MEDLINE and Clinical Practice
One of the major problems in medicine has been finding a way for the clinician actually practicing medicine for the average patient to keep up with the changing world of medicine. The clinician who did not know about the latest discoveries and the newest techniques would instead use outdated knowledge which might actually be detrimental to the welfare of the patient. Many times patients have had to endure more pain and complications than they should have, simply because their attending physician did not know about a relatively new technique which would have made the procedure go more smoothly.
For example, after the disastrous attack on Tenerife in 1797, Rear Admiral Sir Horatio Nelson (later to become Vice Admiral Lord Viscount Nelson of the Nile and Duke of Bronté) suffered through a lengthy convalescence due to complications of the amputation of his right arm. The surgeon aboard his flagship had followed the old practice of leaving the ligatures long and allowing them to separate naturally by necrosis and granulation until a gentle tug could pull them away. This generally took from two to four weeks, but in Nelson's case one of them became trapped and caused him a great deal of pain until it belatedly separated almost six months later. However he would not have needed to suffer so greatly had the surgeon been familiar with a technique developed eleven years earlier by Lancelot Haire, an assistant surgeon at the Royal Hospital at Haslar. In the 1786 London Medical Journal, Haire had described the technique of cutting the arterial ligatures short, near the knot, so that they would make their way out of the healing stump quickly, with little pain to the patient. Because this information had not made its way to the practicing surgeon in the cockpit of the Theseus, the admiral was not able to benefit from that technique and thus had to endure needless suffering. It is arguable that at least some of the pain that his shattered arm gave him for the rest of his life might have been averted had his surgeon known and applied the improved technique.1
Although communication has improved since the close of the eighteenth century, clinicians have continued to lag considerably behind in comparison to the vanguard of medical knowledge. Part of this has been due to the problem of communication. Clinicians often have lived well away from the areas where medical knowledge has been produced (research hospitals and medical schools), which meant that they were seldom able to learn about new developments without particular effort. Since World War II an analogous but opposite problem has also posed itself. Because of the rapid, even overwhelming, growth in the medical knowledge, it has become impossible for a person to keep up with the entirety of new medical knowledge. It would literally take every hour of every day to read all the medical journals, leaving no time for actual patient practice.
Thus it has become essential to find some way to selectively keep up with the growing medical literature. That is, the clinician had to find a way to quickly find the latest medical research on topics relevant to the needs of his or her own patients as those patients presented their medical needs. The solution to this problem of course had two components. First, the articles had be identified and organized in a meaningful way so that the relevant ones could be recognized and located without having to sort through a large number of irrelevant items (the signal-to-noise problem). This was solved by the development of various successive indexing schemes which involved trained indexers going through the various medical journals as they were published and attaching subject labels to various articles and organizing these subjects with their associated article references in alphabetic order. Second, the information had be delivered to the clinician out in Podunk General Hospital rather than remaining trapped in the library of the medical school. The growth of the computer and on-line networks seemed to promise a solution to this problem, if the necessary tools and skills for on-line searching could be put into the hands of the practicing physician.
