Figure
1: A more effective EBM information flow|
The Informatics Review > Thoughts - Evidence based medicine & information flow |
The World Wide Web has changed almost everything. Access to information that was unimaginable ten years ago has become a normal part of contemporary life. Anyone with a web browser and an Internet connection can read the news, shop for consumer items, or find out what movies are playing nearby. In healthcare, any clinician can click on a few links and search Medline. All this is possible without any specialized training in information retrieval.
Access to information is a good thing but it is not the same as skill in information retrieval. While the Internet has increased everyone's access to information, effective literature searches still usually require the expertise of information professionals.
According to the conventional wisdom all information processes should be automated: Wherever humans can be found slowing down information processes, "wetware" should be replaced with silicon. In the long run this might be possible but for now human expertise is still needed in various places. The medical library is one of those places.
I thought I was pretty good at searching Medline until I began a project to analyze the vast body of literature on healthcare outcomes. I needed to run systematic searches to retrieve citations to all kinds of articles that reported outcomes data. My Medline searches grew in complexity from a few MeSH terms to about 150 MeSH terms and textwords, organized into numerous groups that were combined with Boolean operators. It took several months until I was satisfied with the results of my Medline searches for outcome studies.
Clinicians cannot be expected to perform that kind of search on PubMed. And yet that is what evidence based medicine (EBM) requires -- complex searches that retrieve most of the relevant citations, while not retrieving most of the extraneous ones. This is what medical librarians do, and they do it very well.
But that is not the whole story. There are major inefficiencies in the medical library. Information flows in medical libraries are not adequate to meet the increasing need to deliver published evidence to the point of care.
Medical libraries have an interesting relationship with online information. Before there was a World Wide Web, even before there were personal computers, medical librarians were searching Medline. Despite their head start in online access to information, medical libraries have fallen behind in the area of serving information to their customers. A medical library might use the Internet to perform a literature search, but the clinician that requested the search still receives the search results printed out on paper, or pasted into a Word document. Clinicians who need monthly updates get monthly hard copies or Word documents.
The current state of EBM information flow produces fragmented, non-searchable lists of citations. A group of informatics researchers at the University of Missouri started the OutcomeLit project to address these inefficiencies of evidence based medicine. The aim of the OutcomeLit project was to apply web technology to the EBM needs of medical libraries. Our goal was to develop a system for medical libraries with the following characteristics:
None of this is rocket science. In fact, for goal #1, various products like EndNote make the process trivial. Goal #5 is already being met (partially) by medical libraries: The National Library of Medicine provides a service called "Selective Dissemination of Information" (SDI) that makes it possible for libraries to automatically re-run complex Medline searches once a month without the drudgery of re-entering the search terms. (SDI searches result in monthly text files of citations that remain isolated from each other, not updated databases.) The other goals in our list were not challenging either.
The challenge was to integrate these information processes into a system that is easy for librarians and clinicians to use. Our methodology, which is documented in detail on the OutcomeLit web site, accomplishes this and is based on the following applications:
Figure 1 diagrams the EBM information flow that is made possible by the OutcomeLit methodology.
Figure
1: A more effective EBM information flowThis web enabled EBM information flow uses a database to solve the problem of isolated lists of citations. Monthly updates to EBM searches are compiled into problem specific databases of citations. It uses web technology to serve the results of EBM searches to clinicians, and it uses librarians to do what human experts can do better than machines -- conceptual analysis and expert knowledge of a domain.
It is important to recognize the important role of medical librarians. Clinicians frequently ask librarians to perform literature searches without clearly defining either the types of articles they want to retrieve or the types of articles they don't want. A good medical librarian will then interview the clinician to make these clarifications. After a brief conversation the librarian will have a very clear concept of what the clinician needs. The librarian will then translate that concept into a Medline search -- a process that requires an expert level of knowledge of both the medical literature and the MeSH indexing system. While these points might seem obvious, it is all too easy to become mesmerized by information technology. The goal is not to install a Really Cool New System, but to create information flows that maximize the unique contributions that human experts and web technology have to offer.
When we began the OutcomeLit project in 1998, there were no software applications for creating web searchable bibliographic databases. Recently, ISI ResearchSoft has introduced a product called Reference Web Poster that can do this. Reference Web Poster uses the citation libraries from EndNote, Reference Manager, or ProCite, and produces a web page front end for users to search the database. Reference Web Poster is an excellent product, and medical libraries can use it to reengineer their EBM information flows. The OutcomeLit web site includes a comparison of Reference Web Poster and the OutcomeLit methodology. While one of these methods might be more or less suitable for a particular medical library, they both work well.
Medical libraries can use either of these methodologies to reengineer their EBM information flows. The particular methodology is less important than recognizing that inexpensive, easy to use software can be used to substantially improve EBM information flows. Web searchable, locally maintained bibliographic databases are an important next step in the EBM services that libraries can offer to clinicians.
References:
© 2000 Dean F. Sittig
|
The Informatics Review > Thoughts - Evidence based medicine & information flow |