Author Affiliations: Institute for Healthcare Improvement, Cambridge, Massachusetts (Dr Davidoff); Annals of Internal Medicine (editor emeritus), Philadelphia, Pennsylvania (Dr Davidoff); Hartford Medical Society, Farmington, Connecticut (Ms Miglus); and University of Connecticut Health Center, Farmington (Ms Miglus).
The output of published clinical evidence, once a gentle trickle, has increased into a torrent.1 How ironic it is that the volume, richness, and complexity of the clinical literature (undeniably its greatest assets) have now reached the point where they are making it all but unmanageable.1 But that literature is more important than ever, because questions continue to be raised at a rapid rate in clinical practice—about 5 questions per physician per half-day primary care session, about half of which go unanswered.2 Streamlining input to the literature,1 although in some ways an attractive option, carries its own risks and would not address serious problems on the output side, particularly the limited amount of time available to clinicians for consulting the literature. Those problems will remain unresolved unless and until there is a remedy for the gross inadequacy of current mechanisms for finding, extracting, and delivering the best possible information to where clinical decisions are made—the weakest link in the chain of research evidence.
Ultimately patients have the most to gain from strengthening this essential link. But it is also necessary to confront the sobering reality that the inability to bring the right information—when and where it is needed, for every patient, every time—is a major factor in preventing full return on society’s massive financial (and intellectual) investment in biomedical research. Smith3 recently suggested that a properly designed information delivery “machine” could provide the link. But even if such a machine can be built, an exclusively mechanical approach is likely to help only in limited areas of medical information delivery; besides, building and maintaining such a machine is likely to be extremely expensive. Opting for a technological fix ignores a currently underdeveloped but uniquely useful alternative, ie, human information delivery systems, including the services that reference librarians and pharmacists have provided for years, as well as the more recent clinical librarian and informationist programs.
Clinical librarians work from libraries but attend rounds and conferences to identify clinical questions and then deliver quality-filtered information directly to clinicians in timely fashion. Although clinical librarian programs have made considerable headway during the past 40 years in both the United States and other countries,4 only about 12% of Canadian and US medical libraries currently deliver information this way.5 Experimental evidence regarding the effects of clinical librarian services has been difficult to come by, but a wide array of observational reports leave little doubt about the utility of those services in guiding clinical decisions and saving clinicians' time.4
Since it was first proposed more than 10 years ago,6 the new profession of informationist has reframed and extended clinical librarianship in important ways. Informationists are cross-trained in reference library skills and the essentials of specific biomedical disciplines. They work as full-time members of clinical and some research teams while maintaining formal relationships with their home libraries. They search the literature in response to questions, critically appraise the findings, and summarize them in concise written reports. The National Library of Medicine kick-started development of this new path by funding informationist training fellowships for several years, and about 1 medical library of 20 in the United States and Canada provides information through this model.5 Moreover, informationist programs are now a major component of information delivery systems in several institutions, including Johns Hopkins and Vanderbilt University medical centers, and 15 full-time informationists work at the main clinical and research campus of the National Institutes of Health.7 A rigorous controlled study of the Vanderbilt program has demonstrated that informationist services effectively influence the initiation, addition, and change, as well as discontinuation and inhibition, of clinician-dependent actions; the program's primary effect was on use of new or different treatments (odds ratio, 8.19; 95% confidence interval, 1.04-64.00).8
The collective experience with clinical librarian and informationist programs, however limited, has already provided many useful lessons: user buy-in is crucial, the physical presence of information professionals in clinical and research work areas greatly enhances the use of information delivery services, and clinicians who have access to these services actively continue to search the literature themselves and consult expert colleagues. Importantly, even when performed by skilled librarians, sufficient literature searches require at least half an hour, and often many hours (which undoubtedly discourages clinicians from searching the literature on their own), and an adequate literature search by a librarian costs on average about as much as a chest x-ray.4
It is time to build a medical information delivery system worthy of the medical profession. Details of a comprehensive information delivery strategy will need to be hammered out over time, but at a minimum they will need to take into account the multidimensional nature of action-related patient care questions. Generally speaking, action-related questions fall into 3 categories: simple (although not necessarily easy; for example, “How do you cook an omelet?”), complicated (“How do you send a rocket to the moon?”), and complex (“How do you raise a child?”).9 These generic categories parallel the 3 empirically derived categories of clinical questions described by Ely et al2: simple (for example, “What drugs interact with drug A?”), compound (“Can cocaine precipitate a sickle cell crisis?”), and conditional (“What are the screening guidelines for breast cancer when the patient has a family history of breast cancer?”). Electronic decision support, including apps in handheld devices and “info buttons” in electronic health records, can likely continue to handle simple and some complicated clinical questions, as Smith suggests.3 Good answers to complicated questions can be found more quickly and effectively when clinical librarians are involved in the process. Informationists' skills are of particular value in assembling the answers to complex questions.
A truly effective medical information delivery system will require strong political, administrative, and financial support from both public and private sectors. As a value-added clinical service, it should be paid for, at least in part, as a clinical service rather than primarily through library funding. The medical library community will need to play a central role in creating and operating this new, hybrid library-clinical system. That community has been working hard to reinvent itself and is therefore in a strong position to do so. For example, the Johns Hopkins medical institutions will soon close their central library building, replacing it entirely with multiple informationist-staffed units embedded in individual departments backed by extensive digital resources. Medical librarians in many places now coach users on sophisticated information retrieval strategies; others help create new systems that integrate information from multiple sources; some are involved in translational research and bioinformatics. The challenge of clinical information delivery clearly goes well beyond any single constituency, however, and many stakeholders, including hospital systems, insurance companies, accrediting bodies, federal agencies, and patients, will ultimately need to be at the table. The system will need to be built incrementally, with evaluation mechanisms “baked into” sequential build-test cycles, so that its operation and cost-effectiveness are repeatedly and independently assessed, and it becomes, as it must, a continuous learning system.
The recently enacted Patient Protection and Affordable Care Act includes no provision for improving clinical information delivery, a truly unfortunate omission given medicine's heavy information dependence. Despite that lapse—or because of it—passage of the act should galvanize everyone in health care into action. Cochrane had it right in 1979 when he reflected that “It is surely a great criticism of our profession that we have not organized a critical summary, by specialty or subspecialty, adapted periodically, of all relevant randomized controlled trials.”10 But Cochrane did not go far enough: such summaries, indeed all medical discoveries, are of no use to patients and clinicians if they remain buried in the literature. It is surely an equally great criticism that the medical profession has not developed an effective, efficient system for finding relevant clinical information, extracting it, and delivering it when and where it is needed into the hands of everyone who makes medical decisions.
Corresponding Author: Frank Davidoff, MD, 143 Garden St, Wethersfield, CT 06109 (firstname.lastname@example.org).
Conflict of Interest Disclosures: Both authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.
Additional Contributions: We are grateful to Valerie Florance, PhD, National Library of Medicine, for comments on an earlier version of the manuscript. She did not receive compensation for this contribution.
Davidoff F, Miglus J. Delivering Clinical Evidence Where It's NeededBuilding an Information System Worthy of the Profession. JAMA. 2011;305(18):1906-1907. doi:10.1001/jama.2011.619