American medical bibliography had its start shortly after the Civil War. John Shaw Billings started as librarian of the Surgeon General's Office in 1865 and in 14 years took it from 2500 volumes to over 100,000 as part of his determination to create a "national medical library." He also began a catalog of its holdings, both monograph and periodical. In 1875 he produced a specimen volume of a book catalog of the library's holdings. This was to be a single-alphabet subject-author dictionary catalog with each monograph indexed by both author and subject, while journal articles would be indexed by subject alone.2
This became the beginning of Index-Catalog of the Library of the Surgeon General's Office. Unfortunately the indexing and printing techniques of the day simply could not keep up with all the information that was coming out, so a monthly supplement known as Index Medicus was created to list current books and journal articles by subject only. This was produced by the librarians on their off hours, and the first issue came out in 1879. At the end of each year an author index and an alphabetical subject list would be compiled.3 Its early years were perilous, and often the US Army subscriptions formed almost twenty percent of its subscriptions. From 1899 to 1903 it ceased publication, leaving only the French Bibliographica Medica to provide bibliographic access to medical periodical literature. Only with funding from the Carnegie Institution did Index Medicus finally return.4
Even in the nineteenth century the problems of information overload were beginning to appear in the medical profession, and bibliographers were having to develop methods to cope with it. One way was to cover only the "best" material, while another was to look only at the most current materials.5 Furthermore, journal indexing techniques were still in their infancy, which meant that some of the practices that were applied to the early volumes of Index Medicus reduced the usefulness of the product for the end user. In those early days Index Medicus listed each article under a single subject heading, rather than under all headings it would fit, as has become the modern practice. The subject headings were arranged in topical format rather than alphabetical.6 This scheme of indexing had two major drawbacks for the average user. First, it meant that an article covering more than a single topic would be listed under only one of the possible subject headings, to be decided by the indexer, and would not be locatable in a search under the other subject headings. Therefore a person looking for articles on a given subject might never know that an article that covered that topic had been listed under the subject heading for another topic covered within it, and thus would never be able to find the citation. Second, using a topical hierarchy instead of arranging the subject headings alphabetically made searches for a given specific subject heading more difficult by forcing the searcher to think in terms of the hierarchy of ideas used by the indexers.
In 1916 the American Medical Association (AMA) began to publish its own Quarterly Cumulative Index to Current Medical Literature, indexing 150 journals. Unlike the Index Medicus of those days, it arranged its subject headings alphabetically rather than hierarchically by broad concepts and subjects within them. Its indexers also used as many subject headings as were necessary to adequately categorize a given article, so that it might be located under several subject headings. This publication lasted for ten years and grew to cover about 300 journals, about half the number covered by Index Medicus.7
In 1927 the two publications merged, becoming Quarterly Cumulative Index Medicus (QCIM). In 1932 it lost its Carnegie subsidy and the AMA became its sole proprietor. However it began to fall behind in its publication, becoming less useful as a source of timely information. During World War II the National Library of Medicine (NLM) began publishing the Current List of Medical Literature (CLML), a sort of "current contents" produced by shingling slips of paper in order to enable NLM to produce it rapidly and cheaply.8
By the middle of the 1950's QCIM had fallen three years behind in producing its semiannual volumes. Physicians and researchers were using CLML more because it was current. This led to discussions between AMA and NLM about the possibility of combining the two publications. At the same time technological changes were making streamlined information handling and printing possible. In 1960 the two publications merged into a single publication with the original name of Index Medicus, coming out of the old NLM offices, although AMA retained responsibility for cumulation for several more years.9
During this time Calvin Mooers developed the descriptor, a special kind of subject heading which made subject analysis more comprehensive. Other information systems developments having to do with cataloging and classification, including keyword-in-context, Boolean logic and citation indexing, changed the way people thought about indexing. At the same time computers were becoming an important tool for handling large amounts of information. This led to the development of the tape-oriented Medical Literature Analysis and Retrieval System (MEDLARS), used to put together print Index Medicus.10 It had its own special thesaurus of subject headings, the Medical Subject Headings or MeSH.11
By the 1970's a strong perception grew that there was a serious problem with the way in which new medical knowledge was being delivered to physicians engaged in clinical practice. Two physicians, Jeoffrey Stross and William Harlan, undertook a study of the way in which news about the efficacy of photocoagulation in treating diabetic retinopathy was transmitted to practicing clinicians. While attending a continuing medical education conference in Ann Arbor, Michigan, they surveyed almost three hundred clinicians. They discovered that only 33% of the physicians surveyed were aware that photocoagulation was the appropriate course of treatment for both patients described in their survey. Almost half of the physicians surveyed were aware that clinical trials had taken place, but only 28% reported having heard the results of those trials.
