Medical Information on the Internet

The continually changing and expanding body of medical information is increasingly difficult to master. In 1985, Covell and colleagues¹ documented the obstacles physicians face using books and print media as sources of information in practice. Thirteen years later, there are 9.2 million MEDLINE citations with approximately 31000 more added each month.² Combined with the psychological, sociological, and administrative complexities of medical practice, this rapid proliferation of information pressures physicians to make decisions at the margin of what they remember and know. Compounding this problem, changes in health care delivery require practitioners to make more important and complex decisions in less time. Physicians' trouble in applying current health care evidence effectively^{3 - 4} will almost certainly worsen given these trends. Weed⁵ calls this failure to use current health care evidence in the practice of medicine "avoidable ignorance."

As computer systems become ubiquitous and easier to use in the clinic, hospital, and at home, the potential to access information when and where it is needed is enormous. Bibliographic resources such as MEDLINE, textbook collections, full-text journal articles, drug databases, and multiple Internet sites are currently available online. However, several barriers still exist, as discussed in articles in this issue of THE JOURNAL. Electronic information retrieval systems are not used more frequently, Hersh and Hickam⁶ conclude, because they are still difficult to use and often do not provide clinically expedient information. Physicians may be more likely to use new information if it were systematically reviewed and summarized by experts and presented in a more clinically useful and accessible format. But what should that format be? McDonald et al⁷ describe the need for system-to-system standards such as Health Level Seven so that an electronic medical record system can send and interpret messages to and from other computer systems.

Electronic information resources also require specifications and standards to define the clinical questions to be answered. For example, a resource for antimicrobial therapy should provide a simple and reliable means of requesting and receiving a recommended antibiotic for each infectious disease or syndrome.⁸ Structural specifications are needed to define how this information can be searched, presented, and linked to an electronic medical record.⁹ Other barriers exist, including the lack of the necessary hardware and software at many clinics and hospitals to access the Internet efficiently where care is provided.

The true impact of these barriers has not been addressed systematically, and relatively little is known about the information needs of physicians.¹⁰ Systems that attempt to facilitate the access of information in clinical care need to be tested. Unfortunately, an important hurdle remains, one that has received little attention. Medical information for integration in clinical systems requires licensing access, which can raise costs and stymie efforts at integration through idiosyncratic interfaces and constraints. Rather than raising barriers, online access should provide cost savings by eliminating print and distribution costs. At a minimum, authors, publishers, and system developers should work together to create and evaluate systems that have the greatest chance of improving clinical care.

The just-in-time model offers an appropriate metaphor for providing medical information access.¹¹ In the same way that a production line requires specific components from its inventory to be delivered at the proper time, a physician needs specific information from memory or an outside source to be provided at key points during care. Components of this model have been embraced by several academic medical centers,⁷ including ours.¹² The goal is to combine high-quality, full-text medical information resources using a single set of information retrieval skills and in association with other clinical information.

Whether physicians will use medical information in such an integrated system remains to be seen. Hersh and Hickam⁶ raise the concern that information online does not equal information in use. However, Sackett and Straus¹³ found that relevant, accessible evidence at the point of care was used and useful for clinical care. Future studies should measure the actual changes in clinical management and patient outcomes.⁶

The vision of a unified information network that delivers frequently updated, clinically relevant, highly valid, and deeply integrated medical information over the Internet is technically feasible. The success of such a network depends on the technical and economic cooperation of academicians, publishers, practitioners, electronic medical record vendors, and medical societies among others. In the end, conquering "avoidable ignorance" will require the willingness to say "I do not know the answer, but I know where to find it." By providing answers when and where they can make a difference, perhaps we can find the humility and courage to conquer our own ignorance.