Testing availability, positioning, promotions, and signage of healthier food options and purchasing behaviour within major UK supermarkets: Evaluation of 6 nonrandomised controlled intervention studies.

BACKGROUND: Governments are increasingly looking for policies to change supermarket environments to support healthier food purchasing. We evaluated 6 interventions within major United Kingdom grocery stores, including availability, positioning, promotions, and signage strategies to encourage selection of healthier products. METHODS AND
FINDINGS: Nonrandomised controlled study designs were used, except for one intervention that was rolled out nationwide using a pre/post within-store design. Store-level weekly sales (units, weight (g), and value (£)) of products targeted in the interventions were used in primary analyses using multivariable hierarchical models and interrupted time series (ITS) analyses. Stocking low fat chips next to regular chips was associated with decreases in sales of regular chips (units) in intervention versus control stores (-23% versus -4%; P = 0.001) with a significant level change in ITS models (P = 0.001). Increasing availability of lower energy packs of biscuits was associated with increased sales but reduced sales of regular biscuits in intervention versus control stores (lower energy biscuits +18% versus -2%; P = 0.245; regular biscuits -4% versus +7%; P = 0.386), although not significantly, though there was a significant level change in ITS models (P = 0.004 for regular biscuits). There was no evidence that a positioning intervention, placing higher fibre breakfast cereals at eye level was associated with increased sales of healthier cereal or reduced sales of regular cereal. A price promotion on seasonal fruits and vegetables showed no evidence of any greater increases in sales of items on promotion in intervention versus control stores (+10% versus +8%; P = 0.101) but a significant level change in ITS models (P < 0.001). A nationwide promotion using Disney characters was associated with increased sales of nonsugar baked beans (+54%) and selected fruits (+305%), with a significant level change in ITS models (P < 0.001 for both). Shelf labels to highlight lower sugar beverages showed no evidence of changes in purchasing of lower or higher sugar drinks. These were all retailer-led interventions that present limitations regarding the lack of randomisation, residual confounding from unmeasured variables, absolute differences in trends and sales between intervention versus control stores, and no independent measures of intervention fidelity.
CONCLUSIONS: Increasing availability and promotions of healthier alternatives in grocery stores may be promising interventions to encourage purchasing of healthier products instead of less healthy ones. There was no evidence that altering positioning within an aisle or adding shelf edge labelling is associated with changes in purchasing behaviours. TRIAL REGISTRATION: https://osf.io/br96f/.

Introduction

Poor diet is one of the major contributors to preventable morbidity and premature mortality, accounting for around 15% of years of life lost in the UK [1]. The UK National Diet and Nutrition Survey shows that dietary targets for saturated fat, dietary fibre, free sugars, and salt are not being met [2]. Despite dietary recommendations and public heath campaigns, progress has been slow, and socioeconomic inequalities exist, which contribute to variability in long-term health outcomes [3,4].

Interventions to change food purchasing habits at the point of choice offer an upstream opportunity to change food consumption. Supermarkets account for approximately 87% of all UK retail grocery sales [5], and governments are looking for policies to change supermarket environments to achieve population-level change in dietary habits [6-9]. In addition, it is often proposed that environmental interventions are less likely to exacerbate inequalities than individual level interventions because they require less agency from individuals [10]. However the evidence for this in “real-world” contexts is scant, and some evidence points in the opposite direction, for example, the most affluent consumers tend to purchase the most food on promotion and so may benefit most from restrictions of promotions of less healthy foods [11].

Choice architecture interventions in physical microenvironments, such as grocery stores, have been identified and classified in the typology of interventions in proximal physical microenvironments (TIPPME) framework [12]. Interventions based on placement and availability operate by increasing the range, variety, and number, as well as the visibility and accessibility of certain products, while price and promotional strategies can make products cheaper or increase their attractiveness, and systematic review evidence has shown these strategies can help stimulate purchases [13-23]. But most reviews of these interventions have highlighted the lack of high-quality evidence in real supermarkets, especially for interventions to reduce purchases of less healthy options [13,16,18,23]. In practice, it is hard for academics to plan, implement, and evaluate interventions in real stores. For retailers, there are operational costs and challenges of implementation of in-store interventions, coupled with commercial pressures to avoid decreases in sales. In addition, there are concerns about data sharing and customer privacy, which means that many companies are not willing to take the risk of embarking on research collaborations.

The present research was made possible through a collaboration with the Consumer Goods Forum (CGF), which is a membership body of 50 major consumer goods retailers and manufacturers. Under collaboration and data sharing agreements to access the necessary data from 3 major UK food retailers, we conducted an independent evaluation of in-store interventions, designed and implemented by the retailers and aimed at improving food purchasing behaviours.

Methods

This study is reported as per the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline (S1 Checklist).

Study design and data sources

Three major UK retailers comprising 49.5% of the UK grocery market share in January 2019 were involved in this project during which a total of 6 in-store intervention studies were evaluated. These interventions were completely developed and implemented by the retail partners, so we followed the methods recommended in the monitoring and evaluation of natural experiments [24]. Nonrandomised controlled study designs were used to evaluate the interventions in the active intervention stores compared to a matched sample of control stores, except for one study that was rolled out nationwide across all stores, for which a pre/post within-store study design was used. Retailer A implemented 2 interventions in 34 stores, with a matched sample of 146 to 151 matched control stores. Retailer B implemented 3 interventions, 2 in 7 to 8 stores, for which a sample of 7 to 8 control stores was available, and 1 intervention was rolled out nationwide, with a total sample of 37 intervention stores available. Retailer C rolled out 1 intervention in 18 stores with a matched sample of 65 control stores (see Table 1 for details of store samples and data availability). Aggregated data on store-level weekly sales (units, weight [g], and value [£]) of food products that were targeted in the interventions were obtained for intervention and control stores, spanning dates from January 2018 to January 2020 across the 6 studies. Data from nutrients in sales of target categories were also available (e.g., energy (kcal), total fat (g), sugar (g), and fibre (g)).

Table 1

Intervention characteristics, store samples, and data availability.