Stross and Harlan then investigated how well the information had been made available in the journal literature, with particular attention to those journals read by primary-care physicians. They discovered that The Journal of the American Medical Association published only a brief notation, while The New England Journal of Medicine did not mention the study at all in an otherwise excellent article on the treatment of diabetic retinopathy. Medical Letter published a very good review, while American Family Physician had only a letter from the Retinopathic Study Group. The only truly thorough coverage of the study appeared in a specialist journal, American Journal of Opthamology. Because the references in the general-practice journals were so brief, they were likely to be overlooked by the casual reader who was not particularly looking for such material.14
In the 1980's various new interfaces developed with the intent of allowing the clinician to work directly with the MEDLINE system. Beth Israel Hospital in Boston developed PaperChase, a search program designed to allow the user to merely type in the desired terms. The software then constructed a search using Boolean logical operators and retrieved a list of matching citations.
In 1984 Bibliographic Retrieval Services (BRS) of Latham, New York, developed Colleague, a system able to allow anyone with a standard computer and a modem to dial into their system. Its menu-driven search system used keyword-in-context search and retrieval strategies to make it easy for anyone with a minimum of computer skills to use their system. This was a break from the specialized terminals provided by the search service that had been the rule previously.15
However this did not mean that clinicians could simply sit down at the computer and start typing in their request the same way they might call down to a hospital library and make a request of the librarian. It remained necessary to stop and write out a plan for the search beforehand, then look up subject headings in order to construct a search. In order to make the most effective use of the tool it was also necessary to use limiters to select only appropriate types of articles (for instance eliminating foreign language items). Only then could the searcher go on-line. It was still absolutely essential for clinicians wanting to run their own MEDLINE searches to obtain access to a copy of MeSH and familiarize themselves with the annotated alphabetic list and the tree structures of subject headings. Using the controlled vocabulary enabled precise searching and thus reduced on-line connect times.
However the new technology also made it possible to access the literature in different ways. For instance a physician could develop a profile of topics particularly relevant to his or her practice, such as materials related to a long-term condition of a particular patient. This could be updated monthly so that the physician would be notified as new material came in instead of having to come across it while perusing the journal literature. Although such services appeared expensive, their proponents pointed out that in the long run it was likely that these services would actually be cheaper than the time and effort involved in trying to keep up with new developments by hand.16
However this did not mean that clinicians immediately rushed out to start using MEDLINE searches. Many studies performed in the late 1980's revealed that physicians simply were not using computer resources that were available to them. Many physicians still preferred to consult their colleagues rather than use electronic information systems such as MEDLINE to fill gaps in their knowledge. Of course this meant that the information would be only as good as their colleague's understanding of it.
Some pundits suggested that the designers of on-line information systems did not really understand what physicians were looking for and therefore developed systems that were inconvenient to use and often did not provide the desired materials. Because of the nature of medicine as opposed to the other professions, physicians had unique information needs which were not being taken into account, as well as established habits of work and study that did not mesh with the models familiar to other professions and utilized by the designers of many such systems.17 Practicing clinicians generally would phrase their information needs in manners directly related to their practice rather than as queries of the general body of knowledge. For instance they would ask if a given procedure was appropriate for a particular patient rather than asking about the indications for using that procedure in general.18
A study at McMaster University Medical Center examined the difficulties encountered by clinicians attempting to use these databases in the course of their work. Many of them encountered the sorts of difficulties that are common to all beginning database searchers, such as understanding the controlled vocabulary of subject headings or the use of Boolean logic to combine terms. The last problem was quite common in this search. Many clinicians would use the Boolean AND operator when OR would have been more appropriate, since ordinary English usage differs strongly in this regard.
