Credit Card Purchasing Data Helps Hospitals Predict Your Health

After a quick glance through many of today’s Digital Innovation and Transformation Blogs, it was not surprising to see so many articles about how healthcare, life-science, and other companies in the field of medicine are using data and analytics. Recent headlines have focused on how organizations – like the ones many of my classmates have written about – are leveraging Big Data and analysis to better serve their customers. Pharmaceutical companies are mining patient data to better understand unmet needs; health insurance companies are analyzing clinical outcome data to determine potential patient risk and appropriate coverage. We are hearing similar stories to these more and more frequently. There is one organization in this space, however, that is starting to use data in a very creative way: Carolinas HealthCare.

Carolinas Healthcare operates approximately 900 care centers, including hospitals, nursing homes, doctors’ offices, and surgical centers. Unlike most healthcare systems which mine patient outcome and health data to reduce total time in high-cost areas of the hospital, or to maximize utilization of specific hospital units, Carolinas Healthcare has been purchasing customer data from brokers who cull public records, store loyalty program transactions, and credit card purchases – a very different model than its competitors.

Carolinas Healthcare has started to incorporate these data on 2 million people in North and South Carolina into algorithms that are designed to identify high-risk patients. The idea is that doctors can intervene before the patients get sick. While Carolinas Healthcare isn’t using new analytical tools (its algorithms are primarily based off of multivariable regression models), it is using “new” data to help inform its decisions, and is the first healthcare system to apply credit card data to health decisions.

Carolinas Healthcare is still in the early stages of acquiring, analyzing, and using this type of data. It has not yet fully changed its operating model to be able to efficiently leverage its analyses to drive change. In the months to come, it will need to better align its processes to ensure its resources (doctors and nurses, primarily) are able to interpret and make informed decisions based off of this data. Even with this novel idea to use consumer spending behavior to predict health risks, Carolinas Healthcare’s business model is only as strong as its weakest link, which right now appears to be its physicians who may be too busy to use this data.

Carolinas Healthcare executives are optimistic about how much money analyzing this type of data can save the entire system if done appropriately (which, at least hypothetically, would benefit both the consumers and Carolinas Healthcare). Others, however, are less optimistic, including one patient who said she is sick of having healthcare companies try to gather or steal her personal information: “it is one thing to have a number I can call if I have a problem or question; it is another thing to get unsolicited phone calls. I don’t like that…” Privacy experts also worry about the potential that this type of data analysis could erode doctor-patient relationships: “If the physician already has the information, the relationship changes from an exchange of information to a potential inquisition about behavior.”

For me, it will be interesting to see two issues play out: 1) how effective will Carolinas Healthcare analysis of this data be? Will consumer purchasing data really improve health-outcomes? And 2) will privacy advocates eventually win the battle and prohibit healthcare systems from using personal data to make health related decisions? Only time will tell.

Previous:

IBM Marketing Gets Tennis Moneyball Rolling

Next:

Gmail: ensuring a spam-free inbox with Machine Learning

Student comments on Credit Card Purchasing Data Helps Hospitals Predict Your Health

  1. Thanks for the post, Drew! Super interesting approach to using big data for healthcare, which is an industry that has certainly lagged behind the times in leveraging data and analytics in an effective manner. My questions on Carolinas Healthcare are the following: how they are actually going to use credit card information to help them predict consumer health issues, and what flexibility do they have to really do anything about high-risk consumer behavior?
    Are they tracking primarily medical related spending? i.e. – over the counter medications at a CVS, urgent care clinic visits, out of pocket payments at a routine doctor’s visit, etc? These might help predict certain behaviors or issues but it is likely too late to actually do anything. This dataset would also be largely incomplete, as many of these purchases are not made by consumers but are later billed to insurance companies.
    Are they tracking personal spending? i.e. fast-food purchases, sports equipment purchases, travel-related purchases, etc? While this information might certainly help predict patient health or accidents, the link will only be correlative and not causative. It is easy to say that a patient’s heart disease may have been exacerbated by eating McDonald’s every week, but it is not a causal link to say that every patient who eats at McDonald’s will acquire heart disease. Similarly, what can Carolinas Health do about patients with high-risk behavior other than encourage them to come in for preventive care visits or enroll them in population health management programs?
    Would love to hear your thoughts and clarifications about what data they plan to collect and how they plan to use it!

    Thanks!

  2. Great post! Many large academic institutions are flirting with this idea and also believe that this data can help steer their preventative care efforts. Privacy is definitely a concern, but as consumers get more and more used to having their data collected and analyzed, this may actually become a reality sooner rather than later. And users may be accepting of this more when someone who they trust, ie their doctor, is the one using the data to make recommendations – which is far better thanparties that use this information purely for advertising.

  3. This. Is. Awesome. This is super relevant data that I would think predicts health risks more than many other data sets healthcare companies are currently analyzing. I wonder about public backlash, but at the end of the day the medical community has been looking for a way to travel consumer behavior and lifestyle choices vs. just genetic disposition and environmental problems. I do wonder how this translates to me buying things to others (e.g. for kids, or elderly parents)

  4. This is a really interesting use and unique way of using consumer purchase data. Aside from the privacy risks mentioned, I think this has huge potential for people who are unlikely to go to the doctor as frequently as they should for check-ups etc. Lots of people dread going and believe they only need to go when they are sick, but there are lots of benefits of catching things early – including future expenses. I think this technology could be very useful for this cohort of people who aren’t going to the doctors office. One concern could be that consumers who currently do see their doctor view a tool like this as an alternative. It will be really interesting to see how this gets implemented!

Leave a comment