the study investigators on proportion of unhealthier to healthier drinks, to communicate to store managers and their area managers about what is required. The task list will be developed on the basis of merchandising practice observed and photographed for each intervention store at baseline, specifically tailored on the basis of each store’s unique premium high traffic areas such as counter, front- and end-of-aisles, and store entrance. The task lists will indicate the maximum number of shelf facings for sugar, sugar-sweetened beverages, sweet biscuits, and confectionery product categories, with a list of substitutable core food products to fill the space opened. Substitutable core products will be identified with ALPA and indicated to store managers in a 2-page reference guide. Fidelity to no price reductions on targeted products will be maintained from baseline, with no changes during the intervention period, via ALPA’s standard pricing procedures. ALPA will lead the implementation of the intervention and will support the training of its store staff in implementation procedures for each of the intervention stores at start-up. The store will be relayed (ie, implementation of strategy) in consultation with store managers by members of the research team with an ALPA area manager or the ALPA nutritionist with an ALPA area manager. Store managers will use photographs of the newly relayed store to communicate stocking procedures to store staff for strategy maintenance. Thereafter, store managers will be responsible for maintaining the intervention with the support of their area managers. We expect some local adaptation of the intervention because of heterogeneity in the physical design of each store and input on implementation from store directors, but ALPA will aim for standardization of the intervention delivery across stores. Strict monitoring of intervention compliance will be conducted by the research team (see the Implementation Evaluation section below). Any nonadherence identified through monitoring will be communicated immediately to the responsible store manager to correct.
Intervention components.
Targeted discretionary items (all red table sugar, sugar-sweetened beverages, sweet biscuits, and confectionery; price, promotions, and place) | Healthier alternatives (all green and amber items; price, promotions, and place) |
1. No promotional activitya on discretionary products, including no price discounts, volume promotions (eg, 2 for 1 type offers), posters, shelf stripping, and fridge branding | —b |
2. No misleading promotional activity (eg, fruit and vegetable fridge branding on a fridge containing confectionery, or no sugar shelf stripping on shelves with sugary drinks) | No misleading promotional activity (eg, fruit and vegetable fridge branding on a fridge containing confectionery, or no sugar shelf stripping on shelves with sugary drinks) |
3. No visible availability at counter and high traffic areasc of discretionary products (eg, front-and end-of-aisle displays) | Substitute visible availability of core products |
4. Reduced facingsd (ie, number of identical products on a shelf) | Substitute facings of core products in the proximity of targeted product categories where the facings have been reduced |
5. Reduced refrigerator space for targeted drinkse | Substitute refrigerator space for healthier drinks as follows: water, small units of unsweetened fruit juice, and artificially sweetened beverages |
6. In stores with no non-Arnhem Land Progress Aboriginal Corporation competitor store in the community, no units more than 600 ml of targeted soft drinks permitted in refrigerators | — |
7. Shelf stripping warning on target products and floor sticker indicating quantity of sugar in drinks | Floor sticker promoting water as the healthiest drink choice |
aPrice mark-downs with no signal of savings to the customer permitted on short-dated food and drink stock. Marked-down items not permitted in high traffic areas.
bNot applicable.
cProducts considered at high risk of theft to remain at front of store, but in the least prominent location such as under the counter.
dSweet biscuit facings reduced by half; table sugar facings reduced to 1 bay, no multipacks displayed, smaller units at eye level; and confectionery facings reduced by half and no increase in range permitted.
eArtificially sweetened drinks (diet drinks) were classified as amber, not red.
Control stores will be asked to continue usual store practice and told by ALPA that they will be supported to implement the Healthy Stores 2020 strategy at the study end if demonstrated to be effective.
Decisions relating to study design and protocol, ethics requirements, and research dissemination have been and will continue to be made by the research investigator group who meets monthly. Overall, 3 working groups will be established, comprising research investigators and key stakeholders to oversee and advise on development, implementation, and evaluation of the (1) merchandising strategy, (2) customer intercept survey, and (3) qualitative study to identify the policy supports for scale-up of health-enabling policy action in stores. A communication and research dissemination strategy approved by the research investigator group will govern internal and external stakeholder communication. This includes feedback of study findings and lessons learned at the end of the study to the ALPA and Island and Cape boards and participating communities.
