Issue No.07 - July (2006 vol.7)
Published by the IEEE Computer Society
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/MDSO.2006.43
IBM and about 20 organizations, including the World Health Organization and the US Centers for Disease Control and Prevention, announced the Global Pandemic Initiative in May 2006. IBM pledged to contribute several pieces of its health care technology portfolio to the open source community to help launch the initiative's work. However, progress toward electronic health records and networks that could support population-based applications is moving slowly among national government agencies, especially in the US.
Public policy mavens have long believed that comprehensive, population-based, electronic health records and networks could improve healthcare and lower its day-to-day costs. However, existing health networks are extremely complex and often operate in silos that exclude even disparate offices of the same organization from communicating with each other. They also lack clear economic incentives that might encourage private investment.
In May, however, IBM and 20 organizations, including the World Health Organization and the US Centers for Disease Control and Prevention, announced the Global Pandemic Initiative ( http://www-03.ibm.com/press/us/en/pressrelease/19640.wss). The initiative responds to concerns over the recent avian flu outbreaks around the world. IBM pledged to contribute several pieces of its healthcare technology portfolio to the open source community. This offer is intended to speed the flow of information about virulent disease outbreaks. Some technology marketplace observers wondered if the GPI announcement might do for health information networks what the company's support for Linux did for open source software in general and what its release of the Eclipse platform did for development tools.
Richard Waldhorn, a distinguished scholar in the Center for Biosecurity at the University of Pittsburgh Medical Center, served on the GPI's advisory board. He says his participation on the board was at least partially the result of the IBM action: "If a big global company thinks this is important and they are reaching out to experts in the field, they will be calling attention to these issues."
However, Waldhorn also sees a dilemma facing those who might like to see immediate momentum gains for health networks. "In some ways, it's the cart before the horse," he says. "I don't think I can give you an informed opinion on what's the right strategy to develop electronic health records, but drawing on my experience after 28 years as a clinician, if we had such a system, both clinical medicine and public health—and what you might call the strategic intelligence you need for gathering public health information—would be greatly enhanced."
False starts, promising intentions
Certain events of this past May can serve as a microcosm of the current state of interoperable health networks. Early in the month, the officials in charge of developing national electronic health networks in the US, Canada, the United Kingdom, and Australia met in San Francisco. They pledged to begin working more closely, notably in adopting IT healthcare standards. In the middle of the month, IBM and its partners announced the GPI and the release of IBM technology to the open source community. Four days later, David Brailer, the US National Coordinator for Health Information Technology, officially left his position (he announced his resignation in late April).
While the Bush administration has emphasized its desire for a national health information network, its actions to establish one have been limited. They consist mainly of discussions about harmonizing standards among technology and healthcare organizations and establishing pilot programs around four regional health information organizations. The sputtering effort has left some of the US's putative partners wondering what direction the Office of the National Coordinator for Health Information Technology ( http://www.hhs.gov.healthit) will take and how effective it will be. ONCHIT's acting director didn't return a call requesting an interview.
"What is really the mandate of ONCHIT?" asks Dennis Giokas, chief technology officer of Canada Health Infoway ( http://www.infoway-inforoute.ca/en/home/home.aspx), the nation's clearinghouse organization for establishing nationwide interoperable electronic health records. "What are they going to do in the context of this overall mandate? Are they just a coordinator and thought leader?"
Giokas says the Canadian federal and provincial government leaders have empowered Infoway to establish firm guidelines for Canadian healthcare organizations as they invest in health networks. "We have the money to force, if you will—I know that's a strong word—but we have the money to mandate that our partners align with certain things we do in our nine strategic programs." For example, he says, partners must align with Infoway's architecture and standards, he says, to qualify for Infoway's coinvestment, which averages 75 percent of planning and implementation costs for approved projects.
Giokas says the US approach doesn't include that carrot-stick approach and is significantly behind the Canadian effort. "The money they do have is for stuff we did three or four years ago, and a lot of that is grant money to do pilots. We don't believe in pilots. Some projects we invest in might be less than successful and some might fail, but we chose to invest in our vision, not in pilots to test the concepts with respect to our mission."
