Semantic interoperability approach allows multiple HIT systems to understand and make use of incoming data.
Necessity, as the old adage goes, is the mother of all invention - or, in the University of Pittsburgh Medical Center (UPMC)'s case, innovation.
Explosive growth - and an abiding belief that health information technology (IT) plays a vital role in the delivery of quality care - prompted UPMC to seek an avenue for data-sharing across its massive enterprise.
Leadership recognized this would be no trivial task. With 20 hospitals, nearly 500 physician practices, a health plan and a plethora of community services, establishing an infrastructure to satisfactorily meet all user needs represented a daunting challenge. Nonetheless, as a recognized industry thought leader, UPMC welcomed the opportunity to be at the forefront of the emerging interoperability movement.
Over the past eight years, UPMC, where I serve as senior vice president and CIO, "test drove" several approaches that offered the possibility of creating the intricate health information exchange it sought.
Ultimately, UPMC selected an interoperability strategy that allows any information system to access the patient data it needs and present it to caregivers in the format they prefer.
A monolithic approach
Committed to providing integrated health services and seamless information exchange to improve quality, UPMC embraced a technology strategy based on standardization and interoperability. Clinical and IT leadership initially chose to pursue a monolithic approach - based on a single-vendor platform - which promised the delivery of an IT system that would be highly integrated and easy to manage. Although UPMC originally tapped one vendor to provide all ambulatory systems, it subsequently selected another as the enterprise clinical supplier.
UPMC dedicated efforts to rolling out the ambulatory electronic medical record (EMR) to 90-plus practices and 2,200 clinical users between 2000 and 2005, with marginal success. UPMC found that users were reluctant to abandon their best-of-breed systems with their specialized and familiar functionality in favor of a universal system - regardless of its purported future benefits. Sluggish physician adoption hindered the effectiveness of ongoing IT implementation. In addition, even within the realm of a single-vendor clinical strategy, different products - which did not necessarily "mix" well together - were needed to fulfill different roles.
Finally, due to the complexity and size of the UPMC enterprise, the single-source vendor was forced to develop three discrete "zones" to enable the organization to effectively retain and provide timely access to patient information. This, in essence, created a "data silo," which added barriers UPMC had hoped to avoid (in addition to eliminating isolated data sets already existing within disparate legacy systems).
Recognizing the limitations of a monolithic model, UPMC leadership stepped back in 2005 to reassess its strategy. It discovered that many anticipated benefits remained unrealized. Costs, for instance, had not been reduced, while the need for user interfaces had increased. And centralized configuration could not be achieved as long as separate zones were in place.
In the end, leadership concluded that investing in multiple ambulatory EMR products would sustain clinical adoption, increase utilization and, therefore, be more economical in the long run. It also determined that, even in a diverse IT environment, clinicians required context-specific data (medications, allergies, immunizations, problem lists, etc.) at the point of care, regardless of the systems they or their colleagues were using.
Semantic interoperability
At this stage, leaders engaged the Carnegie Mellon University Software Engineering Institute (SEI) to perform an analysis for developing an interoperability strategy. By May 2006, SEI rendered its opinion to the UPMC board of directors: that in building such an integrated system, interoperability was not only viable, but necessary.
SEI recommended a semantic interoperability approach, which allows multiple HIT systems not only to share information, but enable the systems to truly understand and make use of incoming data. At the same time, semantic interoperability preserves the original "meaning" of that data, even if the data has been created and stored in varying systems and formats. SEI also emphasized that, as it pursued its interoperability initiative, UPMC had to address diverse stakeholder requirements in addition to technology considerations, and involve IT, clinical and business/operations staff during all stages of development.
Significantly, the board and executive leadership threw its full support behind the newly christened Interoperability Project, formally funding the initiative and staffing it with a team of clinical and technology professionals.
The search for an interoperability partner that would provide the backbone to an integrated system led UPMC to evaluate large, nationally recognized companies, as well as smaller, niche vendors. UPMC selected dbMotion because it offered an integrated solution that had been in production since 2001, a sophisticated data model, a scalable platform built on open standards, a sound application development framework and deep domain experience. The solution was also constructed around service-oriented architecture (SOA), which enables systems to consume data from each other through functionality that is agnostic to originating system architecture, format and structure. This proved vital, since the multiplicity of EMRs employed at UPMC had their own structure, developmental tools and underlying logic.
Project progression
Because the overarching goal was to enhance patient care and safety, the first phase of the project was to construct the interoperability platform to deliver essential data sets (allergies, medications, problem lists, lab values and demographic information) to all providers at the point of care and across the health care enterprise. To prepare, the Interoperability Project team spent much of 2007 building and loading test records to validate the system.
A pilot group - comprised of UPMC physicians and other clinicians in two emergency departments (EDs), two primary care ambulatory practices and a medical clinic - was created to test whether providers could gain access to a single view of the essential data set via the interoperability platform. At the same time, leadership wanted to ensure that the various clinical systems in use were able to populate the interoperability solution and, in turn, that the solution could handle the volume of transactions that would be required of it.
With these objectives accomplished, the solution was rolled out across the enterprise in February 2008 with about 35 million clinical records on 4.5 million patients. The records had been integrated and mapped into a semantically coherent schema.
Almost immediately, UPMC confirmed the clinical value of the semantically interoperable platform. Physicians in both emergency and outpatient settings relayed incidents where the immediate availability of patient information had a significant and beneficial impact on care.
With these major accomplishments in the "win" column, UPMC has witnessed a dramatic uptick in the number of clinical staff accessing data through the interoperability function. UPMC recorded more than 2,000 users within six months of the system going live in February 2008, with daily increases driven by popular demand.
The UPMC Interoperability Project is delivering upon virtually all of its potential. Reports from across the enterprise indicate that the availability of information - because systems are now interoperable - has had a significant, measurable, positive impact on the quality of care providers are able to deliver.
Mr. Drawbaugh is senior vice president and CIO at the University of Pittsburgh Medical Center (UPMC). Under his leadership, UPMC has pursued an IT strategy that seeks to maximize patient safety and the provision of quality health care.
This article can be found on-line at Advance.