Intervention name, retailer, strategy, and description	Target foods/categories	Intervention and baseline periods	Store sample	Data availability	Store-level characteristics available
Frozen chips range changes (Retailer A, Availability)Stocking lower fat frozen chips next to regular chips in stores where only regular chips were available before	Regular fat frozen chips;Lower fat frozen chips	Intervention period: January 21, 2019 to September 22, 2019Baseline period: January 21, 2018 to September 22, 2018	N = 34 interventionN = 146 control	January 1, 2018 to November 24, 201915,841 data points (store weeks)	IMD; Ethnicity
Biscuit range changes (Retailer B, Availability)Change in biscuits range to increase the availability of lower calorie packs (<100 kcal/serving or individual bag) and decrease the availability of higher calorie packs	Regular range biscuits (≥800 kcal/entire pack);Lower energy range biscuits (<800 kcal/entire pack)	Intervention period: May 19, 2019 to August 11, 2019Baseline period: May 20, 2018 to August 12, 2018	N = 8 interventionN = 8 control	May 13. 2018 to December 29, 20191,344 data points (store weeks)	IMD
Breakfast cereal positioning (Retailer B, Positioning)Positioning higher fibre cereal to be located at eye level within an aisle, displacing regular breakfast cereal	Regular breakfast cereal; Higher fibre breakfast cereal	Intervention period: May 19, 2019 to August 11, 2019Baseline period: May 20, 2018 to August 12, 2018	N = 7 interventionN = 7 control	May 20, 2018 to December 29, 20191,127 data points (store weeks)	IMD
Promotional marketing with Disney characters (Retailer B, Promotions)Educational material that conveys information about nutrition and health using Disney characters, including children packs and collectables, point-of-sale signage, recipe cards, shelf markers, leaflets, emails and newsletters, magazine articles, and in-store displays. This was coupled with loyalty card rewards (points) when purchasing the target products	Selected fruits (mini apples and oranges);Nonsugar baked beans	Fruit promotionsIntervention period: August 18, 2019 to October 6, 2019Baseline period: September 9, 2018 to October 7, 2018Baked beans promotionsIntervention: period: September 1, 2019 to October 6, 2019Baseline period: September 9, 2018 to October 7, 2018	N = 37 intervention(No control stores were eligible because this intervention was implemented nationwide)	September 9, 2018 to January 1, 20202,649 data points (store weeks)	IMD
Fruit and vegetable price promotions (Retailer A, Promotions)Temporary price reductions and promotional space for a selection of seasonal fruits and vegetables	Selected seasonal fruits and vegetables	Intervention period: May 29, 2019 to November 24, 2019Baseline period: May 29, 2018 to November 24, 2018	N = 34 interventionN = 151 control	January 1, 2018 to November 24, 201917,381 data points (store weeks)	IMD; Ethnicity
Shelf labelling beverages of nonalcoholic beverage categories (Retailer C, Signage)Shelf labels highlighting lower sugar and sugar-free beverages within aisles	Regular beverages; Lower/nonsugar beverages	Intervention period: May 27, 2019 to August 26, 2019Baseline period: May 28, 2018 to August 27, 2018	N = 18 interventionN = 65 control	December 25, 2017 to September 17, 2018 and December 24, 2018 to September 16, 20196,474 data points (store weeks)	IMD

IMD, Index of Multiple Deprivation.

By using aggregated weekly sales data, this study was exempt from ethical review and approval. A preregistered protocol (https://osf.io/br96f) was completed and fully available from July 22, 2020 before obtaining data for analysis.

Store selection and matching

Retail partners’ finance and data teams used proprietary analytics to select intervention and control stores for this study. Based on each retailers’ operational considerations and in coordination with the CGF and the project partners, Guy’s & St Thomas’ Charity, a place-based foundation, intervention stores were selected within London boroughs (Lambeth and Southwark, UK). The sample of intervention stores was located in neighbourhoods covering a range of socioeconomic deprivation strata based on the 2019 English Index of Multiple Deprivation (IMD) income domain, the official measure of relative deprivation in small areas (Lower Layer Super Output Areas) across England [25]. Selected intervention stores were all small supermarkets according to a retail food outlet categorisation system previously defined, which includes stores with 1 to 4 manned cash registers [26,27], except for the study that was rolled out nationwide where the intervention happened in larger supermarkets defined as 5+ manned cash registers. Control stores were selected across each retailers own stores, with store size and overall sales performance over the previous year used as the criteria for matching stores (Table 1).

Interventions

An intervention framework was developed to classify all the interventions implemented by retailers according to the TIPPME tool [12] (see Table 1 for details of the interventions, target products, and dates of implementation). Intervention periods ranged from 3 to 8 months. Six interventions tested 4 principal strategies:

Availability: stocking a lower fat frozen chip next to regular chips; increasing the proportion of lower energy biscuit packs (<100 kcal/serving or individual bag) and decreasing the availability of higher energy biscuits;

Positioning: changing shelf location of healthier cereal within aisles to be at eye level;

Promotions: temporary price promotions of seasonal fruits and vegetables; promotional marketing using Disney characters to encourage purchases of healthier products such as fruits and nonsugar baked beans; and

Signage: shelf labelling of lower/nonsugar beverages.

Outcome measures

Primary outcome measures in each study included store-level weekly sales data (units, weight (g), and value (£); with units shown in all graphs for consistency) for foods/food categories that were targeted in each intervention: regular fat and lower fat frozen chips; lower energy range and regular range biscuits; high-fibre/low-sugar and regular cereal; fruits (apples/mandarins); nonsugar baked beans; seasonal fruits and vegetables; and low-alcohol/low-sugar and regular beverages. Secondary outcome measures included nutrient data from sales of specific target food categories (i.e., energy, total fat, sugar, and fibre).

Store characteristics

Store characteristics relating to the customer population included the English IMD and ethnicity (only Retailer A, Table 1). The store postcode was matched to the IMD income domain, the official measure of relative deprivation in small areas (Lower Layer Super Output Areas) across in England [28], which was used as a proxy for the socioeconomic status (SES) of the customer population. The store sample covered neighbourhoods from deciles 1 to 10, regrouped into IMD 1 to 3 (most deprived), 4 to 6 (mid), and 7 to 10 (least deprived). Ethnicity of the store customer population was classified by the retailer using internal proprietary systems and grouped as predominantly white versus other ethnicities.

Statistical analysis

Power analyses were not conducted, and each retailer chose the number of stores to roll out the interventions. Descriptive analyses were used to investigate differences in store demographic characteristics between intervention and control stores using chi-squared tests.

We used data over the year prior to intervention (2018) to define preintervention baseline periods, which matched closely the intervention period (2019) (Table 1). We tested differences in weekly sales of target products over the 2018 baseline periods between intervention and control stores using Student t tests.

Two prespecified statistical models were used for the primary and secondary outcome analyses, using consistent methods for intervention evaluation [29]:

Hierarchical models (negative binomial for unit sales or linear mixed models for weight and value of purchases) were used with a fixed effect adjustment for store demographic characteristics and average weekly sales over the baseline preintervention period. This model was used to investigate differences in weekly sales of target products in intervention versus control stores over the time period while the intervention was active compared to the preintervention baseline period (2018) [30].

Interrupted time series (ITS) analyses and corresponding plots with fitted linear trends were computed using all available data before and after the intervention for intervention and control stores [31]. To assess whether differences visible in the graphs were statistically significant between intervention and control stores and to account for any preintervention differences between groups in the outcome variable, we used a difference-in-difference approach, calculating the mean difference in weekly sales between intervention and control stores and testing whether this time series of differences changed after versus before intervention using a linear regression model. We used a Chow type test for level and trend changes after intervention implementation and Newey–West standard errors with lag 4 to allow for autocorrelation in the time series.

Analyses were conducted using all intervention and control stores with all available data. A prespecified exploratory subgroup analysis was performed by store IMD group (IMD 1 to 3 high deprivation versus IMD 4 to 10 middle/low deprivation), and likelihood ratio tests were used to test the significance of the interaction. Stata version 16 was used for all statistical tests with a 5% significance level.

Results

Differences in store characteristics

For Retailer A, there were significant differences in IMD scores but not in ethnicity between intervention and control stores across the 2 studies (Table A in S1 Appendix).

For retailers B and C, there were no significant differences in IMD scores between intervention and control stores (Tables B and C in S1 Appendix). Overall, across the 3 retailers, intervention stores were mostly located in areas of medium/higher deprivation, which is representative of the population of Lambeth and Southwark (London, UK).

