Analytics means we ‘roll up our sleeves’
July 29, 2013 in Medical Technology
UPMC has invested more than $1.6 billion in its IT infrastructure over the past five years, according to Pamela Peele, chief analytics officer of UPMC Health Plan. That’s more money, she points out, than its home city has spent on three pro sports stadiums combined – “and we take sports seriously in Pittsburgh.”
Those massive investments have paid big dividends, said Peele, speaking July 24 at the The Institute for Health Technology Transformation’s Denver Health IT Summit, and showed how UPMC’s strategies could be useful even for smaller organizations without that sort of financial muscle.
One of the advantages of being an integrated delivery system is that the physicians, patients and payers exist in what’s essentially a “natural laboratory” where different types of data can be mined about what works and what doesn’t, she said.
Robust analytics tools are essential to uncovering inefficiencies and pointing the way toward best practices. But many organizations miss the big-picture when implementing the technology, said Peele: “Analysis is the generation of new knowledge. “It’s not reporting.”
But a crucial first step, of course, is getting data that’s uniform and usable – “fit for consumption,” as she put it. Alas, “the sad part of it is, there’s nothing sexy or fun about making data fit for consumption.”
Whether it’s claims data, prescription information, lab results or some combination thereof, aggregating and normalizing data is key, she said. “You can’t manage what you can’t see.”
But “no vendor tool is going to do this for you,” she added. The way towards “a single source of truth,” is to “roll up our sleeves and make it happen.”
The future is promising, as healthcare organizations are increasingly “using data in a way it was never intended to be used,” said Peele. “Lab data wasn’t meant for pop health management.”
UPMC has gained valuable knowledge from its self-developed care management platform, and has been able to make big strides in predictive analytics. “When someone shows up to the ED we predict the risk that they will be readmitted within 30 days.”
The nature of those doing the analyzing has changed, too. Back several years ago, analysts tended to be business-minded bean counters, unable even to arrive at a consistent number identifying the number of diabetics in a given Pennsylvania county.
Now, UPMC’s analytics personnel each have different tasks, from clinical evaluation to strategic business analysis to database quality to modeling.