Search and Browse
 
by
Marjorie Lazoff, MD
Emergency Medicine
Philadelphia, Pennsylvania
Medical Editor
Medical Computing Today

 
accepted for publication in Medical Computing Today August 1999
 
Originally published in the July 1999 issue of Medical Software Reviews


Sections
Catalog Sites - Free Text Search Engines - Keyword Search Engines - List of Lists

The Web has a number of medical search engines and catalog sites to assist physicians and other healthcare professionals in locating professional information online. Unfortunately, no single tool will consistently and reliably locate all online medical content. Most physicians use a few engines to broadly identify content on line, then supplement their search with the more laborious task of following links and manually browsing their favorite Web sites.

This NetView examines those navigation tools most likely to unearth scholarly and otherwise professional medical content; see the List of Lists section for access to other popular medical and health search engines and catalog sites not included here. Also not reviewed are general search engines such as Alta Vista, Excite, Yahoo, Infoseek, Google or About.com. These and other sites are commonly used, and are especially valuable for looking up new, controversial, or highly specific terms. However, many links will be to commercial fare, and these search engines may contain advertising, sometimes even related to the search topic itself, which will likely distract professionals.

Professional medical content is poorly organized on the Web. There is no universally accepted system of cataloging information such as exists, for example, with Medline or the Cochrane Database of Systematic Reviews (CDSR). HTML standards allow for the coding of reference keywords but these are not consistently and properly incorporated into Web pages. Worse, the Web has no universal cataloging system, such as the National Library of Medicine (NLM)'s Medical Subject Heading (MeSH) terms, for these keywords.

Also unlike Medline or CDSR, which cite only selected articles within medical journals, there is no a priori definition of what distinguishes professional medical content on the Web. Although it wastes time, misinformation or content primarily for marketing purposes is usually easy for physicians to recognize and disregard. But what of information designed, for example, both to inform physicians and to attract public interest in a university research project? Or a labor of love site filled with physician-written but sometimes outdated or unknowingly inaccurate articles? How does a profession go about peer reviewing the entire Internet? More disturbing is content that is highly valuable but inaccessible to search engines, such as editorials from the New England Journal of Medicine or some of the excellent content within Physicians' Online or medical school sites.

The key criterion for excellence is always the quality of the database, not the type of search engine or other navigation tools used to find it. But even the best database is fatally flawed if the user cannot get access to the desired information.

There are two general types of search engines, free text and keyword, each with its strengths and limitations. Free text search engines match the search terms within the full text database. They're ideal for searching relatively rare or narrow topics in a large database, or for searches within a limited database. For example, "Devic's disease" is best approached using a free text search, unlike the broader topic of multiple sclerosis, for which a searcher risks bringing up thousands of hits, many from the same page or article.

In contrast, keyword search engines match the search term within an index that links the selected topic and/or keyword to articles or pages. Indexing the database for keyword searches requires work, but if well constructed they tend to be easier and more accurate to navigate. Keyword search engines are best designed for large databases, or to search out common entities within a database. For example, articles on multiple sclerosis are best identified using a keyword search engine, especially when searches can be limited to specific subsections of the database, but "Devic's disease" is probably too specific a term to warrant separate indexing in most databases.

Of course, the best databases provide for both types of searching formats, or complement a free text search engine with an index. But even the best search engines on the Web have their limits. A recent Associated Press story, Search Engines Lag Behind the Internet, describes an NEC research study that found even the most expansive general search engine, Northern Light, searches only a sixth of the Web. The study also found that it takes more than six months on average for a new Web page to make it into a search engine's listings.

Sections Catalog Sites

These sites classify and list medical sites (not individual pages or articles) by topic or subject. They are good first stops for finding the most popular medical sites, but they don't dig deep into resources within a particular site and so often miss the best online information. Catalog sites themselves may contain search engines, but these search their classifications and not the individual medical sites. Consult free text and keyword search engine sites listed below for more in depth, scholarly, or less common content.

Emory University's MedWeb allows its extensive, well-cataloged database to be either browsed by subject (several generations of collapsing/expanding menus), or searched by keyword. An advanced feature accommodates searches for URLs or free text phrases. This traditionally popular site has undergone major revision in the past few years due to staffing changes. The group that left MedWeb went on to create y-DNA's MedWebPlus, a demonstration Web site for y-DNA products. Not surprisingly, the database and navigation tools are remarkably similar to MedWeb, although its search results page is arguably an improvement. Like MedWeb, this site's database is significantly outdated. Visit both, and then select one or the other to bookmark.

Members of the American Medical Informatics Association Internet Working Group serve on the editorial board of Medical Matrix, which both classifies and rates selected medical sites. Literally thousands of sites are divided into categories, then further subdivided and presented in annotated list form. The database can be browsed or free text searched using entry titles and descriptions. This site has an intuitive, clean interface and a number of worthwhile features but it requires free registration, has advertising banners, and is not as current as one would hope. Worse, the site ratings are not accurate, partly because raters seem to favor practical general medicine information over more limited specialty sites that are often of higher quality, and partly because non-clinical sites are rated using clinical site standards.