Giokas says establishing the Canadian network effort might represent a level of difficulty somewhere between the Connecting for Health ( http://www.connectingforhealth.nhs.uk) effort of the National Health Service in England and ONCHIT in the US. While England's NHS effort is made somewhat easier by elements such as an existing patient identifier system and one overarching agency establishing the technology, Giokas says Infoway officials need to convince 14 provincial deputy ministers of health of the benefits of the organization's vision.
The US, on the other hand, is a vast tapestry of federal, state, and county public-sector healthcare providers and payers, plus private-sector insurers, hospitals, and physicians, often working at cross-purposes with respect to establishing a national network.
VistA: An incomplete vision
If the US government has failed to lead the way in establishing an effective healthcare network, it has developed what might be the "poster child" application for what a public domain, open source development model can accomplish—and what can go wrong when the application's utility is allowed to splinter.
The application is the Veterans Health Information System and Technology Architecture, or ( http://www.va.gov/vdlVistA), which the US Department of Veterans Affairs developed for its Veterans' Administration (VA) medical centers and providers. Recent events surrounding VistA have showcased the possibilities and the limitations of an incomplete vision for a healthcare network and its architecture components.
As an example of its utility for both clinical and public health deployments, VistA records for VA patients are available at any VA facility. In the aftermath of Hurricane Katrina, displaced VA patients from the Gulf Coast were able to receive services from physicians and pharmacists at other VA hospitals with no interruptions, as their records were already in the VA's database.
As a piece of software developed in the public domain and freely available under the Freedom of Information Act, VistA could offer significant savings over proprietary electronic health records (EHRs). Midland Memorial Hospital in Midland, Texas, is deploying a version of VistA developed by Aliso Viejo, Calif.-based Medsphere, as its first-ever comprehensive EHR system. The hospital reports that the Medsphere EHR will cost about US$7 million. This is less than half the $15 to $18 million cost estimated for proprietary EHRs. Medsphere cofounder Scott Shreeve says the company isn't trying to compete with the established EHR vendors. Rather, it's targeting mid-tier hospitals that lack the resources to buy a functional alternative.
However, VistA isn't released under any reciprocal open source license. This has led to development forks, including private efforts such as Medsphere's as well as public nonprofit efforts such as a version developed by Pacific Telehealth Hui. There's no assurance that code changes in one version will work with another and no development model to ensure that beneficial additions to one version will be automatically included in another.
Ken Kizer, who oversaw VistA development while he was undersecretary of health for veterans affairs during the Clinton administration, is currently CEO of Medsphere. He addressed the splintering problem in testimony before the House Ways and Means health subcommittee ( http://waysandmeans.house.gov/hearings.asp?formmode=view&id=4828). Kizer called for Congress to dedicate 5 percent of the VA's VistA development funds toward developing a public-private partnership open source VistA development environment. He also asked the federal government to consider making open source the default choice in software procurement. However, the relevant bill moved out of committee without including Kizer's recommendations.
Progress being made in lower levels
The next step in expanding the interoperable and open source mindset among healthcare developers might not come from patient-facing applications such as VistA. Instead, some healthcare IT experts see it in projects such as the GPI, the new Eclipse Open Healthcare Framework project ( http://www.eclipse.org/ohf/), and the Open Group's Universal Data Element Framework ( http://www.opengroup.org/projects/udef). Chris Harding, who heads the Open Group's UDEF working group on semantic interoperability, describes UDEF as a Dewey decimal-like classification system for data semantics. He thinks it might allow even healthcare organizations with disparate top-level architectures to share metadata: "Because they have a common indexing method," he says, "they'll know what each other is talking about."
UDEF is based on the ISO 11179 standard for metadata registries. Harding says it's also intended to integrate Semantic Web technology, therefore serving as a sort of bridge between the two. Among the organizations interested in testing the UDEF for its potential in improving interoperability is the US National Cancer Institute.
Some experts see the technological challenges of these ambitious projects far outweighed by cultural skittishness among healthcare executives and providers. Obviously, a pandemic would give momentum to addressing both the technological and bureaucratic challenges. However, Waldhorn sees a source for change other than a crisis.
"There is some hope when you talk to interns and residents," he says. They grew up with this stuff and they can't imagine any other way to do business. They demand it."