Other difficulties encountered by clinician searchers had to do with a lack of clear understanding of the way in which the software worked, particularly GRATEFUL MED. In particular, many searchers would place subject headings on the title line, which meant that the software would then only search for that word in the titles of articles.20 GRATEFUL MED was designed to enable a clinician to use it for real information needs after only three hours of training. However the necessary simplification of the search system exacted its price upon the results of searches. Clinicians were generally able to get the things that they wanted, but their technique tended to be inferior to that of trained librarians. Most clinicians failed to use advanced searching techniques that increased precision of result.21
After an initial period of training and free access to the MEDLINE system, the clinicians involved in the study were randomly assigned to one of two groups. One group continued to receive their MEDLINE access free of charge, while the other group was assessed a charge for using the system. The group that had to pay for their access did fewer searches than those who continued to receive access for free. Presumably they were more likely to think twice about using the system when it came with a price tag attached. However the individual searches that they performed were only slightly shorter than those performed by their non-paying colleagues and incurred only a slightly smaller on-line charge. Thus the clinicians who had to pay for their searches applied selectivity primarily in determining which cases merited an on-line search rather than in the performance of the searches themselves. An examination of the results of the searches performed by the subjects of the study showed that there was no significant difference in the quality of the searches performed by either the control or the experimental group. In fact both groups were rated as having perfomed their searches at a level comparable to an experienced reference librarian.22
The National Library of Medicine ran a study to determine how physicians used MEDLINE for clinical practice and how it affected patient care. This involved interviewing a randomly selected group of 763 MEDLINE users, asking them about how they used the system in their clinical practice. The results of this study verified that physicians were indeed using MEDLINE to fulfill a wide variety of information needs in the clinical setting. Many physicians and allied health professionals were able to learn about new diagnostic and treatment options from materials found on MEDLINE. Many of these health professional were convinced that being able to obtain important information on MEDLINE enabled them to better serve their patients, even to the point of saving a patient's life. Most of them cited currency of information or ease of location as their reasons for preferring MEDLINE to their personal collections of textbooks and journals.23
However researchers began to discover that low-cost access and a user-friendly interface were not enough to get clinicians to use MEDLINE on a regular basis. This led medical educators to search for some teaching method which would ensure that medical students would continue using MEDLINE once they were in regular practice. They discovered that it was essential to integrate on-line searching into the curriculum as an important tool for accessing and managing the burgeoning medical literature in order to improve patient care, rather than treating the skill as an end unto itself. Furthermore, students had to be put in situations in which they had to use MEDLINE regularly in the course of their studies, preferably no less frequently than once a month, to cement in their minds the idea of turning to MEDLINE for their information needs.24
Recently yet another new way to search MEDLINE has appeared. A company known as HealthGate has begun providing access to the MEDLINE database on the World Wide Web, and has arranged with several of the Web search services so that an advertisement for it appears whenever a person enters a search for the term MEDLINE. Depending on how extensive their version of the database is and how frequently they update it from NLM, it may eventually become practical for clinicians to drop their subscriptions to pay-per-search database systems such as DIALOG and simply connect to the Internet via a flat-rate Internet Service Provider and perform MEDLINE searches on the Web. However being able to use a graphical browser such as Netscape Navigator or Microsoft Internet Explorer to reap the full benefits of the Web would require a more powerful computer than one needs in order to run terminal emulation for command-line interfaces typical to services such as DIALOG.25
Through the years physicians have struggled with the problem of keeping abreast of developments in their field. Aware that the quality of patient care often depends upon clinicians being aware of new techniques, the medical community has constantly striven to improve the channels of communication. With the advent of modern telecommunications and computer technology it has become possible for clinicians to access "virtual medical libraries" no matter what their physical location. This ability to have the wealth of medical knowledge at one's fingertips has enabled physicians to identify and apply the latest techniques even as patient needs present themselves.
Notes
- P. D. Gordon Pugh, ed. Nelson and His Surgeons (Nelson Chirurgiique): Being an Account of the Illnesses and Wounds Sustained by Lord Nelson and of his Relationship with the Surgeons of the Day Edinburgh: Livingstone, 1968. 15. Almost nothing is known about Haire beyond this one contribution to medicine. Even his vital statistics have eluded record.
