UPMC Leans On Business Intelligence For Accountable Care
The University of Pittsburgh Medical Center is using natural language processing and interoperability to achieve accountable care.
Accountable care organizations (ACOs) are all the rage in healthcare as providers look to focus more on quality than on volume of services delivered. But Dr. Robert Margolis, CEO of HealthCare Partners Medical Group, Torrance, Calif., has likened ACOs to unicorns because there has never been a verified sighting of one.
Clinical leaders at the University of Pittsburgh Medical Center (UPMC), a $10 billion enterprise that includes 20 hospitals, 400 outpatient treatment sites, and a health plan, believe that the organization has a technology-driven roadmap in place to achieve accountable care.
More Healthcare Insights
- How Healthcare Payers are using Customer Communications to Improve Productivity and Effectiveness
- Learn how Kettering Health Network maximized clinician patient time by virtualizing clinician access to data
White PapersMore >>
"This is really ground zero of where all of the wishes of an accountable care organization translate into reality," Dr. Rasu Shrestha, vice president for medical information technology, told InformationWeek at the Center for Connected Medicine this week. The center, on the 60th floor of Pittsburgh's U.S. Steel Tower, is a showcase for many cutting-edge technologies and processes already in use at UPMC. (The organization made the InformationWeek 500 list for 2010.)
[Is it time to re-engineer your Clinical Decision Support system? See 10 Innovative Clinical Decision Support Programs.]
"I think ACOs, like a lot of other things in health IT, have now become the buzzword," Shrestha said. "But if you look at the core impetus behind an ACO, an accountable care organization, we like to think that we have a lot of those components at hand within UPMC."
Shrestha, who also is medical director of interoperability and imaging informatics and chief of radiology informatics for the integrated delivery system, said that the idea is to take a patient-centric approach to delivering healthcare services. "It's really about transforming the way that we practice medicine from volume-based to value-based," he explained.
The effort will rely heavily on business intelligence and interoperability. "A connected environment is absolutely critical," said Dr. Andrew Watson, medical director of the Center for Connected Medicine and a heavy user of telemedicine for consultations and postoperative care in his surgical practice. "The heart of the connected environment is the data center."
UPMC, already in a $402 million partnership with IBM that soon will incorporate the "Watson" supercomputer platform, recently added Nuance Communications as a partner in the Center for Connected Medicine. Nuance, which recently sponsored a media day at the center, is touting cloud-based "clinical language understanding," or CLU, its name for natural language processing of unstructured text.
"It's really important for us to leverage the power of CLU to index the documents and other unstructured blobs of information, and that's across our multiple different silos of repositories, such that we are able to get at meaningful and actionable data and then bring this back in a closed-loop manner to the clinical environment," Shrestha said. He talked about how healthcare is moving from mere information to knowledge. "It's really about bringing the data to life," Shrestha said.
"We like to think that we have multiple components to contribute to an ACO and we are steadfastly moving toward that direction right now. An initiative such as what we're doing with CLU--getting more data and facts extracted and making that more meaningful and bringing it into the clinical work environment-- those will definitely contribute to what we're striving for."
Early this year, UPMC began aggregating data from both the provider and payer sides of the organization and applying analytics to that information to assure accuracy. This can be tricky because billing records often do not match clinical records.
Shrestha said UPMC has multiple checks and balances in place to assure data accuracy. "Data quality is one thing that drives us across the board in any initiative at UPMC, but especially around more of a groundbreaking, innovative thing like this payer-provider aggregation," he said.
The claims data can help fill in some of the gaps in patient records, such as for care delivered outside the UPMC provider network or whether prescriptions actually got picked up from the pharmacy. "As a clinician, say, in the ED, I'm interacting with this patient who's come in with acute abdominal pain. I'm able to look up all of the medications that this patient is on across the board, across all of the hospitals from UPMC, in an aggregated, semantically harmonized manner. But in addition to that, I'm also able to look at prescriptions that have been filled, claims-based information that's coming in from the payer side," Shrestha said.
This kind of knowledge is useful for medication reconciliation, something that likely will be part of Meaningful Use standards for electronic health records (EHRs) in the future. It also can help identify "frequent fliers" who go from hospital to hospital in search of multiple prescriptions to drugs they are hooked on. "If this patient has been doctor shopping, going from one ED to another, including non-UPMC facilities, we actually can see that and we can treat the condition as opposed to [feeding] the patient's addiction," Shrestha explained.
Healthcare providers must collect all sorts of performance data to meet emerging standards. The new Pay For Performance issue of InformationWeek Healthcare delves into the huge task ahead. Also in this issue: Why personal health records have flopped. (Free registration required.)