Task 1.1: Needs, characteristics and preferences of the target population
We will perform subgroup analyses of previously meta-analyzed studies and quantitative analysis in large prospective studies (the Hoorn Study, SPOTLIGHT, EPIC-NL and KORA) to test for moderation by SES, sex, BMI, eating behaviors and stress, and to study which specific dietary and physical activity behaviors are related to cardiometabolic risk across subgroups. Second, Participatory Action Research (PAR) methodology (e.g. focus groups, participatory mapping, transect walks) will be used to evaluate quantitative findings and identify preferences towards nudging and content of a physical activity application.
Task 1.2: Needs, characteristics and preferences of the supermarkets
A multidisciplinary qualitative assessment following PAR principles, including qualitative interviews on perceived extent of the problem, responsibility and ideas for intervening and scaling-up, will be performed with stakeholders from different levels within supermarkets. Secondly, an interactive workshop for validation and further input using Venn diagramming, including mapping actors and their relationships, root cause analysis and preference ranking (for comparison of options) will be held.
Task 1.3: Co-creation of intervention components
We will create attractive designs for the intervention components and adapt the content of the virtual supermarket environments and messages delivered by the physical activity app. This includes the use of a Virtual Reality (VR) devices, such as the HTC Vive head-mounted displays (HMDs).
Task 2.1: Testing of pricing and nudging supermarket intervention
We will test the efficacy of different types of nudging (‘social proof nudges’, ‘default nudges’ and ‘salient nudges’) in a lab setting, with comparison conditions of a pricing strategy and a control condition, using a randomized between-subjects design with five factors. The most promising of theses nudges will be piloted in controlled virtual supermarket settings, using the same controls and the same study design, complemented with exit interviews to test for nudging transparency and awareness of nudge exposure.
Task 2.2: Testing of intelligent and individually tailored physical activity intervention
We will perform a content-valuation and feasibility study with participants thinking out loud while using the tailored physical activity app. Next, a single case design will be used to establish preliminary efficacy in participants testing the app during 8 weeks while a set features of the app (i.e., geo-fencing, self-monitoring, social comparison, goal setting) will be switching on/off every week. Finally, qualitative interviews about the user experience, and preferences for app features will be performed.
Task 2.3: Interim evaluation and adaptation of intervention components
Based on Task 2.2, the intervention components will be adapted to be suitable for implementation in a real-life setting. A workshop will be organized with representatives of the target group and supermarkets to formulate an evaluation framework, including the selection of process and outcome indicators and methods of data collection.
Task 3.1: Implementation of the intervention
Using the most promising intervention components from WP 1 and 2, we will implement the intervention in a real-life setting of 8-12 single-supermarket municipalities characterized by a low SES population. Shoppers will be offered a free medical checkup and will be characterized in terms of diet and physical activity, including questionnaires, supermarket purchases and accelerometers.
The supermarkets will be randomized to one of four intervention arms:
1) a control group
2) intervention group receiving ‘nudges’ only (environmental-level nudges, e.g. product placement, product promotion, foods at checkout)
3) intervention group receiving ‘nudges’ and the physical activity app (i.e. tailored, time-specific and context-specific feedback and support to improve physical activity
4) intervention group receiving ‘nudges’, the physical activity app and ‘pricing’ (i.e. lowering of prices of healthy products, and/or increasing prices of less healthy products
The intervention phase will be 12 consecutive months to account for seasonal variation in shopping and physical activity behavior and allow measurement of long-term effects.
Task 3.2b: Evaluation of the intervention
The effect of the intervention will be evaluated according to the intention-to-treat principle at 3, 6 and 12 months. Primary outcomes include: changes in blood lipids, blood pressure, waist circumference and HbA1c at 6 and 12 months. Secondary outcomes include weekly purchases of targeted foods based on grocery till receipts, dietary quality and physical activity, at 3, 6 and 12 months. Tertiary outcomes include habit formation of health behaviors and acceptance of nudges by customers, at 12 months. Intervention effects will be analyzed using multilevel analyses with a random intercept at the subject level and including supermarket as a fixed effect to adjust for clustering of individuals within supermarkets, and accounting for subgroup differences (e.g. based on sex).
Task 4.1: Identify relevant meso- and macro-level actors and factors to facilitate scaling up
We will map the actors and factors influencing the supermarket environments (e.g., contracts with suppliers, competition, profit margins) through a systematic literature review and policy document analysis. Relevant actors will be invited for an in-depth interview to identify core elements and leverage points in the web of interacting factors. Additionally, we will explore options for supermarkets to adopt and promote the tailored physical activity application. An example could be allotting vouchers for free fruit and vegetables after reaching a certain physical activity goal. Advice will be asked from our User Committee that also includes a health care insurer and local policy makers.
Task 4.2: Monitoring of the identified factors
We will use observations, interviews, questionnaires and focus group discussions with supermarket stakeholders to monitor the identified factors that may facilitate or constrain horizontal and vertical up-scaling of the intervention and develop strategies to overcome the identified barriers in order to stimulate the process of scaling up. A reflection session with relevant stakeholders (in which (preliminary) results are discussed, barriers are listed on a so-called ‘Dynamic Learning Agenda’, underlying causes are reflected upon and strategies are formulated) will be organized twice a year according to the Reflexive Monitoring in Action framework.