- John B. Blake, "Billings and Before: Nineteenth Century Medical Bibliography" in Centenary of Index medicus, 1879-1979, edited by John B. Blake, Bethesda, MD: US Dept of Health and Human Services, Public Health Service, National Institutes of Health, National Library of Medicine, 1980. 31.
- Ibid. 32.
- Frank B. Rogers, "Index Medicus in the Twentieth Century," in Centenary of Index medicus, 1879-1979, edited by John B. Blake, Bethesda, MD: US Dept of Health and Human Services, Public Health Service, National Institutes of Health, National Library of Medicine, 1980. 53.
- Blake, 36.
- Rogers, 54.
- Ibid.
- Ibid. 55-56.
- Ibid. 56.
- Ibid. 57-8.
- Morris F. Collen and Charles D. Flagle. "Full Text Medical Literature Retrieval By Computer: a Pilot Test," JAMA, 254 (November 15, 1985) 2768.
- Ibid. 58.
- Collen and Flagle, 2769.
- Jeoffrey K. Stross and William R. Harlan, "The Dissemination of New Medical Information," JAMA 241 (June 15, 1979) 2622-4.
- Collen and Flagle, 2768-70.
- R. Brian Haynes, Anna McKibon, Dorothy Fitzgerald, Gordon H. Guyatt, Cynthia J. Walker and David L. Sackett, "How to Keep Up with the Medical Literature: V. Access by Personal Computer to the Medical Literature." Annals of Internal Medicine 105 (November, 1986) 812-4.
- Steven H. Woolf and Dennis A Benson. "The Medical Information Needs of Internists and Pediatricians at an Academic Medical Center." Bulletin of the Medical Library Association. 77 (October, 1989) 337.
- "David G. Covell, Gwen C. Uman and Phil R. Manning, "Information Needs in Office Practice: Are they Being Met?" Annals of Internal Medicine 103 (October, 1985) 596-9.
- Warling, Brian, Gilman, Lelde B. "Manual Versus MEDLINE Searches," The American Journal of Psychiatry, 1991 v 148, Number 5 P 686 (reply by John S. Lyons, David B. Larson, and Joseph C Bareta, PP 686-7) (in response to an article by Bareta, Larson and Lyons, "A Comparison of Manual and MEDLARS Reviews of the Literature on Consultation-Liaison Psychiatry." AJP, 1990, 147:1040-1042)
- Cynthia J. Walker, Ann Mckibbon, R. Brian Haynes and Michael F. Ramsden. "Problems encountered by clinical end users of MEDLINE and GRATEFUL MED." Bulletin of the Medical Library Association 79 (January, 1991) 67-69.
- K Ann Mckibbon, R Brian Haynes, Cynthia J. Walker, Michael F. Ramsden, Nancy C. Ryan, Lynda Baker, Tom Flemming and Dorothy Fitzgerald. "How Good are Clinical MEDLINE Searches? A Comparative Study of Clinical End-user and Librarian Searches." Computers and Biomedical Research 23 (1990) 583-593.
- R Brian Haynes, Michael, F. Ramsden, Ann McKibbon, Cynthia L. Walker, et al. "On-line Access to MEDLINE in clinical settings: Impact of user fees." Bulletin of the Medical Library Association 79 (October, 1991) 377-80.
- DA Lindberg, ER Siegel, BA Rapp, KT Wallingford, and SR Wilson. "Use of MEDLINE by physicians for clinical problem solving," JAMA , 269 (1993) 3124-9.
- ML Pao, SF Grefsheim, ML Barclay, JO Wooliscroft, BL Shipman and M McQuillan, "Effect of Search Experience on Sustained MEDLINE usage by students" Academic Medicine 69(November, 1994) 914-20.