The primary outcome is difference in free sugars (g/MJ) from baseline in intervention versus control stores derived from store sales data. Free sugars contributed 26% of total energy purchased from 20 community stores, more than double the World Health Organization recommendation of less than 10% of total energy intake [
We will assess the impact on store revenue as a secondary outcome. This measure was considered by ALPA as important in determining the impact on retail performance. A necessary outcome for ALPA is that revenue is maintained throughout the intervention. Retail measures useful in evaluating specific merchandising effects on business operations will be examined where data are available, including number of products purchased per transaction (basket size), number of unique transactions, and category share of store sales. We will also collect data on intervention costs, including costs associated with strategy material production, implementation, and evaluation.
Our use of store sales data captures all foods and drinks sold, enabling an assessment of total nutritional quality of purchases as well as exploring impact on specific product types. We will examine the impact on purchases of targeted discretionary products (table sugar, sugar-sweetened beverages, sweet biscuits, and confectionery), total discretionary products, and nontargeted products (eg, water, diet drinks, fruit, and vegetables). We will examine the impact on the nutritional content of all food and drinks sold, including total energy, energy density, and nutrient density.
We will assess intervention implementation using a Merchandising Checklist and the Store Environment tool (Store Scout App). The Merchandising Checklist will capture the intervention activities. Store managers from control and intervention stores will be requested fortnightly during the 12-week intervention period to respond to a brief checklist of intervention components delivered by a member of the research team via phone, and they will be requested to provide photos of premium locations (hot spots for product displays, such as high traffic areas, defined for each store following baseline data collection) to verify degree of compliance with intervention components. Store managers will be asked to comment on observations about community-level incidents that they perceive may affect sales (eg, weather events or festivals). They will also be asked questions related to their perceptions of intervention implementation and effectiveness.
Store Scout App (developed and piloted by our team in 2016 in 6 remote stores) assesses the overall store as a consumer environment (ie, the retail choice architecture), through measures of stocking and merchandising of 7 categories of food and drinks (including fruit and vegetables, drinks, snacks, meals and convenience food, breads and cereals, dairy products and eggs, and meat and seafood). We will train government and Aboriginal Community-Controlled Organization employed public health nutritionists to conduct assessments in control and intervention stores. The tool will be completed at the end of the baseline, intervention, and postintervention periods.
In the final 2 weeks of the intervention, we will conduct customer intercept surveys (1080 unique customers; 45 per store) from control and intervention stores to identify the following:
Differences among customers in discretionary food and drink purchasing in intervention and control communities.
Customer characteristics that predict discretionary product purchases.
Proportion of customers who regularly purchase food and drinks from other food outlets in or outside of the community.
Using a structured close-ended question survey in electronic format (iSurvey) with standardized scripts, trained surveyors with support from store staff will interview customers after they complete their purchase. Surveyors will be scheduled to survey customers at store front (postshop) over a period of 6 hours (3×2 hour sessions over the course of the day) per day for up to 3 days, to capture a broad spectrum of customers and shopping purpose, planned or unplanned, each subject to different levels of impulse shopping propensity. We consulted with ALPA to ensure this is feasible on the basis of number of transactions per day. We anticipate surveys to take 5 to 10 min per customer.
Upon exiting the store, customers will be invited to provide their receipt (a photograph of the receipt will be recorded) and respond to a short-item questionnaire to gather information on characteristics of the customer (age range, gender, shopping alone versus with child or others, and impulse shopping propensity). Data on payment method (eg, cash or card) and food shopping frequency at other retail outlets will also be collected.
The data source for the primary and secondary outcome analysis will be weekly sales reports generated by ALPA for each store for the entirety of the study and sent electronically to the research team. These data will include product identifier (stock keeping unit or barcode), product description, quantity sold, and dollar value. Store products will be linked to nutrient data using a database that we have developed specifically for this purpose, which is mostly derived from the Australian Food, Supplement, and Nutrient Database [
Longitudinal data analysis models will be used on fortnightly store sales data aggregated from weekly sales. This will enable the effect of the intervention to be expressed as a relative change. Analyzing fortnightly data reduces variation because of income cycles, as observed in our previous analyses with sales data [
Descriptive statistics will be used to describe contextual data. To assess implementation fidelity (as described in the implementation evaluation), a dichotomous variable of high or low fidelity will be derived from the repeat measures collected throughout the study in intervention and control stores using the Merchandising Checklist. Contextual and qualitative data will be analyzed to determine factors influencing implementation.
A store environment global score will be derived from use of the Store Scout App and compared across the 3 time periods. We will assess the degree to which the intervention has impacted the consumer environment and been sustained at 24 weeks postintervention in control and intervention stores.
