Could Shopping Data Be Used to Predict Cancer and Diagnose It Earlier?

A recent study suggested that shopping data collected via loyalty cards may help predict the risk of cancer.¹ Now, researchers are exploring how to use that information and the implications it could have for patients.

The study showed that patients with ovarian cancer were more likely than those without ovarian cancer to have purchased over-the-counter (OTC) medications for indigestion and pain in the months prior to their diagnosis.¹ The study was conducted using data participants agreed to share from their loyalty card purchases at 2 retailers in the United Kingdom — a pharmacy and a grocery store.

To build on this work, the researchers are starting a bigger study to explore whether shopping data may be associated with the diagnosis of 10 different cancers. If the findings of the initial study can be replicated, the team will consider approaches for informing consumers if their purchasing patterns are linked to an increased risk of cancer.

However, experts note that it will probably be difficult to identify specific purchases that predict the risk of cancer or any single disease. And purchases would have to be highly predictive to justify informing — and potential scaring — the shoppers making those purchases. In addition, consumers and advocacy groups have raised concerns about the use of shopping data in research, such as whether the data could be used for marketing and for other uses consumers didn’t agree to. 

“We can potentially do very good things in health care [with consumer data], to try to detect conditions earlier and engage patients earlier,” said David Grande, MD, of the University of Pennsylvania, who researches consumer attitudes toward sharing personal data and was not involved in the UK study. 

“A lot of consumer data is far more predictive and informative about aspects of health [than medical records]…, but, at the same time, there is a lot of risk where patient privacy becomes a major concern,” Dr Grande said. 

Shopping Data and Cancer Risk: Findings and Implications 

The UK study, which was published in JMIR Public Health and Surveillance, included 153 women with ovarian cancer who were recruited from clinics across England, Scotland, and Wales.¹ The study also included 306 women of about the same age who did not have ovarian cancer and were recruited online. All the participants consented to share their loyalty card data from 2 retailers going back up to 6 years. 

The researchers found a significant association between buying indigestion and pain medication and receiving a diagnosis of ovarian cancer, and the association was highest at 8 months before diagnosis (odds ratio [OR], 2.91; 95% CI, 2.07-4.12). Indigestion medication was the major driver of the association, and its purchase started to spike at 9 months before diagnosis (OR, 1.38; 95% CI 1.04-1.83). 

The strongest association was seen among women who received a diagnosis of late-stage ovarian cancer. For this group, the purchase history was able to predict 68% of patients with ovarian cancer. 

Patients with late-stage disease “are the patients who are going to have the worst prognosis, compared with early-stage, and they are the ones that are experiencing symptoms, and the ones we really need to diagnose earlier,” said study author James M. Flanagan, PhD, of Imperial College London in the UK. 

Early diagnosis of ovarian cancer can be challenging because symptoms are vague — often indigestion and stomach and back pain — and even doctors may miss them. 

The ultimate goal of this research is to get patients to the doctor earlier and shorten the time from when they first experience symptoms to when they seek medical help, Dr Flanagan said. However, he expressed doubt that the purchase pattern will ever be associated strongly enough with ovarian cancer to justify telling consumers they may have a higher risk of ovarian cancer.  

Dr Flanagan and his colleagues are now delving deeper into the loyalty card data to explore whether there are combinations of purchases — perhaps OTC purchases plus certain food or vitamins — that may be strong predictors of ovarian cancer risk. 

It may also be the case that certain purchase combinations would be a more general signal of ongoing bloating, which could be due to ovarian cancer, other types of cancer, or conditions such as irritable bowel syndrome. And this signal could warrant sending a message to consumers that they may have some sort of digestive condition and may want to talk with their doctor.

There is already general guidance, such as from the National Health Service (NHS) in the UK, that people should see a doctor if they have ongoing bloating, noted Elizabeth Dolan, a researcher at the University of Nottingham in the UK who is doing computational analysis of the data sets from Dr Flanagan’s study.² 

Dolan noted, however, that the NHS guidance is not very specific. “I think these shopping data could help us get something more specific,” she said.  

Even if this research doesn’t lead to earlier diagnosis of ovarian cancer — or until it does — these findings can provide valuable information about the everyday behaviors and trajectories of people leading up to an ovarian cancer diagnosis, said James Goulding, PhD, of the University of Nottingham, who collaborates with Dolan and Dr Flanagan. 

“Just understanding the experiences that people go through is lacking in the data sets,” Dr Goulding said. He is also studying whether loyalty card data, and specifically the purchase of OTC cold medications, can improve upon the models that are used to predict when hospitals will see an uptick in COVID-19 cases and deaths.

This article originally appeared on Cancer Therapy Advisor