- URL http://www.healthgate.com/HealthGate/MEDLINE/search.shtml
Bibliography
Blake, John B. "Billings and Before: Nineteenth Century Medical Bibliography" in Centenary of Index medicus, 1879-1979, edited by John B. Blake, Bethesda, MD: US Dept of Health and Human Services, Public Health Service, National Institutes of Health, National Library of Medicine, 1980. 31-52.
Collen, Morris F, and Charles D. Flagle. "Full Text Medical Literature Retrieval By Computer: a Pilot Test," JAMA, 254 (November 15, 1985) 2768.
Covell, David G, Gwen C. Uman and Phil R. Manning, "Information Needs in Office Practice: Are they Being Met?" Annals of Internal Medicine 103 (October, 1985) 596-9.
Haynes, R Brian, Michael, F. Ramsden, Ann McKibbon, Cynthia L. Walker, et al. "On-line Access to MEDLINE in clinical settings: Impact of user fees." Bulletin of the Medical Library Association 79 (October, 1991) 377-80.
Haynes, R. Brian, Anna McKibon, Dorothy Fitzgerald, Gordon H. Guyatt, Cynthia J. Walker and David L. Sackett, "How to Keep Up with the Medical Literature: V. Access by Personal Computer to the Medical Literature." Annals of Internal Medicine 105 (November, 1986) 812-4.
Lindberg, DA, ER Siegel, BA Rapp, KT Wallingford, and SR Wilson. "Use of MEDLINE by physicians for clinical problem solving," JAMA , 269 (1993) 3124-9.
Mckibbon, K Ann, R Brian Haynes, Cynthia J. Walker, Michael F. Ramsden, Nancy C. Ryan, Lynda Baker, Tom Flemming and Dorothy Fitzgerald. "How Good are Clinical MEDLINE Searches? A Comparative Study of Clinical End-user and Librarian Searches." Computers and Biomedical Research 23 (1990) 583-593.
Pao, ML, SF Grefsheim, ML Barclay, JO Wooliscroft, BL Shipman and M McQuillan, "Effect of Search Experience on Sustained MEDLINE usage by students" Academic Medicine 69(November, 1994) 914-20.
Pugh, P. D. Gordon, ed. Nelson and His Surgeons (Nelson Chirurgiique): Being an Account of the Illnesses and Wounds Sustained by Lord Nelson and of his Relationship with the Surgeons of the Day. Edinburgh: Livingstone, 1968.
Rogers, Frank B. "Index Medicus in the Twentieth Century," in Centenary of Index medicus, 1879-1979, edited by John B. Blake, Bethesda, MD: US Dept of Health and Human Services, Public Health Service, National Institutes of Health, National Library of Medicine, 1980. 53-61.
Stross Jeoffrey K, and William R. Harlan, "The Dissemination of New Medical Information," JAMA 241 (June 15, 1979) 2622-4.
Walker, Cynthia J, Ann Mckibbon, R. Brian Haynes and Michael F. Ramsden. "Problems encountered by clinical end users of MEDLINE and GRATEFUL MED." Bulletin of the Medical Library Association 79 (January, 1991) 67-69.
Warling, Brian, Gilman, Lelde B. "Manual Versus MEDLINE Searches," The American Journal of Psychiatry, 1991 v 148, Number 5 P 686 (reply by John S. Lyons, David B. Larson, and Joseph C Bareta, PP 686-7) (in response to an article by Bareta, Larson and Lyons, "A Comparison of Manual and MEDLARS Reviews of the Literature on Consultation-Liaison Psychiatry." American Journal of Psychiatry, 1990, 147:1040-1042)
Woolf, Steven H. and Dennis A Benson. "The Medical Information Needs of Internists and Pediatricians at an Academic Medical Center." Bulletin of the Medical Library Association. 77 (October, 1989) 337.
This paper was originally written as part of a course in the history of medicine, taught by Dr. John Haller of Southern Illinois University at Carbondale
Last updated November 5, 2010.