Study Design: Randomised controlled trial

Study Participants: Healthy volunteers from the UK

Planned Sample Size: 500

Planned Study Period: June - December 2021

Objectives and Outcome Measures:

1. Primary

   • Objective: To investigate the effect of promotions on the total number of calories selected
   • Measure: Difference in the absolute energy (kcal) of the final basket between the two groups

2. Secondary

   Objectives:

   • To investigate the effect of promotions on the nutritional content of the items selected overall, and across the individual target food categories
   • To explore the impact of removing promotions by socioeconomic status

   Measures:

   • Difference in the absolute sugar (g, kcal, %E), salt (g), and saturated fat (g, kcal, %E) of the final basket between the two groups
   • Difference in the absolute energy (kcal), sugar (g, kcal, %E), salt (g), and saturated fat (g, kcal, %E) between the two groups across the individual target food categories
   • Analyses of the primary and secondary outcomes specified above by gender, age group, ethnic group (White vs Non-White), BMI stratified into <30 and ≥30kg/m2 groups, education level (lower vs. higher), and household income (lower vs. higher)

3. Exploratory outcomes:

   • Objective: To investigate the acceptability of the intervention
   • Measures: Rating scores and open-ended answers from the follow-up questionnaire

BACKGROUND AND RATIONALE

As a nation people eat more energy, free sugars, saturated fat and salt than recommended for good health. These nutrients contribute to the burden of major chronic diseases, including cardiovascular disease and diabetes, principally through effects on blood cholesterol, blood pressure, insulin sensitivity or body weight. In addition, persistent inequalities in dietary intake underpin inequalities in long-term health outcomes. Despite decades of health promotion, dietary change is slow. Growing evidence suggests people need more than general knowledge about the healthfulness of foods to change behaviour.

Food purchasing is a key antecedent of food consumption and interventions to change food purchasing habits may be more impactful than interventions intended to change consumption at the moment of eating. Supermarkets account for the majority of the weekly expenditure on food and drinks, estimated to be around 87% of all UK retail grocery sales. Impactful, scalable and sustainable approaches within supermarkets are particularly needed in order to shift population-level intakes to be closer to the recommendations for calories, sugar, salt and saturated fat.

Previous systematic reviews of supermarket interventions have identified effective strategies to support dietary change. Evidence is generally stronger for economic (e.g. price interventions or price promotions), positioning and availability interventions. However, many of the interventions described in the literature are multi-component, and there is a challenge in attributing their effectiveness to any single strategy. In the 2020 Obesity Plan, the government in England laid out plans to introduce legislation to restrict the promotion of high fat salt sugar (HFSS) foods, by restricting volume promotions such as 'Buy One Get One Free'. This is based on observational evidence of purchasing habits. Specific evidence of the impact of removing promotions is of interest to policymakers and could help in shaping effective but proportionate policies.

Online supermarkets offer unique opportunities to deliver and support complex nutrition interventions that can often be tested and scaled up more rapidly and feasibly using digital technology than in-store interventions. There is some evidence that baseline purchases made online tend to be healthier than purchases in physical stores, but less is known about the effectiveness of interventions delivered online to change behaviour.

The purpose of this project is to test the effectiveness of the removal of price promotions on confectionery and snacks on the energy, sugar and saturated fat content of items selected while shopping online in a representative sample of UK adults. The proposed intervention can be easily implemented, and could reach large numbers of people, meaning that if it is effective it could have a significant population impact and be very cost-effective, even if the effect size is smaller than more intensive interventions.

INTERVENTION This study uses a bespoke virtual online supermarket shopping (OLS) platform, hosted by The University of Oxford, which emulates a real online supermarket for research purposes relating to food purchasing interventions (www.woodssupermarket.co.uk). The site is populated with approximately 23,000 current supermarket products drawn from foodDB (April 2019), a weekly updated database of food and drinks available for purchase in six UK online supermarkets.

Participants will be asked to imagine they are buying snacks for a night in with 4 friends, with a maximum limit of £10. They will be asked to select products from the following categories, which are aligned and named in the way items are presented in the store:

• Confectionery
• Biscuits and crackers
• Crisps, nuts and snacking fruit
• Cakes and tarts

The shopping task is estimated to take participants 10 minutes to complete. Participants will only be asked to complete the shopping task once. Participants will read a message with the following instructions:

"We would like you to do some online grocery shopping on a supermarket website. This is not a real supermarket, and you will not be asked to spend your own money.

We would like you to imagine you are buying snacks for a night in with 4 friends, up to a limit of £10. You should select products from the following categories:

• Confectionery
• Biscuits and crackers
• Crisps, nuts and snacking fruit
• Cakes and tarts We would ask you not to select any additional items from your usual shopping list.

