Without significant support from the government and the private sector alike, the creation of Internet-based clinical-information networks—a longtime dream of many health care experts—faces serious challenges. McKinsey research shows that to build and operate such networks profitably, large health care institutions would have to spend $2.2 million annually. Only the most financially sound hospitals could afford such expenditures.¹

Over the past ten years, several high-profile yet unsuccessful efforts have attempted to use medical technology to link groups of patients, physicians, hospitals, and other providers. These networks, also known as community clinical-data exchanges, are intended to facilitate the sharing of digital medical records, laboratory results, prescriptions, and insurance information (Exhibit 1). They have been inspired by the belief that up-to-date, clear, and easily transmittable information—accessible to doctors and patients no matter what their location—would cut the number of medical errors, improve the quality of clinical care through the more consistent practice of evidence-based medicine, and save money by making processes more efficient.

Potential members of such networks have been discouraged, however, by their expense and by the disincentives facing first movers. Indeed, only about 13 percent of US hospitals and a small number of doctors' offices have pursued this idea.² The slow pace of development has frustrated its advocates in government, including the Bush administration, which in 2004 appointed Dr. David Brailer as the nation's first health information technology coordinator. Community clinical-data exchanges are critical stepping-stones toward the creation of a national network that could save $140 billion a year on health care costs and help the federal government track trends, such as epidemics, more successfully.

Metropolitan and regional health care communities have thus far lacked sufficient data to justify the cost of building and operating health information networks. Our cost-benefit analysis demonstrates that even larger communities must overcome sizable financial and organizational hurdles, such as achieving a critical mass in numbers of participants, to gain financial benefits that would make the investment worthwhile.

Our analysis focused on quantifiable costs and benefits. Savings from work flow efficiencies and the concomitant reduction in the number of full-time equivalents³ were included, for example, but anticipated though as yet unproven reductions in clinical-practice pattern variations⁴ were not. We divided the networks' benefits intotwo categories. The intrinsic benefits are those a health care institution derives from posting its data on the Internet. The network benefits are the incremental advantages it gains when it receives relevant information from external sources. The costs of setting up and operating a system include network and site-specific hardware, software, and personnel. By analyzing a series of hypothetical information exchanges, we found that when participation is low their net value is modest in large and midsize communities and negative in small ones. Substantial returns become apparent only as the size of the exchange grows and participation rates increase. The aggregate costs for a large community's network were $2.2 million a year, with benefits of $7.9 million (Exhibit 2).

Although our study found that large networks could reap a maximum annual net benefit of $5.7 million, first movers are at a distinct disadvantage. The network initiator would likely bear a significant portion ($290,000 annually) of the exchange's costs and derive too little reward to get its money's worth unless a full slate of constituents were on board from the start. While physicians have the most to gain from a network ($2,400 for a $40 investment per physician), they are also the least likely to participate, given their historical reluctance to pay for information technology and to share data. The most likely network organizers are hospitals, which would gain substantially from the streamlined transfer of laboratory and radiology data.

Given the first-mover disadvantage, however, we believe that hospitals would not form data exchanges without support from the government and private foundations, which, unlike many hospitals, can afford to finance the development of health care information networks. Many states are already supporting such efforts.⁵ The California HealthCare Foundation and health care leaders in Santa Barbara County, California, are developing an information network in cooperation with CareScience, which specializes in building these systems. The Santa Barbara County Care Data Exchange includes physicians' notes, demographics on patients, radiological images and reports, lab results, prescriptions, and insurance coverage information. The system, in the final phase of testing, should be operational in early 2005.

A way to assess how data exchanges improve the quality of health care is still lacking. Here again, local governments and foundations can play a critical role by investing in tools to measure such progress; the California HealthCare Foundation is conducting this kind of research. Studies that quantify the clinical benefits—from lives saved to medical errors avoided—might help persuade health care communities of all sizes to establish information networks.