Analyses will be conducted according to the intention to treat principle applied at store level; sensitivity analyses will be conducted looking at the stores’ implementation fidelity. Statistical analyses will be performed using Stata version 15.
Percent discretionary product to total food and beverage dollars (dependent variable) will be calculated for each receipt collected during the Customer Intercept survey. Associations between the dependent variable and dichotomous variables indicating exposure or no exposure to the intervention will be estimated using mixed-effects linear models. Effect of the intervention will be expressed in terms of percent difference relative to the control group adjusting for baseline differences in merchandising. Mixed models will include a random intercept for community to account for within community correlation (clustering effect). As this is a convenience sample of customers, the potential confounding effect of baseline exposure to merchandising of targeted products, age range, and gender will be explored by including these factors in the models. Store Scout will be conducted at baseline in control and intervention communities to provide a proxy of merchandising exposure at baseline. Multivariable linear mixed models (with random intercept for community) will be conducted to identify customer characteristics (eg, payment type) associated with purchase choice.
Our intervention targets products that collectively account for 87% of free sugars from all product purchases in remote communities. The effects of the strategy are likely to be seen from the beginning of the intervention period, and we expect that sales (gram weight) of targeted products will be reduced by approximately 10% and free sugars (g/MJ) by approximately 8% to 9% throughout the intervention; we will test if this is sustained postintervention. A mean effect size of a 10% reduction in targeted discretionary product categories purchased is based on Batis et al (2016) [
Under the assumption of an intracluster correlation equal to 0.01 and a cluster size equal to 45 (on the basis of 24 communities), we calculated a sample size allowing a study design of 1.4. With a sample of 1080 customers, and assuming a relative SD of 40% for the outcome discretionary product dollars of total dollars, the study will have 80% power to detect an 8% difference (from 50% in the control group to 42% in the intervention group) in the discretionary product dollars and total dollars at an alpha of .05.
As of August 2018, 20 stores consented to participate and were randomized to receive the intervention or continue usual practice. The 12-week strategy ended in December 2018. The 24-week postintervention follow-up will occur in May 2019. Results are expected for 2019.
There is strong evidence to show that merchandising is used to drive sales of discretionary foods [
The outcomes will indicate the level of effectiveness and feasibility of the proposed strategy and identify reasons for these outcomes. We will then build on these results to make recommendations for policy, if appropriate, or for the next research study, to identify effective and feasible healthy retailing interventions. The evidence from this study will directly inform the nutrition policy of ALPA and will indirectly influence policies of other retail organizations and community stores through the leadership and influence of ALPA in remote retailing and through the broader effort of the Commonwealth Department of PM&C, which is responsible for food security in remote Indigenous Australia.
Comprehensive multicomponent interventions are necessary to improve the quality of the customer food environment [
Examining the impact of modifying retail food environments for improved diet and health outcomes is a rapidly expanding research area worldwide [
Peer review from funding agency.
Authors response to peer review.
Arnhem Land Progress Aboriginal Corporation
megajoule
National Health and Medical Research Council
northern territory
Prime Minister and Cabinet
stores healthy options project in remote indigenous communities
The authors are grateful to the project staff, Anthony Gunther (Healthy Stores 2020 Project Manager), at Menzies School of Health Research; ALPA and the Island and Cape Board and community store owners for their commitment to this project and the in-kind support in study design and development of implementation materials; Mark Chatfield (University of Queensland) and Federica Barzi (Menzies School of Health Research), Senior Statisticians; Professor Marj Moodie (Health Economist, Deakin University); other members of the Healthy Stores 2020 research collaborative. LMM gratefully acknowledges funding from the Canadian Cancer Society through a Career Development Award in Cancer Prevention (grant # 704744). This study is funded by the Australian National Health and Medical Research Council (NHMRC; ID 1138629). The contents of the published material are solely the responsibility of the administering institution, a participating institution or individual authors, and do not reflect the views of NHMRC. This funding source had no role in the design of this study and will not have any role during its execution, analyses, interpretation of the data, or the decision to submit results.
JB, MF, EM, and CM conceived and all authors designed the study. All authors contributed to refinement of the study protocol and approved the final manuscript.
MF, EM, EM, AP, LMM, TW and CM declare that they have no competing interests. JB is a non-executive director of Outback Stores Pty Ltd and declares no other competing interests. KDS is an employee of the ALPA and declares no other competing interests.