Availability interventions

One intervention aimed to switch consumers to purchase a lower fat frozen chip by stocking them next to regular chips in stores where only the regular chips were available previously. This intervention ran for approximately 8 months starting in January 2019. Over the 2018 preintervention baseline period, there were no significant differences in sales (units, weight, and value) of regular frozen chips between intervention versus control stores (Table D in S1 Appendix). During the study period, intervention stores reduced their weekly sales of regular frozen chips by −23% (units and weight) and −14% in value, while the control stores reduced sales by −4% (units and weight) but increased +8% in value, with statistically significant differences between intervention versus control stores (P = 0.001, Fig 1, Table E in S1 Appendix). The absolute difference in weekly sales in regular frozen chips (−2,998 g or −3.3 units) was coupled with an absolute increase in weekly sales in lower fat chips (+3,361 g or +3.7 units) in the intervention stores. ITS analyses showed a statistically significant difference in level change in weekly sales of regular chips at the point of intervention (P = 0.001, Fig 2) and in the trends afterwards. There were no significant differences in the total energy or fat content of all chips sold during the study period between intervention and control stores (Table E in S1 Appendix).

Fig 1

Weekly sales of target food categories (units) at baseline and comparison of changes before/after intervention in intervention versus control stores.

*Data shown are presented according to intervention strategy (availability, positioning, promotions, and signage) with estimates coming from 5 of the interventions included in this study, in the following order: Frozen chips range changes (regular frozen chips); Biscuit range changes (regular range biscuits, lower energy range biscuits); Breakfast cereal positioning (regular breakfast cereal, high-fibre breakfast cereal); Fruit and vegetable price promotions (seasonal fruits and vegetables); and Shelf labelling beverages of nonalcoholic beverage categories (sugar beverages, low/non sugar beverages). Baseline periods used: Frozen chips range changes January 21 to September 22, 2018; Biscuit range changes May 20 to August 12, 2018; Breakfast cereal positioning May 20 to August 12, 2018; Fruit and vegetable price promotions May 28 to November 24, 2018; Shelf labelling beverages of nonalcoholic beverage categories May 28 to August 27, 2018. IRRs were obtained from hierarchical negative binomial models with fixed effect adjustment for store demographics and average sales per week over the baseline 2018 period. IRR, incidence rate ratio.

Fig 2

ITS analysis showing level and trend changes in weekly sales of target food categories (units/store/week) with availability interventions.

*Solid dots/lines represent intervention stores, and white dots/dotted lines represent control stores. ITS, interrupted time series.

A second intervention manipulated the range of biscuits and cookies to increase the proportion of lower energy options available within the range, largely by offering smaller pack sizes. This intervention ran for approximately 3 months starting in May 2019. Over the 2018 preintervention baseline period, there were no significant differences in sales (units, weight, and value) of the regular or the lower energy biscuit range between intervention versus control stores (Table D in S1 Appendix). Over the 2019 intervention period versus baseline, intervention stores decreased their weekly sales of the regular biscuit range by −3% in weight, −4% in units, and −15% in value, while the control stores increased sales by +8% in weight, +7% in units, and +3% in value. Only the change in value of sales was statistically significantly different between intervention and control stores (P = 0.011). Over the same period, intervention stores increased their weekly sales of the lower energy biscuit ranges by +16% in weight, +18% in units, and +20% in value, while the control stores saw a decline in sales by −0.4% in weight, −2% in units, and smaller increase in value, +7%, although the differences between stores were not statistically significant (Fig 1, Table E in S1 Appendix). ITS analyses showed a statistically significant difference in level change at the point of intervention of both the regular range (P = 0.004 and the lower energy range (P < 0.001, Fig 2) and a significant difference in the trend afterwards in sales of lower energy biscuits (P < 0.001). There were no significant differences in total energy content of all biscuits sold during the study period between intervention and control stores (Table E in S1 Appendix).

Positioning interventions

One intervention manipulated the position of breakfast cereals within the aisle to locate healthier (higher fibre and/or lower sugar) cereals at the eye level in exchange for regular (lower fibre and/or higher sugar) cereal packs. This intervention ran for approximately 3 months starting in May 2019. Over the 2018 preintervention baseline period, there were no significant differences in sales (units, weight, and value) of regular or high-fibre cereals between intervention versus control stores (Table D in S1 Appendix). There were no significant changes over the 2019 intervention period versus baseline period in the intervention stores compared to control stores in sales of regular or higher fibre cereal (Fig 1, Table E in S1 Appendix). Counter to the expected results, ITS analyses showed a significant reduction in weekly sales of the high-fibre cereal at the point of intervention compared to control stores (P = 0.003, Fig 3), but increased sales of regular breakfast cereal more than control stores (P < 0.001). There was a significant difference in the trend afterwards in sales of regular breakfast cereal (P < 0.001).

Fig 3

ITS analysis showing level and trend changes in weekly sales of target food categories (units/store/week) with positioning interventions.

*Solid dots/lines represent intervention stores, and white dots/dotted lines represent control stores. ITS, interrupted time series.

Promotions

One intervention implemented a multicomponent intervention using Disney characters to promote healthier products, including selected fruits and nonsugar baked beans, and was implemented nationwide for approximately 6 weeks (August to September 2019).

Over the 2019 intervention period versus 2018 period, stores increased their weekly sales of selected fruits by +232% in weight, +305% in units, and +135% in value (Table E in S1 Appendix). Over the same period, stores increased their weekly sales of nonsugar baked beans by +54% in weight, +54% in units, and +79% in value. Single-group ITS analyses showed statistically significant level changes in weekly sales of selected fruits (P < 0.001) and nonsugar baked beans (P < 0.001) (Fig 4), but a significant trend towards preintervention sales thereafter. Data from similar nonpromoted products, including regular baked beans and other fruits, were used in supplementary ITS analyses as a comparator, showing no significant level changes in weekly sales of nonpromoted products (Fig A in S1 Appendix).

Fig 4

ITS analysis showing level and trend changes in weekly sales of target food categories (units/store/week) with promotional interventions.

*Solid dots/lines represent intervention stores, and white dots/dotted lines represent control stores. ITS, interrupted time series.

In another intervention, temporary price reductions and increased promotional space were used to encourage purchases of selected seasonal fruits and vegetables. This intervention ran for approximately 6 months starting in May 2019. Over the 2018 preintervention baseline period, there were no significant differences in sales (units, weight, and value) of selected fruits and vegetables between intervention versus control stores (Table D in S1 Appendix). Over the 2019 intervention period versus baseline period, intervention stores increased their weekly sales of fruits and vegetables by +10% in weight and units and +12% in value, while the control stores increased sales by +3% in weight, +8% in units, and +5% in value, although these differences between intervention versus control stores were not statistically significant (Fig 1, Table E in S1 Appendix). However, ITS analyses showed a statistically significant difference in level change at the point of intervention (P < 0.001, Fig 4) as well as in the trend afterwards (P < 0.001).

Signage

One intervention used shelf labels highlighting lower/nonsugar beverages within aisles and ran for approximately 3 months starting in May 2019. Over the 2018 preintervention baseline period, there were no significant differences in sales (units, weight, and value) of lower/nonsugar beverages between intervention versus control stores; but sales of regular beverages in intervention stores were significantly lower compared to control stores at baseline (Table D in S1 Appendix). There were no significant changes over the 2019 intervention period versus 2018 period in the intervention stores compared to control stores in sales of regular or lower/nonsugar beverages (Fig 1, Table E in S1 Appendix). ITS analyses showed nonsignificant differences in level changes in weekly sales of regular or lower sugar beverages at the point of intervention (Fig 5), but there was a significant difference in the trend afterwards in sales of lower/nonsugar beverages (P < 0.001).