The University of Nottingham's OMNI engine searches the titles, descriptions, and keywords of selected medical, health, and management sites. Among other features, an advanced engine accommodates simultaneous searching of online nursing, dental, and/or biological databases as well. Users should also browse the alphabetical list of NLM sections, a complementary though less comprehensive database. The emphasis at OMNI is on UK sites and resources, and it includes many non-academic sites.

Two premed undergraduates, now medical students at Tufts, created the Multimedia Medical Reference Library (MMRL), a site with almost 20,000 links. The database is divided into categories for browsing, with a keyword search engine also available. The site's design and interface are problematic; browsing large topics is time consuming, and search hits are sometimes difficult to read. The last entry on the site is from August 1998 (alas, the first year of med school is time consuming). Several dozen medical audio files are listed as well, although most are from the Family Health Archives, a broken link.

The Doctor's Page is the creation of G.S. Nace, MD. Once a personal labor-of-love member page on AOL, Doctor's Page is now affiliated with The Wall Street Journal. It sports a new -- although not necessarily improved -- interface with a flashing advertising banner on top. Patient Education is now the first and most extensive section, followed by Medical Information (general medical physician sites), Coding Tools, Other Docs, and several sections housing lighter fare. Gone are the extensive lists of resources; this site is now one of the better introductions for new medical Web users and lay persons rather than professionals searching for specific medical content. Likewise, Medicine on the Net magazine maintains an extensive annotated list that will please those feeling lost on the Web, but frustrate others who seek content depth and updated links.

Two specialty catalog sites are worthy of mention: those interested in international public health will appreciate University of Pittsburgh's Global Health, and Neuroscience on the Internet has done an excellent job of cataloging neurology and related sites.

Sections Free Text Search Engines

The Society of General Internal Medicine Medical SmartSearch selects several online evidence-based medicine databases to search by natural query: Medline, AHCPR's National Guideline Clearinghouse, the British DARE (part of the Cochrane Library), British Medical Journal, NEJM, Journal of Pediatrics, and Circulation. The site is rich with extra features. For example, subscribers to the latter three journals can enter login information to access articles directly. A dialog window offers the opportunity to focus one's search, and ReutersHealth News can be read while awaiting search results. Other databases are added or avoided depending on the query. SmartSearch even suggests alternate searches when the pickings are slim. This site, hosted by the University of Texas at San Antonio, is a good one to bookmark.

CiteLine.com healthcare industry covers over 6,000 selected pharmaceutical and healthcare Web sites. Searches can be limited to disease and treatment, organizations, news and journals, and/or research and trials. Results appear as an annotated list of pages ranked by relevancy and include the page's last update. The search engine is slower than most and lists multiple hits from the same page, but frequently identifies pages and sites other engines miss entirely. One very nice feature is its Electronic Library Search, a link to six general search engines, each of which can be used to automatically search the same term(s). CiteLine.com has an elegant interface with minimally intrusive advertising and a left navigation panel linking to its commercial products, which emphasize search and strategy tools of interest to corporations and competitive business. The company has just been acquired by Caredata.com.

At five years old, America's Health Network's EINet's Galaxy: Medicine is part of the oldest searchable/browsable Web directory. This well designed user-friendly site has cataloged thousands of pages and sites. Both basic and advanced search engines are available, including Boolean operators and a search option limited to just page titles or URLs. Although actively maintained, the site is largely populated with older resources.

Sections Keyword Search Engines

OHSU's CliniWeb International organizes nearly 10,000 clinically oriented Web pages using the Medical Subject Headings (MeSH) disease and anatomy classifications. The assignment of MeSH terms is done manually by medical librarians, assisted by concept-mapping software. "International" refers to its multi-lingual search capabilities, which allows for searches in English, French, German, Spanish, and Portuguese. Its database can be searched or browsed. MeSH terms are also linked to canned PubMed searches on reviews, therapy articles, disease articles, or all articles. There is no advertising.

Created by two information retrieval specialists, Medical World Search (MWSearch) uses the NLM's Unified Medical Language System (UMLS) to search its selective database. If not used with care, this sophisticated search engine, here called a query processor, will bring up many extraneous hits without apparent relevancy ranking.

Sweden's Karolinska Institutet Biomedical Information Resources & Services by Subject or Diseases, Disorders and Related Topics is not very intuitive or user-friendly, and the Results page format varies depending on the section. Some searches brings up an intermediate page to sections related to the search term; just ignore the descriptions and click on the title of the section that sounds most appropriate, which will bring forth an extensive list of pages arranged in outline format. Compiled by medical librarians, it is less comprehensive than one would expect but sometimes contains rare treasures, including some from academic institutions around the world.

Sections List of Lists

OMNI's Launchpad houses more than a dozen medical and health search engines conveniently located on one page. Search terms must be entered manually for each engine. Internet Sleuth: Health uses a similar format but is far more extensive. It divides its engines into specialties and common diseases, each accessing a large variety of medical, journal, health, and government databases. Both are good sites to bookmark.

Scroll down EINet's Galaxy home page for a comprehensive collection of medical directories. Medical Matrix's Stat Search offers access to selected pharmacology, medical news, and clinical textbook databases, and selected health and general Web search engines. The technology is temperamental and the search engines don't always search the best online databases, but some may find this format useful.


 
Comments or questions for posting?
 
Archives of other articles
 
Introduction and List of other NetView articles