The past 50-plus years have seen dramatic advances in medical discovery. Only 51 years have elapsed from the identification of the double-helix structure of DNA (1) to the publication of the first estimate of the human genome (2). Furthermore, the specialty of rheumatology is also relatively new. The name "rheumatologist" was first introduced in 1940 by Bernard Comroe (3,4), and the first subspecialty board exams took place in 1971. As a field of study, rheumatology changed radically in the early 1950s largely because of the discoveries of rheumatoid factor, the lupus erythematosus cell, and the therapeutic effect of cortisone (4).Keeping pace with medical discovery has been an exponential increase in the volume of published literature, as well as a dramatic improvement in the technology available to search this literature. The PubMed database contains over 16 million citations published between 1951 and 2005. This is remarkable considering the fact that the National Library of Medicine's IndexCat (a database for citations older than those available in PubMed) lists only 2.5 million references to publications from late 17th to the early 20th centuries.The ability to access over 50 years of publication data electronically with uniform indexing, paired with scientific developments, invited our approach to examining publication trends in the field of rheumatology. We were interested in studying how the number of articles on particular topics changed over time. We theorized that a number of patterns would be displayed that would reflect the history of our specialty.
PubMed searchWe reviewed the indices of Arthritis & Rheumatism (A&R) from volume 1 (1958) to the present and developed a list of diseases and concepts considered central to the field of rheumatology. These terms were searched in PubMed. For all but two searches, we allowed PubMed's automatic term mapping to enhance the terms entered (Table 1). This function searches the database not only for the phrase entered as text in the title or abstract of the citation, but it also maps the phrase to the appropriate Medical Subject Heading (MeSH). For example, the search for superantigen in PubMed mapped to the MeSH term Superantigens and performed a search that included all articles indexed with that term. Nearly 25% of the citations in the resulting set did not have the words superantigen or superantigens in the title or abstract. Additionally, the subject headings are automatically "exploded," meaning that if a term has corresponding specific subject headings within the MeSH tree structure, these headings are included in the search. For example, the phrase rheumatoid arthritis automatically maps to the MeSH term Arthritis, Rheumatoid and includes citations with the more specific subject headings of Arthritis, Juvenile Rheumatoid; Caplan's Syndrome; Felty's Syndrome, etc.Automatic term mapping was deemed inadequate for two of the searches: the illnesses associated with silicone breast implants and rapeseed oil poisoning. For these topics, we constructed more compl...