Fig 5

ITS analysis showing level and trend changes in weekly sales of target food categories (units/store/week) with signage interventions.

*Solid dots/lines represent intervention stores, and white dots/dotted lines represent control stores. ITS, interrupted time series.

Differences by store deprivation

There were no statistically significant interactions with IMD group for any of the interventions analysed, although there was some evidence of heterogeneity (Table F in S1 Appendix). For example, one availability intervention was associated with higher sales of the lower energy biscuits in stores located in higher deprivation areas (IRR 1.42 [1.13 to 1.79]) but not in stores within lower deprivation areas. In contrast, price promotions on seasonal fruits and vegetables was associated with increased sales in stores located in lower deprivation areas (IRR 1.12 [1.00 to 1.26]) but not in stores within higher deprivation.

Discussion

This wide-ranging analysis of 6 in-store interventions showed that increasing the availability of healthier options within a category (e.g., lower fat frozen chips or lower energy biscuit packs) was associated with significant increases in purchases of the healthier items. Promotions were associated with a significant initial uplift in sales of target products, which declined over time. There was no evidence that altering the positioning of healthier cereals within an aisle or shelf edge labelling of lower/nonsugar beverages were associated with changes in purchasing behaviours. Overall, there was no evidence that the results of these environmental interventions varied according to the level of deprivation in the area where the store was sited.

There is an emerging body of evidence on effective in-store interventions to help change purchasing behaviours, although still few studies have been conducted in “real-world” settings or in more than 1 store. Consistent with our results, previous studies have shown that lower prices and promotions on healthier options are associated with increased sales of these products because they incentivise purchasing [13,21-23]. However, observational studies have suggested that price promotions on less healthy food items are more prevalent and more likely to influence food purchasing than promotions on healthy items [19]. A recent cluster randomised controlled trial in remote Australian stores tested a complex intervention to limit in-store promotional activities (including those in prominent areas) targeting high-fat/high-sugar products and showed decreases in sales of sugary drinks and significant reductions in free sugars [32]. Positioning products in prominent locations, such as checkouts or the end of an aisle, increases visibility and accesibility to products, and this can stimulate purchases [12,15,16,18,22]. Limited evidence around altering the position of food products within the shelf or aisle, for example, removing less healthy products from eye-level positions, has also shown positive influences on food choice [15], but our study did not provide evidence of changes towards purchasing more higher fibre cereal options. On the other hand, increasing the availability of healthier items with a corresponding decrease in less healthy items was effective in increasing selection or purchase of healthier items, with enhanced effects if combined with positioning strategies [16,17,22,33]. Our study provides evidence that increased availability of healthier options within a range (e.g., lower fat chips) is associated with significant results on food purchasing behaviour, although it is possible this may be more important within small-scale stores (as here), where options are more limited than in larger format stores. Finally, evidence for information/education interventions to convey information about product properties (e.g., shelf tags, signage, posters, flyers, recipes, and taste testing) is mixed, and we found no evidence of changes in purchasing behaviours in our analyses [13,16,20,22,23].

In the context of the increasing gap in dietary inequalities and long-term health outcomes, it is also important to understand if supermarket interventions help reduce, or at least do not exacerbate, dietary inequalities. Systematic reviews have identified a limited number of studies generally showing mixed results on the impact of supermarket interventions on inequalities [18,23]. It has been argued that environmental-level (as oppose to individual level) approaches and, relatedly, interventions that trigger automatic (rather than conscious) behavioural responses [10,34] are less likely to increase health inequalities. The results from 2 interventions here suggested that some strategies might show small differences by store deprivation area in opposite directions: promotions favouring the more affluent and one intervention increasing availability of healthier options favouring the least advantaged. However, our evaluation generally showed no evidence of large differences in intervention results across the 6 studies.

These results contribute to the underresearched field of interventions that are effective in supporting healthier choices in real supermarkets, which may be of particular interest to policymakers considering opportunities for new policy actions. Our study provides evidence on potentially effective interventions (e.g., availability of healthier options next to regular ones) that can help reduce purchases of less healthy options, especially food categories contributing energy, saturated fat, and free sugars to the UK diet [2], for which good quality evidence is particularly scarce. However, it also highlights the weak impact of other interventions, such as signage and repositioning of items within aisles. While it is hard to draw generic conclusions from these few interventions targeting different behavioural mechanisms, it is clear that, overall, structural interventions targeting automatic behavioural responses are associated with stronger changes in the intended direction compared to those targeting conscious decisions, which is generally consistent with recent work [35].

It is important to note that the interventions here focused on encouraging swaps from a less healthy to a healthier option or increasing overall sales of healthy foods, such as fruits and vegetables. This is an attractive commercial proposition, but the impact of these interventions on the overall energy content of food purchases is likely to be considerably smaller than interventions that specifically seek to reduce impulse of discretionary purchases, such as the removal of foods high in fat, sugar, and salt from prominent locations such as end of aisles [36].

This collaboration with food retailers was established to facilitate the industry–academic dialogue and enable rapid access to the necessary data for this evaluation. From the many interventions planned by the retailers, the study team chose for evaluation those where there was a clear target food category for intervention, an identifiable behavioural mechanism, and considerable duration of the intervention. This evaluation provides proof of concept that it is possible to establish these multiretailer collaborations and learn important lessons to design larger and more definitive intervention studies. Future studies should include plans to evaluate legacy effects of the most effective strategies, as some of the interventions analysed here showed significant differences in trends from ITS models after implementation suggesting that some interventions may be short lived. To better study the impact on health inequalities, future research should seek to analyse changes in purchases at a household rather than store level using data from customer loyalty cards [37,38].

A major strength of this study is the use of large data sets over an extended time period drawn from interventions conducted in real supermarkets. The results of these studies reflect true shopping behaviour and provide important insights to inform population-level interventions to encourage healthier food purchasing. However, these real-world data present analytical challenges. Adjustment for confounding and other sources of heterogeneity was approached in several ways. First, control stores were matched to intervention stores, with more than 1 control store per intervention store in most cases. Matching was done using store demographic factors and overall sales over the previous periods, which, in most cases, resulted in nonsignificant differences in baseline characteristics between intervention and control stores. However, there were significant differences by IMD due to the fact that stores in less deprived areas were underrepresented in the intervention sample, but in any case, all available baseline demographic characteristics were adjusted in the models. The difference-in-difference approach used in models also helps to remove the effect of any small absolute differences in sales between the intervention and control stores. Finally, with access to extended periods of time (2018 and 2019), we were able to use the 2018 period as a control in the models. However, the sample of stores was smaller (<8 stores) for 2 studies and in one case did not include control stores, limiting the power to detect any significant effects.

