Eating plenty of fruit and vegetables is key to staying healthy and avoiding diseases such as heart disease and stroke. But it’s often easier said than done.

Places where many people eat poorly are often called “food deserts”, and their existence has typically been blamed on a lack of nearby supermarkets or grocery stores.

However, my colleagues and I have discovered food deserts exist in the heart of one of Europe’s biggest and most cosmopolitan cities, surrounded by local shopping options.

In new research published today in PLOS Complex Systems, we analysed hundreds of millions of Tesco supermarket transactions across London and discovered surprising patterns in who buys what kind of food and where they do it.

Our results show the factors that influence how people eat are complex – with implications for nutrition in cities around the world.

The rise of ‘food deserts’

The term “food deserts” emerged in the late 1990s to describe areas where residents were denied access to affordable, healthy food due to a lack of supermarkets or poor transport links. As a result, food deserts have usually been defined by distance to supermarkets.

More recent research has revealed the picture is more complex. It’s not just how close the person is to a supermarket and how affordable the food is. There are other factors, such as how many shops are available, and whether the shops stock culturally appropriate foods and accept different forms of payment.

Our new paper builds on this. It takes a different approach to identifying food deserts – based on what people actually put in their shopping baskets.

420 million shopping lists

We used a dataset of Tesco grocery purchases containing 420 million anonymised transactions from 1.6 million London Clubcard holders to analyse residents’ food buying, based on the areas linked to shoppers’ loyalty cards.

Two clear purchasing patterns emerged from the data – one involving sugary, processed and high-carbohydrate foods usually considered unhealthy, the other involving purchases of fresh fruits, vegetables and meat, usually considered to be healthier.

We then mapped the areas of London where each of these purchasing patterns was most common. This revealed distinct geographic patterns.

Inner northwest London had the most nutritious purchasing behaviour – with high fruit, vegetable and fish purchases. The east and outer west of London followed a less nutritious pattern, high in sweets and soft drinks.

Because our analysis is based on supermarket purchases, it doesn’t capture all food consumption – such as meals eaten out, takeaway orders, or shopping from smaller local stores.

Still, using real transaction data offers a major advantage over traditional surveys, which often rely on what people say they eat rather than what they actually buy.

Even in cities with stores nearby, some neighbourhoods are still ‘deserted’ of nutritious options. Joel Carrett/AAP

Lower income linked to less nutritious food

Here’s what emerges when we define food deserts by what people actually buy. Even in cities with stores nearby, some neighbourhoods are still “deserted” of nutritious options. Often, it’s not about distance at all – it’s about economic and social factors.

We then analysed how demographic and socioeconomic factors such as age, income, Black, Asian and minority ethnic populations, car ownership, and walk time to stores relate to diet quality across London.

We found that income and the proportion of Black, Asian and minority ethnic residents were among the strongest factors linked to diet quality. But their influence varied across the city. Lower income was linked to less nutritious food purchasing throughout London, and this effect was strongest in parts of the east and west.

This suggests that affordability and social disadvantage shape what’s within reach – even when supermarkets are nearby.

Some factors that might be expected to influence diet had surprisingly little effect. For example, car ownership was linked to less nutritious purchases in certain areas, while walk time to stores had very low association with diet quality.

Together, these patterns suggest two things: the reasons people eat unhealthily are local and vary from place to place, and they’re shaped more by social and economic conditions than by how close shops are.

Global relevance

While our study focuses on London, the findings have relevance beyond the United Kingdom.

The same inequalities that shape London’s dietary health also exist in Australian cities. Australia is highly urbanised, with around 73% of the population living in major cities.

Here too, poor diet is one of the nation’s leading causes of preventable disease. In 2022, 66% of Australian adults and 26% of children were living with overweight or obesity, according to the Australian Institute of Health and Welfare (AIHW). On top of that, the average servings of fruit and vegetables have declined across all age groups since 2017–18, according to AIHW data.

A similar data-driven approach using anonymised grocery transaction data from sources such as Woolworths Everyday Rewards card or Coles’ Flybuys programs could help reveal which communities face the greatest nutritional constraints, and why.

Another important takeaway from our work is that food access is not a one-size-fits-all problem. Understanding what people buy – not just where they live – is key to creating healthier and more equitable food environments.

Focusing on actual purchasing behaviour allows policymakers to design more effective, community-informed interventions that promote fairer, healthier food choices.