When doing your shopping, try to imagine you are doing your own grocery shopping and choose foods that you and your friends would eat. You should choose the things you normally buy or wouldn't mind eating.

To make sure you get your reward, please make sure you buy at least £5 worth of products within the suggested food categories"

PARTICIPANT IDENTIFICATION AND RECRUITMENT 7.1. Study Participants Eligible participants will be UK adults who are responsible for a substantial proportion of their household grocery shopping, able to read and understand the instructions provided and able to provide online consent to take part in the study. Participants will be required to have access to a computer and the internet.

People who are following a vegan, gluten-free, dairy-free or sugar-free diet will not be eligible to participate to minimise the risk of unbalancing the groups with people choosing from a restricted selection of products.

STUDY PROCEDURES 8.1. Recruitment Participants will be recruited from the volunteer panel Prolific Academic (https://www.prolific.co).

8.2. Screening and Eligibility Assessment Panel members will be sent an email introducing the study and including a link to the participant information sheet. Panel members who are interested in taking part in the study will be asked to click on a web-link from the email, which will take them to the study registration website (on Qualtrics). At this registration website, they will be asked screening questions indicating their responsibility for shopping, if they have any dietary restriction, if they regularly buy products from the target categories for their household, and to confirm their age and country of residence. They will also be able to re-read the detailed participant information before consenting to participate.

8.3. Informed Consent After reading the detailed participant information sheet and answering screening questions to confirm they meet the inclusion/exclusion criteria; and if they are happy to proceed, participants will be asked to give consent electronically to take part in the study.

Written electronic versions of the Participant Information and Informed Consent will be presented to the participants detailing no less than: the exact nature of the study, what it will involve for the participant and the implications and constraints of the protocol. It will be clearly stated that the participant is free to withdraw from the study at any time for any reason and with no obligation to give the reason for withdrawal. The participant will be allowed as much time as wished to consider the information. Electronic Informed Consent will then be obtained by means of clicking on a tick box.

8.4. Randomisation Randomisation will be performed by the survey platform Qualtrics via computerised random number generation on a 1:1 basis with random block sizes and participants will be directed to a website that introduces an online shopping task. Allocation concealment is achieved, as participants are recruited from independent research panels and are being directed by automatic randomisation in the survey platform. Details of the intervention are described in Section 6.

8.5. Blinding Investigators will be blinded to intervention allocation and will not be able to manipulate any study parameter following the initial study set up, as all study procedures are taking part in the online platform. The outcome assessment is blinded, as it happens automatically in the online platform. The statistician who will analyse the data will be blinded to intervention allocation. Participants will only be aware of the trial arm that they are exposed to and will be unaware of the other trial arm.

8.6. Baseline Assessments Following consent and completion of the shopping task, participants will complete a baseline questionnaire on demographic, shopping, and health data

8.7. Follow-up Participants will complete a short post-intervention survey about the acceptability of the intervention in the online shopping task and their usual shopping behaviours.

8.8. Discontinuation/Withdrawal of Participants from Study Each participant has the right to withdraw from the study at any time. Participants will not be replaced as a sufficient sample size will be recruited to allow for non-completion rate. Withdrawal from the study, based on the pre-defined completion criteria, will result in exclusion of the data for that participant from analysis.

8.9. Definition of End of Study The end of study is the completion of the baseline questionnaire and post-intervention survey following the shopping task.

STATISTICS AND ANALYSIS 9.1. The Number of Participants There are no previously reported standard deviations of the mean difference from similar trials (e.g. where a one-off shop is done) to guide the estimation of the effect size and standard deviation, and, thus, the calculation of the sample size. However, the study by Brimblecombe et al. 2020 implemented a complex intervention within real supermarkets in Australia where promotions on high sugar products were removed, including confectionery and sugary drinks. This study found a total reduction of -22% in sales (g/MJ) in confectionery, a difference of -4.5% (SD 15) compared to control; and a reduction of -8% in sugar-sweetened soft drinks, a difference of -13% compared to control.

A total sample of 500 participants (250 in each group) would be needed to detect a minimum effect size of -4.5% in total kcal from the target categories, with 90% power, 5% alpha and 10% attrition.

9.2. Analysis of Outcome Measures

Completion of the task will be defined as spending at least £5 on products from the target categories.

Primary Analysis Linear regression models will compare the total energy (kcal) purchased in the two groups. Any baseline variables (e.g. demographic characteristics) which appear to have been imbalanced between groups will be adjusted in the model.