Other limitations to note include the lack of randomisation, residual confounding from unmeasured variables, and absolute differences in trends and sales between intervention versus control stores which in our case were not statistically significant. There could have been other interventions in stores running alongside the ones tested here, which could have influenced the observed effects, but the use of control stores should help adjust for this. In addition, we have no independent measures of intervention fidelity, and we had to rely on the retailer implementation plans, which means poor implementation may have diminished the apparent effects of some interventions. These interventions were selected, developed, and implemented by the retailers, without the direct involvement of the research group. It is not possible to know the extent to which this was influenced by behavioural theory, prior commercial insights or awareness of government thinking, although it is probable that all contributed to greater or lesser extent. There was also limited data on store characteristics, and only one retailer provided restricted data on the ethnicity of the customer population. The very broad categorisation of ethnicity is unlikely to have removed all of the confounding related to ethnicity in our results, although there were no significant differences in the distribution of ethnicity between intervention and control stores. Similarly, the IMD used as a measure of store deprivation may also be a very crude proxy for the SES status of the customer population, particularly when people drive to larger out-of-town supermarkets or for smaller stores located in city centres with a large proportion of nonlocal customers. Finally, for the most promising interventions, we studied the effects in 2 different food categories (e.g., frozen chips and biscuits for availability interventions; seasonal fruits/vegetables and beans/fruits for promotions), and from this, we make inferences on the potential effects of these specific intervention modalities. However, for other interventions, we had data on only one food category (e.g., beverages; breakfast cereal), and it remains to be tested whether the effects are also observed across categories. Indeed, one of the research questions facing this field is whether behaviourally similar interventions are equally effective across food categories or whether there are interactions between intervention modality and food category.

In conclusion, these results make a substantial contribution to the emerging evidence of the effect of in-store interventions on food purchasing behaviour. Some structural interventions within stores, including availability and promotions, were associated with significant changes in food purchasing in the intended direction, although the changes observed with promotions on consumer behaviour may diminish with time and are less likely to be sustainable for retailers over longer time periods. Strategies aiming at informing customers about healthier options are unlikely to work in isolation. This research has important implications for the development of policies by retailers or governments to bring dietary intakes closer to recommendations for good health.

STROBE Checklist.

STROBE, Strengthening the Reporting of Observational Studies in Epidemiology.

(DOCX)

Click here for additional data file.

Supporting information tables and figures.

Table A: Store demographic characteristics (Retailer A). Table B: Store demographic characteristics (Retailer B). Table C: Store demographic characteristics (Retailer C). Table D: Differences in weekly sales of target food categories in intervention versus control stores over the preintervention baseline period. Table E: Average weekly sales of target food categories in intervention versus control stores during the intervention period and comparison of changes before/after intervention between intervention versus control stores. Fig A: ITS analysis showing level and trend changes in weekly sales of products not promoted during the promotional intervention using Disney characters (units/store/week). Table F: Comparison of changes in sales of target food categories (units/store/week) before/after intervention between intervention versus control stores, by store IMD group. IMD, Index of Multiple Deprivation; ITS, interrupted time series.

(DOCX)

Click here for additional data file.

16 Sep 2021

Dear Dr Piernas,

Thank you for submitting your manuscript entitled "Natural experiments testing availability, positioning, promotions and signage of healthier food options: Analysis of six non-randomised controlled trials within major UK supermarkets" for consideration by PLOS Medicine.

Your manuscript has now been evaluated by the PLOS Medicine editorial staff as well as by an academic editor with relevant expertise and I am writing to let you know that we would like to send your submission out for external peer review.

However, before we can send your manuscript to reviewers, we need you to complete your submission by providing the metadata that is required for full assessment. To this end, please login to Editorial Manager where you will find the paper in the 'Submissions Needing Revisions' folder on your homepage. Please click 'Revise Submission' from the Action Links and complete all additional questions in the submission questionnaire.

Please re-submit your manuscript within two working days, i.e. by Sep 20 2021 11:59PM.

Login to Editorial Manager here: https://www.editorialmanager.com/pmedicine

Once your full submission is complete, your paper will undergo a series of checks in preparation for peer review. Once your manuscript has passed all checks it will be sent out for review.

Feel free to email us at plosmedicine@plos.org if you have any queries relating to your submission.

Kind regards,

Callam Davidson

Senior Editor

PLOS Medicine

11 Nov 2021

Dear Dr. Piernas,

Thank you very much for submitting your manuscript "Natural experiments testing availability, positioning, promotions and signage of healthier food options: Analysis of six non-randomised controlled trials within major UK supermarkets" (PMEDICINE-D-21-03912R1) for consideration at PLOS Medicine.

Per our recent correspondence, the editorial team have decided to rescind our original decision stating that the manuscripts ought to be combined into a single paper and instead will proceed to consider them as independent submissions. At the bottom of this email you will find editorial comments and reviewer comments. Any accompanying reviewer attachments can be seen via the link below:

[LINK]

We would like to consider a revised version of this manuscript that addresses the reviewers' and editors' comments. We cannot make any decision about publication until we have seen the revised manuscript and your response, and we plan to seek re-review by one or more of the reviewers.

In revising the manuscript for further consideration, your revisions should address the specific points made by each reviewer and the editors. Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments, the changes you have made in the manuscript, and include either an excerpt of the revised text or the location (eg: page and line number) where each change can be found. Please submit a clean version of the paper as the main article file; a version with changes marked should be uploaded as a marked up manuscript.

In addition, we request that you upload any figures associated with your paper as individual TIF or EPS files with 300dpi resolution at resubmission; please read our figure guidelines for more information on our requirements: http://journals.plos.org/plosmedicine/s/figures. While revising your submission, please upload your figure files to the PACE digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at PLOSMedicine@plos.org.

We hope to receive your revised manuscript by Dec 02 2021 11:59PM. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

We ask every co-author listed on the manuscript to fill in a contributing author statement, making sure to declare all competing interests. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. If new competing interests are declared later in the revision process, this may also hold up the submission. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT. You can see our competing interests policy here: http://journals.plos.org/plosmedicine/s/competing-interests.

Please use the following link to submit the revised manuscript:

https://www.editorialmanager.com/pmedicine/

Your article can be found in the "Submissions Needing Revision" folder.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

We look forward to receiving your revised manuscript.

Sincerely,

Callam Davidson,

Associate Editor

PLOS Medicine

plosmedicine.org

-----------------------------------------------------------

Requests from the editors:

Please use the reviewers’ comments to guide your structuring of the revised manuscript. It was generally felt that a more consistent and systematic presentation of the various interventions would facilitate interpretation of your findings. Reviewer 1 suggests that Supplementary tables 1 and 2 could be combined and presented as Table 1 in the main text to facilitate reader comprehension.

Please update your title such that the setting occurs before the colon ("Natural experiments testing availability, positioning, promotions and signage of healthier food options within major UK supermarkets: Analysis of six non-randomised controlled trials").

Please add a final line to your abstract ‘Methods and findings’ which summarises the main limitations (it should begin ‘The limitations of this study include’ or similar). Given that the manuscript covers a variety of studies, it may be best to focus on the limitations relating to study design more generally as opposed to detailing limitations specific to any one of the experiments presented.

Please include line numbering in the margins throughout.

Citations should occur before punctuation.

As reflected in the reviewers’ reports, please provide further clarification as to the definition/classification of ethnicity in your study and consider discussing it as a limitation.

The PDF conversion process appears to have introduced some minor formatting errors – an example can be found in line 1 of page 7 line 3 (a boxed question mark appears where it should not). Please check throughout for other errors.

Please remove the funding information from the acknowledgements section (in the event of publication this will be published as metadata based on your responses to the submission form).

Please also remove the Contributors, Competing interests, and Data sharing sections from the main text for the same reason as above.

Please remove the Ethical approval statement from the end of the main text as this is already stated in the methods section.

In your references, please only use et al. after listing the first six authors. Our guidelines can be found here: https://journals.plos.org/plosmedicine/s/submission-guidelines#loc-references

Please provide a completed STROBE checklist. When completing the checklist, please use section names and paragraph numbers rather than page numbers.

