agsandrew - Fotolia
Granite Healthcare Network in Concord, N.H., found what appeared to be an obvious variant in claims data. However, that variation turned out to be false, said Bob Kay, director of population health at Granite Healthcare Network.
What solved the puzzle and brought the truth to light was clinical big data, Kay said. "The context of those claims is very important."
In Granite Healthcare's case, this false variation was due to differences in how claims are paid and how hospitals process claims. Kay explained that a health system charged for a service under an incorrect category which "artificially drove up the rate and the expense for that service category."
Ultimately, Kay said Granite Healthcare figured out that this was indeed a false variation after having their health systems provide their clinical data as well as explain their billing practice so that the pieces could be put together.
"It's probably a universal issue with claims data because claims data is really just used to pay bills, and sometimes what you see in claims data is the artifact of how those bills get paid rather than true variation," Kay said.
Unfortunately, Granite Healthcare doesn't have claims data merged with clinical data yet -- and they are not alone in this big data and data analytics predicament.
Joel Vengco, vice president and CIO at Baystate Health based in Springfield, Mass., said during a panel at the recent Health IT Summit in Cambridge, Mass. that his organization ran into a problem similar to Granite Healthcare's with its diabetes population. By analyzing only claims data, analysts missed about 35% of people in the health system's overall population who should have been in the diabetes programs. Vengco said analysts discovered this disparity after they had integrated clinical data in with their claims data.
Many healthcare professionals acknowledge the importance of data analytics and big data in healthcare, but due to lack of interoperability and the healthcare community not yet grasping how to derive value from big data, many are unable to take full advantage of the benefits, like precision medicine, population health management and value-based care.
"The ship has left the dock. We're all going to have to make this journey," said Deane Morrison, RPh, CIO at Capital Region Healthcare and Concord (N.H.) Hospital, during the panel.
Before a hospital jumps up and implements analytics, Vengco urged the audience to remember that the data itself is key. "You've really got to work on the data and make sure that you've transformed it, you've localized it, you've standardized it, you understand what the knowledge management and the master data management components are for that data so you can evolve it over time," he said. "The analytics piece will come. But if you don't have the right data you're never going to have the right analytics."
Morrison also advised that "when you [go] through and you look at all the inferences you … really have to challenge yourself [and] say, 'Can I really draw this conclusion from the data or is the data misleading me?' And that's our challenge."
Granite Healthcare's solution to the problem of disjointed claims and clinical data was to get help piecing the two sets together from athenahealth Inc. in Watertown, Mass., a cloud-based electronic health records company that also sells medical billing services.
Competition and siloed data barriers to interoperability
One challenge many healthcare professionals face when it comes to big data in healthcare is the unwillingness to share data. The experts on the panel attributed this barrier to the changing roles of healthcare providers as well as to vendors working hard to obtain and then keep customers.
Kay used Granite Healthcare's experience with insurance carrier Cigna as an example.
"We have a relationship with Cigna, we're a part of their collaborative accountable care program, which is an ACO-like arrangement, and we have a shared savings component to that. So we're responsible for managing the care of all the patients attributed to us," Kay said. Granite Healthcare asked Cigna for the claims data report so that Granite Health could do its own analytics. "We rely on the reports they give us but we also go a step further and use that claims data ourselves to look for variation and ways to improve care."
Cigna was hesitant to hand over that report because it included financial data, which is critical to the insurance company's business needs, Kay explained. Ultimately, Cigna agreed to share its report after Granite Healthcare put restrictions on the data so that only two people in the company could see it, Kay said.
Granite Healthcare also started a health insurance company in partnership with Tufts Health Plan that will be available in 2016.
"Roles are changing very quickly," Kay said. "It used to be: 'We're the provider' [and] 'We're the payer', but those roles are really becoming meshed together, and I think everybody's just trying to figure out where they fit in."
Vendor data offers added hurdles
Vendors, on the other hand, present a different situation, said Chris Blessington, senior director of marketing at ExtraHop, a company in Seattle, Wash., that develops wire data analytics platforms. "Vendors in the space have traditionally taken those siloed data sets that you work with and siloed them even further, making that data proprietary and difficult to get out of [the vendor's] tools."
To Vengco, that's the wrong way to go about generating revenue, he said. "I think that's the completely wrong sort of market focus and … interoperability should be free."
And although the 21st Century Cures Act draft -- introduced to the U.S. House of Representatives in May with the goal of promoting the development approval of new drugs and devices -- has strong language encouraging vendors to share their data, Vengco said it isn't enough. "[It's up to] us to … really continue to put pressure on our vendors to open up that data."
At the end of the day, Jeff Harner, chief analytics officer at Rise Health in Oak Brook, Ill., said, the debate over sharing data in healthcare comes down to patient care. "It's the patient in the middle that's the really only important thing here," Harner said. "People who are siloing that [data] off are doing a disservice to the patient."
Learn more about big data in healthcare:
Attract the best talent to make healthcare data analytics happen
Maine HIE takes next step with data analytics
Vast data doesn't have to be a barrier to population-scale genomics