A Network of Networks

One of the key efforts of the Office of the National Coordinator for Heath Information Technology (ONCHIT) is to move forward on a set of contracts — made in late 2005 for $18.6 million — to build prototypes of a national health information network (NHIN).

The four prototypes are supposed to be part of the seeding activities that will eventually develop into a network of networks, to include geographical networks and "non-geographical networks," such as the current connections for pharmacies.

In that vision, the NHIN will eventually become, "the glue that brings maybe several hundred of those networks together in an interoperable fashion," said John Glaser, CIO of Partners HealthCare System in Boston, who spoke about the NHIN future at a recent American Health Information Community (AHIC) meeting.

But for now, work on the four prototypes reached a landmark early this year as each of the four contracted consortia displayed the architectures they have constructed to AHIC and at a special forum on the prototypes.

The groups demonstrated, for example, how with their products patients could pull in discharge summaries, medication histories, home monitoring device readings or other information into a personal health record. In addition, patients could add providers to allow them to view information, and shield specific information from particular providers or from all providers.

In other examples, the prototypes illustrated how an "invited" clinician could view patients' PHRs before the individuals are seen in an office or hospital and how providers, through their electronic medical record system, could search for information on a patient through the network.

One of the key differences in the prototypes, according to consortia representatives, is in where data are stored and if they are in a central location or if they are available only by query.

John Loonsk, M.D., director of the Office of Interoperability and Standards within the federal HIT effort, noted that last year the prototype consortia devoted about 80 percent of their effort to the underlying architecture of the networks and thus, the software demonstrated reflected only about 20 percent of the work to date. Over previous months, Loonsk said, "We have expected them to do functional requirements, security models, business models... and a variety of other deliverables."

• The Accenture (Hamilton, Bermuda) group works in Kentucky, Tennessee and West Virginia.

• The Computer Sciences Corporation (CSC, El Segundo, Calif.) consortium is working in parts of Indiana, Massachusetts and California.

• The IBM-led group (Armonk, N.Y.) is working in areas of New York and North Carolina.

• The Northrop Grumman (Los Angeles, Calif.) consortium is working in parts of California and Ohio.

The next phase of the prototypes' work, underway now, is to connect the prototypes and state and regional health information exchanges in trial implementations.

Meantime, in discussion at the AHIC meeting on the long-term vision for the national network of networks, Glaser said different parts of the connected networks will provide different services. But throughout, there should be certain core capabilities including exchanging records, providing necessary security and locating records wherever they are geographically.

AHIC also heard assessments from economists indicating a national network may be able to sustain itself through its own business transactions in under eight years. That assessment was true for network models that would have significant central agencies or functions, and those with very little central function.

Glaser warned, however, that financial success would be dependent on a large number of conditions, including a sufficient base of EHR adoption and broad adoption of the accepted standards. He also stressed that those successful models foresaw a critical part of the revenue, up to half in some cases, coming from secondary uses of the data, such as for public health purposes, research and other analyses. He emphasized that currently data through such networks are often not good enough for those purposes.