Please reference the checklist in your methods ("This study is reported as per the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline (S1 Checklist)."

When reporting p-values, please use P<0.001 rather than P<0.0001 (please also check your figures and tables to ensure P<0.001 is used consistently).

Comments from the reviewers:

Reviewer #1: This is an interesting attempt to analyse six non-randomised controlled trials within major UK supermarkets in one paper on testing availability, positioning, promotions and signage of healthier food options. The study design and statistical methods and analyses are mostly adequate. However, there are a few major issues needing attention.

1) In figure 1, only 1 of 7 trials in stores (intervention vs control) is statistically significant while most trials aren't, which suggests the authors need to tone down all the claims in the paper including in the abstract and discussion. Overall, the trials results are not conclusive with majority of non-significant results therefore the intrepretations need to be balanced, modest and fair. Although the ITS analyses identified a few significant changes before and after intervention, it's mainly technical and secondary and can't substitute the main findings in figure 1. Also it's a bit confusing as why there are 7 trials in figure 1 but throughout the paper only 6 trials are mentioned?

2) Analysing 6 trials in one paper is a bit challenging and the writing sometimes is difficult to follow. Without good and clear understanding of the designs of these 6 store trials, it's difficult to continue to read. However, supplementary table 1 and 2 are extremely informative on design, sample size and intervention of these trials. I'd like to suggest to combine supplementary table 1 and 2 to make it Table 1 in the main paper so that the readers can understand the framework of the trials and then can read the results in the context of design, sample size, intervention and etc. Also, it would be useful to match and align the trials/interventions between figure 1 and new table 1 so that the trial results match the design and sample size.

3) Sample size. From the supplementary table 1, we can see only two trials from retailer A have relatively decent sample size (34 vs 151 and 34 vs 146). All the rest trials have relatively sample size (No. of stores) especially 3 trials from retailer B (8 vs 8 and 7 vs 7 and 37 (no control)). The authors might argue there are many data points (weekly and etc) available but it is really the number of stores that matters. The small sample size may contribute to the mostly non-significant results of these 6 trials as lack of statistical power. This issue needs to be properly discussed as a limitation.

4) The presentation of p-values needs to be consistent throughout the paper. For example, in the abstract, it appear 'P <0·0001' but all the other p-values only have 3 digits. Can authors please check and make sure p-values are presented with 3 digits throughout the paper?

Reviewer #2: I was asked to review these papers together. Given their similarity many of my comments apply equally to both and so I have provided a combined report. I've noted where comments refer particularly to one or the other paper and described these as the 'combined' or 'easter' papers. Overall, I wasn't convinced by the value of separating these papers. Whilst the combined paper is certainly complicated, I don't think that adding the easter analyses would make it appreciably more so. I think perhaps the only value of the distinct easter paper is that it speaks very specifically to current UK public health policy decisions. On the other hand, the specific high value placement message could be drawn out in post-publication dissemination and, arguably, UK public health policy needs to be thinking beyond just the current policy decisions on the table to some of the strategies presented in the combined paper.

These papers report on a rather disparate set of interventions in supermarkets aiming to improve the healthiness of purchases. I applaud the authors on the partnership they have developed with the respective retailers and agree that this represents a substantial achievement both in terms of interventions achieved and in data made available. Unfortunately I found the report (particularly in the combined paper) of, what is undoubtedly a complicated collection of work, rather hard to follow and a little underdeveloped in terms of both theoretical introduction and interpretation of findings.

The nature of the partnership between the researchers and the retailers is a little unclear. Although there are a number of different definitions of natural experiments, and natural experimental evaluations, key is that the nature and implementation of the intervention is outwith control of the researchers. It is not clear if this was the case here. If there is uncertainty in definitions, non-randomised controlled trials is an unambiguous description of what was done.

I'd also like to understand more about why this particular group of interventions was selected for study. Was this informed by theory, pragmatic considerations or something else? A key limitation on how the findings can be interpreted is the possibility of an intervention modality by food category interaction - as different food categories were selected for each intervention modality, we can't tell if the effects reported are specific to the food category, the intervention modality, or the interaction of the two. I'd like to see more consideration given to this throughout.

I also thought there was a missed opportunity to raise the issue of industry versus government led public health initiatives in this paper. Again, it wasn't clear to me whether the decision to implement these interventions was driven by retailers, the industry body mentioned, the charity mentioned, or the researchers. Shedding some further light on this may help illuminate how such interventions might be implemented more consistently.

Whilst a lot of information about the interventions and data used is provided in supplementary material (particularly in the combined paper), I would have liked to see this brought into the main text. The scant descriptions of the interventions on p6 of the combined paper left we many unanswered questions that I didn't feel were appropriate to be left to supplementary material. This includes exactly what products were impacted; how eg lower fat and low sugar are defined; whether products are added or moved from elsewhere in store etc.

The application of IMD and ethnicity to stores was poorly explained and, I felt, crude. Was this using lower level super output area of location? If so, how likely is it that supermarket customers are drawn only from the surrounding LSOA? At best, I think the limitations of this aspect of the work need further consideration. Perhaps it should be cast in an 'exploratory' light?

I'd recommend both manuscripts are reviewed by a statistician. I wasn't clear what the hierarchical vs ITS analyses were contributing and how I should reconcile the information provided by both. More information on what specific questions these two different strategies addressed would be helpful. I found the inconsistent presentation of different information on the results for each intervention in the combined paper additionally confusing. Eg information on nutritional information and differences in pre-intervention sales are sometimes presented, sometimes not.

There were some apparent inconsistencies in the results that could do with more explanation and discussion. For example, in the ITS, I wasn't clear why there seemed to be substantial increases in the level for control stores at the point of intervention in many cases - even if not statistically significant, what do these reflect? I also wasn't clear if seasonal differences had been taken into account (both eg seasonal events such as Christmas, and seasonal patterns such as temperature changes in summer). This seems particularly important in the easter paper. Finally, the changes in some, but not all, nutrients of interest in line with sales (in the easter paper particularly, but I think in the combined paper too, but reporting of nutrients here was less consistent) could be explored further - there appears to be some displacement of purchases to other foods, but not total?

I would like to see more consideration in the discussion section (of the combined paper) of why some intervention modalities seemed more promising than others. The authors present a general background of inconsistent previous literature. What do these findings add? Are we any further forward in working out in what contexts different intervention modalities may be more or less effective? How could we make progress on this?

Overall the manuscripts need attention to terminology (and consistency) of this throughout - eg cost vs value; natural experiment vs natural experimental evaluation; interventions being short-lived vs intervention effects being short lived etc; the 'opposite' of the population approach is the high-risk approach, not the individual approach; some assertions are without reference (eg socioceoncomic inequalities exist); 'hypotheses' are referred to in the results, but none were presented in the aims.

In relation to the consideration of Adams et al (2016), please be clearer that the hypothesis concerning 'agency' is proposed, not stated; and that the agentic effort referred to is specifically of individual potential beneficiaries of interventions - interventions don't have agency, they require individual recipients to use their agency and mobilise their resources in order to benefit.

