Data-driven approach to evaluating and improving Scottish food environments

Abstract

The persistent rise in the prevalence of diet-related chronic diseases has prompted a shift from traditional individual behaviour change interventions to policies to improve the healthfulness of food environments. The food environment is the interface between people and the wider food system. It encompasses all places where people access food, including retailers, restaurants, pubs/bars, cafés/coffee shops, takeaways, mobile food vans, schools, universities, workplaces, and charities as well as deliveries from these places. Whilst research on food environments in the UK and globally is rapidly expanding, very few studies have been conducted in Scotland. Scotland has disproportionately high rates of diet-related diseases, particularly in socioeconomically deprived areas. To address this, one of the two initial priorities of Scotland’s Population Health Framework 2025-2035 is to develop a whole system approach to improve food environments, ensuring a healthy diet is accessible and affordable to all.

The overarching goal of this thesis was to provide evidence to inform data-driven improvements to the Scottish food environment. To achieve this goal, four major aims were developed, each of which forms a chapter in the thesis. The first aim, Chapter 2, was to identify and narratively synthesise recent evidence regarding UK food environments. To accomplish this aim, I conducted a systematic review of 7 databases. After reviewing 31,457 titles and abstracts, I identified 312 articles on the UK food environment published in the past two decades. The majority of articles (32%) focussed on only one of six food environment domains: availability, defined as the presence of specific food items or outlets within a given physical space. Of the remaining domains, 30% of articles assessed product characteristics/quality, 10% assessed promotion, 6% assessed sustainability, 5% assessed affordability and none assessed convenience. With regard to the type of food environment studied, most articles (67%) focussed on food retailers and only 16% assessed restaurants. Finally, obesity was the only health outcome studied extensively. This systematic review highlighted research gaps and laid the foundation for subsequent chapters, each of which addressed a specific gap. Namely, the domain of convenience (Chapter 3) and studies in the geography of Scotland (Chapters 4 and 5) covering the entirety of the food environment, from food retailers to takeaways to full-service restaurants (Chapter 4).

The aim of Chapter 3 was to develop a conceptual model and metric for measuring convenience in food environments. This was done through a scoping review on whether and how food environment frameworks integrated convenience and in-depth interviews with food environment experts. Convenience was proposed to be objectively measured as the weekly time spent on food-related tasks. Food-related tasks included planning meals, procuring food, cooking, consuming food and cleaning up. The model also included two sub domains within convenience: product and personal characteristics as well as interaction with other food environment domains. Product characteristics encompass the inherent attributes of a food item, such as the level of processing and packaging, which contributes to or hinders its convenience. Personal characteristics (i.e., income, family size, gender, food choices, cultural beliefs and practices) interact with product characteristics and influence time spent on food-related tasks and whether the food environment is perceived as ‘convenient.’

The aim of Chapter 4 was to characterise food outlets across Scotland and to assess the association between the food environment and neighbourhood deprivation. Using publicly available food business data from the Food Standards Agency (2024), and a novel approach combining text similarity matching with the GPT-4 large language model, I classified 31,135 food businesses as out-of-home (OOH), including restaurants, pubs, cafés and takeaways; retail, including supermarkets and other establishments that primarily sell non-food products with a limited range of food products such as pharmacies; and other, including mobile caterers, charity organisations and home caterers. Most (60%) of the food outlets were OOH, 28% were retail and 12% were other. The density of OOH outlets (1.9 per km2) was over twice that of retail (0.8 per km2).

Glasgow City had the highest OOH density (18.4 per km2), while Argyll and Bute, Western Isles and Highlands had the lowest density for both OOH and retail (≤0.03 per km2). Compared to the most deprived neighbourhoods, the least deprived neighbourhoods had more restaurants/cafés (39% versus 23% of food outlets, respectively) and fewer takeaways/sandwich shops (15% versus 23% of food outlets, respectively). This analysis revealed that OOH outlets far outnumber retail in all of Scotland, with takeaways more common in deprived areas and restaurants/cafés in affluent ones. It also highlights unique food environments in different local authorities.

The aim of Chapter 5 was to quantify the consumption of home-prepared, ready-to-eat and OOH foods and its association with diet quality among adults 16+ years living in Scotland. Using 24-hour dietary recall data from the nationally representative Scottish Health Survey (2021), I classified all reported food items as home-prepared, ready-to-eat or OOH.

Home-prepared foods were defined as those cooked at home using raw ingredients (such as fruit, milk, flour, etc.); frozen ingredients (including frozen vegetables or fruit); ingredients obtained from sources like farmer’s markets, butchers or fishmongers; and items explicitly designated as home-cooked, home-baked or home-grown. Foods and drinks eaten as-purchased (e.g., biscuits, crisps, breakfast cereals, juice and soft drinks) and pre-prepared convenience foods that only require heating (e.g., frozen and refrigerated ready meals) were classified as ready-to-eat. All food and drinks prepared outside the home including restaurants, pubs and cafés, as well as food prepared outside the home but eaten in the home (e.g., takeaway and delivery food) was defined as OOH.

The results showed that the weighted mean energy contributions from home-prepared and ready-to-eat foods were similar (~45% of daily calories), while the contribution from OOH foods was relatively low (9%). The highest consumption of home-prepared foods was observed in those aged 65–74 years, accounting for weighted mean (SD) 53% (20%) of total calories.

Participants aged 16–24 years were the highest consumers of ready-to-eat foods [weighted mean (SD) 51% (21.5%) of total calories]. Those aged 25-34 years were the highest consumers of OOH foods [weighted mean (SD) 13% (20%) of total calories]. Participants in the least deprived areas consumed a higher proportion of energy from home-prepared foods, while those in the most deprived areas obtained more calories from ready-to-eat and OOH foods: weighted mean (SD) intake ranged from 43% (23%) to 50% (20%) for home-prepared foods across the most to least deprived quintiles; and likewise, from 48% (22%) to 43% (19%) for ready-to-eat foods and 9% (18%) to 7% (14%) for OOH foods. Adults with a larger proportion of energy from home-prepared foods were more likely to adhere to the Scottish Dietary Goals, whilst those with a larger proportion of energy from ready-to-eat and OOH foods were less likely to adhere to the Goals.

Overall, the results of this thesis highlight and begin to address research gaps in food environment research in Scotland and provide local evidence essential for understanding Scottish food environments. These findings can inform the development of forthcoming Good Food Nation Plans by local authorities and health boards, as well as national policies to improve food environments in Scotland.