Could Your Shopping Save Lives?

Each purchase is a piece of data, as is the date, time, and location of purchase. Stores already use that data for monitoring stock and price trends, as well as for advertising. But what if that data could be used for more than marketing? Some researchers have begun to look for the links between what we buy and our health in the hope that their discoveries could save lives. 

What does our shopping say about us?

To illustrate how, let’s revisit our friend in the queue with the flu medicine. Let’s imagine they are buying it for a friend who has COVID-19. Researchers have explored how sales of cold and flu remedies could improve predictions of the spread of COVID. Usual models for predicting disease spread make use of factors such as socio-demographics, but the one that also included shopping data was a better match for the actual deaths from COVID. 

To massively over-simplify, sales of cold and flu medications corresponded with more COVID cases. But since shopping data is already being collected automatically, there is potential for it to be an efficient way to keep track of how a pandemic is evolving and stay one step ahead of its spread.   

The data is already there, every time you swipe a loyalty card it is saved, and you can go back years easily.

Dr Romana Burgess

Some patterns are less obvious, like seeing disparities in health and wealth through how many red, blue, and purple foods people buy. 

Fruits like strawberries owe their color to a chemical compound called anthocyanin. It has many potential health benefits, ranging from improved diabetes control to reduced risk of heart disease and some cancers. Researchers wanted to know how much and from where people were getting their anthocyanins in their diet. 

Approximately half of the UK’s anthocyanin intake came from wine (I’m not sure what this says about us as a country) with the rest coming predominantly from seasonal berries, plus a smattering of other purply food like red onions and black olives. When they looked at which neighborhoods had the highest anthocyanin intake, they were generally those known to be wealthier and healthier too. 

On a more individual level (as “individual” as a dataset going into the millions of transactions can be), researchers have looked to the humble meal deal. The lunchtime staple of main, drink, and snack for a set price – a mainstay of British supermarkets and convenience stores – makes it a surprisingly powerful tool. When everything costs the same, it is easier to see what people are prioritizing besides price. 

The researchers in this case wanted to understand more about the factors affecting weight changes and how people’s caloric intake varied over time. By examining meal deals, they found that people who bought lunch earlier in the day had less calorific meals on average compared to people buying later in the day. The same applied to the time of year – as the colder months of November and December rolled around people’s averages inched towards more calorie-dense options. In January the calories slid back down as New Year’s resolutions marked the end of the festive season. 

They also found that people who were most affected by the daily and seasonal trends tended to eat more calories overall. By understanding what factors affect people’s patterns of consumption, it might be possible to find ways to help people make healthier choices.  

The potential of this data is clear, and it has several advantages. Dr Romana Burgess from the Digital Footprints Lab at the University of Bristol explained to me: “The data is collected in real time in the real world, and unlike questionnaires it is objective (you either bought or didn’t). The data is already there, every time you swipe a loyalty card it is saved, and you can go back years easily.” 

Despite this, there are many gaps, because shopping data contains so little information beyond what was bought, when, and where. Researchers can only see data linked to that specific loyalty card and persuading retailers to work with them at all is a challenge in itself. Dr Burgess describes how “companies can be reluctant to work with us – they worry that they’ll be put at a disadvantage if the results indicate that people should stop purchasing products”. 

Not every store even has a loyalty card system, meaning it’s impossible for researchers to build a complete picture of everything people are buying. Nor can they tell for sure who is using the products (like our flu medicine buyer getting it for a friend), or if they’re getting used at all. However, the level of scrutiny required for this is more than most consumers are likely to be comfortable with. 

Where next for this research?

To get a better idea of what people are comfortable with, the Bristol-based team have been consulting the public in a recent project based at science center We The Curious. Visitors of all ages shared their thoughts through the medium of a mock shop complete with toy food and toddler-sized baskets.  

People were generally positive towards the idea of sharing shopping data for health research, with the condition that they would want their data anonymized and assurance that it wouldn’t be used for marketing. There was no shortage of suggestions for future research directions. Nutrition was a popular topic, along with the impact of ultra-processed foods and cost of living on health. Questions about how specific health conditions and shopping habits intersect also came up.  

To answer these kinds of questions, and find new connections between purchases and people’s health, linking shopping data to health records such as prescriptions or blood tests could be key.  

So far the studies doing this have been on a smaller scale, such as one aiming to improve diagnosis and early treatment of ovarian cancer. It’s difficult to catch early – smear tests do not detect it and frustratingly vague symptoms like bloating, fatigue, and pain in the abdomen and/or pelvis overlap with many other illnesses. The researchers recruited women who had ovarian cancer and looked for any telltale signs in what they were buying prior to being diagnosed. Sure enough, antacids, more comfortable clothes, and vitamins related to fatigue were among the things women were using to self-treat symptoms. The specific constellation of purchases could be seen up to a year before their eventual diagnosis.  

Discoveries like this could be the first step towards a future where shopping data could be used by doctors as a tool to help with diagnosis. 

There’s no need to rush off to download your loyalty card data yet – which you can do, by the way, at least in the UK. Just be prepared to venture into the depths of company websites to receive a jumble of numbers (the Digital Footprints team are also working on creating a tool to summarize these in a more human-readable way). But the next time you swipe your loyalty card, or go to judge a stranger’s shopping basket, spare a thought for how the information contained there could perhaps help save a life.