Reviewer #3: The authors present the results of a series of observational studies on the effectiveness of various nudges to promote healthier grocery purchases. They rely on data from large chain grocery stores across the UK and examine the preliminary success of low touch interventions such as placing healthier options at eye level, coordinating grocery displays with promotional advertising and using signage and labeling to help direct consumers to healthier alternatives. On the one hand, the data are very noisy and the lack of pure randomization makes the (mostly null) results difficult to interpret. I think the authors do a good job of highlighting the limitations of their approach and tried their best to statistically control for the cluster effects, non-randomization and the reality that relying on grocers/retailers to implement these interventions is noisy business. I also worry less about Type 2 errors than Type 1 errors in this situation. Overall, the results are not terribly surprising -- low touch nudges have very small effects on purchasing behavior and the effects are short-lived. I think it's important for policy makers to know this, especially since publication bias often leads to an over-representation of one-off studies showing big effects of these small interventions over relatively short periods of time. Here, we see much more modest effects and they are short-lived. This suggests consumers are willing to try these items but then revert to their preferred brands because the healthier options are less satisfying. I also liked that the authors were realistic in terms of how likely it is that grocers (if given the option) will comply with directives to give premium shelf space to healthy items that don't sell well. I wish the authors would be more specific in their recommendations -- what do they think would work based on their observations? What should policy makers take away from this set of studies? It's an interesting discussion/debate because we see a lot of proposed legislation directed at making (un)healthy options (less)/more available - which doesn't appear to be all that effective in terms of getting consumers to change their behavior. The authors could also do a better job of tying their results in with results from other studies at grocery stores (see Cardario & Chandon, 2019 for a very comprehensive review) or that focus on healthy eating.

Any attachments provided with reviews can be seen via the following link:

[LINK]

5 Dec 2021

Submitted filename: Response to reviewers CGF 6 trials 02 12 21.docx

Click here for additional data file.

17 Jan 2022

Dear Dr. Piernas,

Thank you very much for submitting your revised manuscript "Natural experiments testing availability, positioning, promotions and signage of healthier food options within major UK supermarkets: Analysis of six non-randomised controlled trials" (PMEDICINE-D-21-03912R2) for consideration at PLOS Medicine.

Your paper was evaluated by an associate editor and discussed among all the editors here. It was also discussed with an academic editor with relevant expertise, and sent back to the reviewers. The reviews are appended at the bottom of this email and any accompanying reviewer attachments can be seen via the link below:

[LINK]

In light of these reviews, we will not be able to accept the manuscript for publication in the journal in its current form, but we would like to consider a revised version that addresses the reviewers' and editors' comments. We cannot make any decision about publication until we have seen the revised manuscript and your response, and we plan to seek re-review by one or more of the reviewers.

In revising the manuscript for further consideration, your revisions should address the specific points made by each reviewer and the editors. Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments, the changes you have made in the manuscript, and include either an excerpt of the revised text or the location (eg: page and line number) where each change can be found. Please submit a clean version of the paper as the main article file; a version with changes marked should be uploaded as a marked up manuscript.

In addition, we request that you upload any figures associated with your paper as individual TIF or EPS files with 300dpi resolution at resubmission; please read our figure guidelines for more information on our requirements: http://journals.plos.org/plosmedicine/s/figures. While revising your submission, please upload your figure files to the PACE digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at PLOSMedicine@plos.org.

We hope to receive your revised manuscript by Feb 07 2022 11:59PM. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

We ask every co-author listed on the manuscript to fill in a contributing author statement, making sure to declare all competing interests. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. If new competing interests are declared later in the revision process, this may also hold up the submission. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT. You can see our competing interests policy here: http://journals.plos.org/plosmedicine/s/competing-interests.

Please use the following link to submit the revised manuscript:

https://www.editorialmanager.com/pmedicine/

Your article can be found in the "Submissions Needing Revision" folder.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

We look forward to receiving your revised manuscript.

Sincerely,

Callam Davidson,

Associate Editor

PLOS Medicine

plosmedicine.org

-----------------------------------------------------------

Requests from the editors:

Please revise your title to ‘Testing availability, positioning, promotions and signage of healthier food options within major UK supermarkets: evaluation of six non-randomised intervention studies’.

Line 20: Please refer to the studies as ‘non-randomised controlled intervention studies’ (including 'controlled' where appropriate) rather than non-randomised controlled trials (please check throughout the manuscript for consistency).

Lines 24 and 26: Please amend text to “… was associated with decreases in sales”, or similar at line 26.

Your author summary requires some trimming – please aim for 2-3 bullet points per section (I appreciate that, given the number of interventions, the ‘What did the researchers find’ section may need to be slightly longer than this, so aiming for no more than nine bullet points (preferably single sentences) across the three sections would be OK.

The final line in your Financial Disclosure (‘We thank the consumer goods…’) would be better placed in an Acknowledgements section at the end of the main text (as it does not pertain to funding received).

As in the previous draft, it appears the PDF conversion process has introduced some formatting errors (see line 188 for an example).

Please avoid the term “effects” at line 434. Refer instead to associations.

Please check your references consistently follow our guidance (https://journals.plos.org/plosmedicine/s/submission-guidelines#loc-references) e.g., use of ‘et al’ only after listing the first six authors.

Comments from the reviewers:

Reviewer #1: Thanks authors for their great effort to improve the manuscript. The authors have comprehensively addressed all my comments. I am satisfied with the response and revision. No further issues needing attention.

Reviewer #2: Thanks for responding to my previous comments. There are two issues that I still think deserve further attention.

1. IMD. I am still not clear at what geographical level IMD was calculated. Was is lower level super output area, middle level super output area, ward, local authority? All of these would involve linking postcode to another geography, but it is not clear from your ms which other geography you have used. I suspect you have used lower level super output area. The have an average population of just 1500 people. Hence the IMD of LSOA of location is likely to be a very poor marker of the deprivation of the people who shop in a store - particularly large out-of-town stores which most people drive to, and city center stores with high non-local usage. I still don't see this point discussed in consideration of the equity of effects seen.

2. I still don't see discussion of my previous comment "A key limitation on how the findings can be interpreted is the possibility of an intervention modality by food category interaction - as different food categories were selected for each intervention modality, we can't tell if the effects reported are specific to the food category, the intervention modality, or the interaction of the two. I'd like to see more consideration given to this throughout." The conclusions are stated in terms of modality. On occasion there are examples of categories studied in brackets. But it's not made clear that the effects seen may be specific to the categories studied or a modality x category interaction - ie some intervention modalities only work for some food categories.

Reviewer #3: Thanks to the authors for a thoughtful revision and set of comments on both papers. I have read the revised manuscripts, the revision notes, and my previous reviews. I remain positive about both papers, but many of my initial concerns remain. First, I don't think the authors can make the causal claims their titles or discussions of the results suggest. Other reviewers have also pointed this out. In my view, these papers are about seeing how various nudges are playing out in the field not about testing interventions. I equate this to non-experimental data that comes in after a drug that did well in clinical trials has been on the market for a while. You cannot overcome a lack of randomization with statistical massaging and/or a pre-post design. You cannot overcome underpowered studies with statistical massaging either. I don't think the word experiment should appear anywhere these papers. There is no natural experiment, as it would require some kind of exogenous shock that resulted in equivalent stores with equivalent populations be essentially randomized to treatment vs. control. I also have reservations about treating these studies as experimental given how little access the researchers had to interventions that were "selected, developed, and implemented" by the retailers -- who I assume are not experimentalists or data scientists. I understand the authors' desire to make a strong policy argument here and it's impressive to have so much data from several retailers trying to implement various nudges simultaneously. However, given the lack of data that's available for controls and robustness checks, I am very hesitant to sign off on a paper that makes causal claims about the viability of certain interventions. I think there are many interesting insights to be gleaned from the data but I think the authors should tone down their language that these results provide any sort of definitive test. I belabor this point because it's easy for keywords that appear in titles and abstracts to become sound bites for journalists and agenda-driven policy makers who never dig a little deeper into the work to realize that the claims cannot be substantiated by the data.

