By Adam Stone
Feb. 12, 2008 | On December 15, 2007, following two years of preparation, the University of Pittsburgh Medical Center went live with a new enterprise master person index (EMPI) from Initiate Systems. The goal is to boost efficiency and improve patient care by stripping out hundreds of thousands of duplicate records from throughout its many data systems.
Part of the challenge was coping with UPMC’s daunting size. It serves more than 4 million patients each year in 29 western Pennsylvania counties and has 45,000 employees, 20 hospitals, 400 physician offices/outpatient centers, and a health plan. In a typical year, UPMC has more than 167,000 inpatient admissions, 3 million outpatient visits and 400,000 emergency visits annually.
The IT environment is also large and complex, with over 1,100 IT professionals supporting more than 2,000 applications. Since 2000, UPMC’s legacy EMPI had proven its value and benefited patients and UPMC.
As described by project manager Karen Kohl and client services manager Julie Prough, deploying the new EMPI was an ambitious effort, requiring a tricky on-the-fly handoff from the legacy system. The new EMPI must tap nine different source systems of patient data, seven source systems of provider data, and one source of organization information. It also must function in lock step with a recently implemented interoperability platform from dbMotion.
The legacy vendor, however, had not fully committed itself in the EMPI space, says Prough. The legacy system couldn’t effectively interface with ancillary systems, scalability was a problem and the vendor’s attention was focused on other product lines.
Two years ago Prough and Kohl began reviewing options and expanded their effort to include a VP of technology platforms and an operational VP. Their RFP, which included demanding technical criteria for a robust algorithm to identify duplicate records and for system scalability, drew four bids.
Initiate’s singular focus on EMPI impressed the UPMC team. “This particular vendor is basically an EMPI vendor,” Prough says. “This is not part of a larger suite of products. Their full focus and the full effort of their developments really is the EMPI.”
In the brief time the new system has been running, it has already identified about 460,000 duplicate records dispersed among the hospital’s many data systems. Once the duplicates are reconciled and removed, that number likely will drop to 160,000 legitimate records, Kohl says.
With 10 staffers reconciling records, in addition to doing their other work, Kohl anticipates it will take about a year to get the job done. The strategy is to clean up duplicates among the newest records first, as these patients likely are still active in the medical system, and then go after the older material.
The ability to prioritize de-duping runs deep into the EMPI system. Consider two common problems in de-duping: a patient is identified in two different ways (a misspelled name, perhaps); one patient’s information is put into another patient’s file.
The latter is clearly more urgent. The Initiate EMPI algorithm can differentiate between these issue types, and prioritize which to tackle first. Kohl calls that a significant capability.
Kohl says Initiate also delivers the ancillary connections missing in the legacy system. Radiology, lab systems, and other areas now can query the EMPI, which helps keep patient data accurate and consolidated.
To put the new EMPI in place, IT leaders opted for a “rip and replace” approach, flipping the entire system at once rather than phasing it in. The decision rested largely on the sheer complexity of UPMC’s information systems. “We could not sustain two different EMPIs at the same time. There were too many things that were intertwined, too many processes in there that all would need to get their data from the same source,” Kohl said.
To get it right on the first try the team started small, running simulations at the unit level for two to three weeks. The next round of testing increased the volume of records, and eventually the team did two go-live tests at full production volume.
In the past, UPMC’s EMPI ran in the background largely unnoticed by users. Given the new system’s expanded capabilities, a steering committee formed in January to explore business uses that could potentially reach into all aspects of the organization’s information management apparatus.
“There is a lot of excitement about the ability of this tool,” Prough said. “Now people at the executive levels are really asking: How can we best leverage [it]?”