Walmart is an early example of a company that took its financial transaction data – what people were buying in its stores plus what it was purchasing in supplies – and realizing how to use it for operational, marketing, and supply chain advantage. They were so effective at it that entire industries began to copy them. Now, consumer marketing companies aggregate purchasing data and sell it to retailers all the time. Try to name one large consumer company today that doesn’t do it as part of their core strategy.
Health insurance companies with claims data go beyond what Walmart, Amazon, Google, and other companies like them have in transaction data. Health insurance companies have complete financial transaction data on every patient interaction that is insured with clinical dimensions added on to every claim. Think about that: we have complete, clinically-annotated transaction data on people. The complete history (while they are enrolled) of a person’s care and the trajectory of care over time is in the claims data alongside financial and supply chain data. As we move to value-based care, how could that not be more important than ever?
Sure, there are legitimate and serious concerns about the data. Providers use it for reimbursement accuracy, not clinical accuracy. The data is lagging, arriving too late to drive real-time interactions. It’s part of fee-for-service and we need to move to a value-based approach where we don’t track every little thing we do that drives a burdensome administrative system and is exhausting our clinicians and their support staff. These are all true. Yet, it is still the best source of system-wide data out there.
The only alternative to it is clinical data, captured in electronic health records (EHRs). And many argue clinical data is better – more accurate and richer in clinical detail. I agree. It is better, in theory, but not in practice. There are serious issues with clinical data today when we consider its utility as a system-level data source. The explosive growth in EHRs was due to the billions in federal incentive dollars, not because of a business model or a business transaction where the quality of the data captured mattered. You could literally have built a digital filing cabinet that does minimal productive work and received federal incentive dollars. Trying to extract meaningful clinical value out of that is challenging, to put it mildly.
In a head-to-head comparison, I think claims data handily wins today: relative to clinical data, it is more standardized, complete, well-defined, integrated, and accurate. When we are trying to derive enterprise-wide insights and action, these qualities of a data system matter a lot. And while claims may not be quite as timely, un-adjudicated claims are not too bad (most claims flow through transaction systems fast, it’s the big ones that are held up, or those coming from out of network), and it doesn’t mean we can’t add in more timely data as a supplemental data signal. And of course, we should absolutely be investing in clinical data, because yes, it will be better over time than claims in its clinical richness and accuracy. At the same time, we should think very carefully before we set off on a path where we lose a valuable data source (claims) before the alternative data source (clinical) becomes mature enough to replace it.
Jay Chyung, MD, PhD, serves as Chief Operating Officer and Co-Founder of 10xHealth. 10xHealth is an intelligence technology company serving healthcare markets with cloud-based business intelligence solutions.