Any attachments provided with reviews can be seen via the following link:

[LINK]

20 Jan 2022

Submitted filename: Response to reviewers CGF 6 trials 20 01 2022.docx

Click here for additional data file.

15 Feb 2022

Dear Dr. Piernas,

Thank you very much for re-submitting your manuscript "Testing availability, positioning, promotions and signage of healthier food options within major UK supermarkets: evaluation of six non-randomised controlled intervention studies" (PMEDICINE-D-21-03912R3) for review by PLOS Medicine.

I have discussed the paper with my colleagues and the academic editor and it was also seen again by two reviewers. I am pleased to say that provided the remaining editorial and production issues are dealt with we are planning to accept the paper for publication in the journal.

The remaining issues that need to be addressed are listed at the end of this email. Any accompanying reviewer attachments can be seen via the link below. Please take these into account before resubmitting your manuscript:

[LINK]

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

In revising the manuscript for further consideration here, please ensure you address the specific points made by each reviewer and the editors. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments and the changes you have made in the manuscript. Please submit a clean version of the paper as the main article file. A version with changes marked must also be uploaded as a marked up manuscript file.

Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. If you haven't already, we ask that you provide a short, non-technical Author Summary of your research to make findings accessible to a wide audience that includes both scientists and non-scientists. The Author Summary should immediately follow the Abstract in your revised manuscript. This text is subject to editorial change and should be distinct from the scientific abstract.

We hope to receive your revised manuscript within 1 week. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

We ask every co-author listed on the manuscript to fill in a contributing author statement. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT.

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript.

Please note, when your manuscript is accepted, an uncorrected proof of your manuscript will be published online ahead of the final version, unless you've already opted out via the online submission form. If, for any reason, you do not want an earlier version of your manuscript published online or are unsure if you have already indicated as such, please let the journal staff know immediately at plosmedicine@plos.org.

If you have any questions in the meantime, please contact me or the journal staff on plosmedicine@plos.org.

We look forward to receiving the revised manuscript by Feb 22 2022 11:59PM.

Sincerely,

Callam Davidson,

Associate Editor

PLOS Medicine

plosmedicine.org

------------------------------------------------------------

Requests from Editors:

Please update the manuscript title to ‘Testing availability, positioning, promotions and signage of healthier food options and purchasing behaviour within major UK supermarkets: evaluation of six non-randomised controlled intervention studies’ (my apologies for omitting this suggestion in my previous revision letter).

In the Data Availability Statement, please provide further details as to how an interested researcher could establish contact with the retailer to request permission to access the data (this could be via an intermediary such as the University Research Office or the Institutional Review Board – please note that it cannot be via a study author). Feel free to contact me directly to discuss (cdavidson@plos.org).

Line 27: For consistency with the rest of the sentence, please refer to ‘lower energy biscuits’ rather than ‘healthier’.

Line 46 – please remove the protocol registration details from the abstract.

Line 54 – please replace the full stop on this line with a comma (‘…may be effective, but most…’).

Line 56 – please begin the sentence ‘As part of a multi-retailer partnership…’ as a new bullet point.

Line 63 – please begin the sentence ‘However, there was no evidence’ as a new bullet point.

Line 75 – in the interest of brevity, please remove this final bullet from the Author Summary.

Line 113 – please updated to ‘…by the retailers and aimed at improving…’.

There are no corresponding asterisks in Figure 1 despite their use in the legend. I would suggest removing the asterisks in the legend rather than adding them to the figure as I do not feel they are necessary for clarity.

Related to the above, I would suggest updating the Figure 1 legend to read ‘…estimates coming from five of the interventions included in this study, in the following order: Frozen chip range changes (regular frozen chips); Biscuit range changes (regular range biscuits, lower energy range biscuits)…’ etc. I personally find the use of the alphabetical labels in the legend confusing as it suggests a corresponding label in the figure which is not present.

For the relevant panels in Figure 2, please either show the y-axes beginning at zero or show a break in the axes.

For Table F in S2 Appendix, ensure all symbols in your legend are present in the Table.

Please remove subheadings from the Discussion section.

Line 374: ‘provides’.

Consider separating the strengths and limitations paragraph into two paragraphs.

Please check all references for accuracy against our Submission Guidelines (https://journals.plos.org/plosmedicine/s/submission-guidelines#loc-references), in particular those for official reports (there are some issues with the authors listed for references 2, 6, and 7, and references 9 and 25 are also not presented consistently).

To help us extend the reach of your research, please provide any Twitter handle(s) that would be appropriate to tag, including your own, your coauthors’, your institution, funder, or lab.

Comments from Reviewers:

Reviewer #2: Thanks for responding fully to my previous comments. I've not further comments. Congratulations on some great work.

Reviewer #3: Thanks to the authors for their responsiveness to the requests of the entire review team. I am happy with the changes and congratulate them on two very nice papers!

Any attachments provided with reviews can be seen via the following link:

[LINK]

17 Feb 2022

Submitted filename: Response to reviewers CGF 6 trials 17 02 2022.docx

Click here for additional data file.

23 Feb 2022

Dear Dr Piernas,

On behalf of my colleagues and the Academic Editor, Dr Jean Adams, I am pleased to inform you that we have agreed to publish your manuscript "Testing availability, positioning, promotions and signage of healthier food options and purchasing behaviour within major UK supermarkets: evaluation of six non-randomised controlled intervention studies" (PMEDICINE-D-21-03912R4) in PLOS Medicine.

Before your manuscript can be formally accepted you will need to complete some formatting changes, which you will receive in a follow up email. Please be aware that it may take several days for you to receive this email; during this time no action is required by you. Once you have received these formatting requests, please note that your manuscript will not be scheduled for publication until you have made the required changes.

When making the formatting changes, please also review your references to ensure they meet our guidelines (https://journals.plos.org/plosmedicine/s/submission-guidelines#loc-references). Journal name abbreviations should be those found in the National Center for Biotechnology Information (NCBI) databases (https://www.ncbi.nlm.nih.gov/nlmcatalog/journals).

In the meantime, please log into Editorial Manager at http://www.editorialmanager.com/pmedicine/, click the "Update My Information" link at the top of the page, and update your user information to ensure an efficient production process.

PRESS

We frequently collaborate with press offices. If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximise its impact. If the press office is planning to promote your findings, we would be grateful if they could coordinate with medicinepress@plos.org. If you have not yet opted out of the early version process, we ask that you notify us immediately of any press plans so that we may do so on your behalf.

We also ask that you take this opportunity to read our Embargo Policy regarding the discussion, promotion and media coverage of work that is yet to be published by PLOS. As your manuscript is not yet published, it is bound by the conditions of our Embargo Policy. Please be aware that this policy is in place both to ensure that any press coverage of your article is fully substantiated and to provide a direct link between such coverage and the published work. For full details of our Embargo Policy, please visit http://www.plos.org/about/media-inquiries/embargo-policy/.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Thank you again for submitting to PLOS Medicine. We look forward to publishing your paper.

Sincerely,

Callam Davidson

Associate Editor

PLOS Medicine