Secondary Analysis Linear regression models will compare the (i) total sugar (g, kcal, %E), (ii) total salt (g, kcal, %E) and (iii) total SFA (g, kcal, %E) purchased in the two groups.

Secondary analyses will also compare differences in total energy (kcal), total sugar (g, kcal, %E), total salt (g, kcal, %E) and total SFA (g, kcal, %E) across the individual target categories.

Secondary analyses of the primary and secondary outcomes specified above will be conducted by gender, age group, ethnic group (White vs Non-White), BMI stratified into <30 and ≥30kg/m2 groups, education level (lower vs. higher), and household income (lower vs. higher) provided there are sufficient numbers within each subgroup (n>=30).

Descriptive analysis The exploratory outcome measures will be discussed descriptively. A Statistical Analysis Plan will be finalised before conducting the statistical analysis.

DATA MANAGEMENT 10.1. Access to Data Direct access will be granted to authorised representatives from the University of Oxford for monitoring and/or audit of the study to ensure compliance with regulations.

10.2. Data Handling and Record Keeping All study data will be captured in a password protected secure database. The participants will be identified by a unique study specific number and/or code in any database. Personal data will not be collected in this study. Research data and records will be retained for at least three years after publication or public release of the work of the research and reviewed thereafter.

QUALITY CONTROL AND QUALITY ASSURANCE PROCEDURES The study will be conducted in accordance with the current approved protocol, relevant regulations and standard operating procedures.

ETHICAL AND REGULATORY CONSIDERATIONS 12.1. Declaration of Helsinki The Investigator will ensure that this study is conducted in accordance with the principles of the Declaration of Helsinki.

12.2. Approvals The protocol, informed consent form, participant information sheet and any proposed advertising material will be submitted to the ethical committee for approval.

The Investigator will submit and, where necessary, obtain approval from the above parties for all substantial amendments to the original approved documents.

12.3. Participant Confidentiality All data collected by participants will be anonymous. The participants will be identified only a participant ID number on all study documents and any electronic database. All documents will be stored securely requiring password-protected access and will be only accessible by study staff and authorised personnel.

Privacy and confidentiality of data is particularly hard to manage in Internet-based research because researchers are not in control of online communication networks, leading to the risk of third-party interceptions. The use of a secure link at the consent stage will be encouraged. Participants will be informed that their IP addresses will not be collected. As the data will be anonymised, it does not constitute personal data and the duties and obligations of the Data Protection Act do not apply. The research agencies will only share with the researchers a participant ID and no personal data.

12.4. Other Ethical Considerations Internet based research issues

1. Authentication The research agency who will recruit participants uses a number of mechanisms to authenticate responders such as checking for duplicate respondents by evaluating variables such as email address, matches across several demographic data, and device-related data through use of digital fingerprint technology.
2. Participant rights Participants will be free to withdraw themselves and their data at any point in the research. During the shopping task, clicking on a clearly displayed "exit here" button will lead participants to a quick debrief page that will give them the option to confirm that they do not wish their data to be retained for the study purposes. Participants will be clearly informed before giving consent that anonymity makes withdrawal following completion of the study difficult.

FINANCE AND INSURANCE 13.1. Funding The study is funded by the National Institute for Health Research (NIHR) Applied Research Collaborations (ARC Oxford).

13.2. Insurance This is a simple online task and any unintended or adverse effects due to participation cannot be foreseen. The University of Oxford maintains Public Liability and Professional Liability insurance that will operate in this respect.

PUBLICATION POLICY The Investigators will be involved in reviewing drafts of the manuscripts, abstracts, press releases and any other publications arising from the study. Authors will acknowledge that the study was funded by the NIHR Oxford ARC. Authorship will be determined in accordance with the ICMJE guidelines and other contributors will be acknowledged. As the participants' contact details will not be stored, the results will not be disseminated directly to study participants. A dissemination plan will be followed to engage with public health agencies, the industry, the media, and the

Inclusion Criteria:

• UK adults, aged ≥18 years.
• Able to speak and read English.
• Willing and able to give informed consent for participation in the study.
• Having access to a computer and Internet.

Exclusion Criteria:

• They are following a dairy-free or sugar-free diet.

A data sharing plan will be drafted after study closure and before data analysis

Investigators will be blinded to intervention allocation and will not be able to manipulate any study parameter following the initial study set up, as all study procedures are taking part in the online platform. The outcome assessment is blinded, as it happens automatically in the online platform. The statistician who will analyse the data will be blinded to intervention allocation. Participants will only be aware of the trial arm that they are exposed to and will be unaware of the other trial arm