Beyond ACP Journal Club: how to harness MEDLINE for prognosis problems
ACP J Club. 1995 July-Aug;123:A12. doi:10.7326/ACPJC-1995-123-1-A12
This is the 4th editorial in the series on how to harness MEDLINE for clinical problems (1-3). Much of the clinically applicable literature addresses questions of therapy and diagnosis, followed by studies of etiology or harm. Prognosis studies are uncommon in the literature and, accordingly, are difficult to track down. Using methodology terms is, however, an effective way to retrieve articles that are clinically relevant and scientifically sound for patient care decisions. As with the previous editorials, we begin with a clinical scenario.
Your patient is a 32-year-old man. Several months ago he developed optic neuritis that was not associated with viral disease. In consultation with the local neurologist, you have treated him with intravenous methylprednisolone and oral prednisone. Although the disease was distressing at first, recovery was uneventful and vision had returned to normal within 3 weeks. He has booked a routine check-up and has told your receptionist that he wants to talk about long-term outcomes of optic neuritis. The neurologist had indicated to your patient in a brief encounter that there was a “pretty good chance” that he would develop multiple sclerosis (MS). He knows about the agonies of MS because a friend has it. He is very anxious and wants to hear from you just what a pretty good chance means.
Diagnostic and therapeutic information is often easily available in texts, but no text equivalent exists for prognostic information. One annual text that librarians often consult for prognostic information is Current Medical Diagnosis and Treatment by Tierney and colleagues (4). The 1994 edition states that “in patients with clinically isolated optic neuritis, as many as 75% will have developed MS within 15 years” and supports the discussion of the topic with an article by McDonald and colleagues from 1992 (5). Scientific American Medicine (6) is updated regularly and has a firm commitment to providing evidence to support its text. It states that “optic neuritis is suggestive of MS,” “optic neuritis may be the first manifestation of MS,” and “in 13% to 85% of patients with optic neuritis, MS subsequently develops.” References for these statements go back to 1985. The electronic version of ACP Journal Club (which is available on disk or through ACP Online) has 3 articles that mention both optic neuritis and MS. Because of the time lag, the article by Rodriguez and colleagues from the Mayo Clinic abstracted in this issue (7) is not included in the electronic version. One of the 3 studies by Beck and colleagues (8) for the Optic Neuritis Treatment Trial is listed. It reports that at 2 years, 7% to 17% of patients with optic neuritis developed MS depending on their initial treatment for optic neuritis. The lower rate is for patients prescribed methylprednisolone and prednisone.
The 75% affected number from Tierney and colleagues seems arbitrary and was not supported by the cited evidence; the 2-year data from Beck and colleagues are not sufficient for an accurate prognosis; and the range of 13% to 85% in Scientific American Medicine is too broad to be helpful. You decide that a MEDLINE search might be the best approach.
Optic neuritis is an “explodable” index term in MEDLINE and is grouped with neuromyelitis optica. You explode it (see the search strategy below) and ask that it be the major focus of the article (by putting an asterisk in front of the term). “Major focus” affects searching in the following way. Indexers at the National Library of Medicine assign a term to any idea that is described in 1 or more paragraphs. This results in most articles being assigned 10 to 20 terms. 2 to 3 of the most important terms are also identified as major headings, reflecting the most important aspect of each article. You can restrict retrieval to articles with a term indicated as the major focus by putting an asterisk in front of the term. Because optic neuritis is the major focus of our interest, we have designed the search strategy to represent this. Articles that mention optic neuritis in passing will not be retrieved.
This “weighted” search reduces the 7 million citations included in MEDLINE to 197 in the 1992 to 1995 segment—still too many to read. You will then want to select a methodologic term that will sift out all but the most appropriate studies. For prognosis, the strongest studies involve following a cohort of patients forward from the inception of their problem (9). On the basis of a careful empiric assessment, such studies are best detected using the search terms in the Table below (10). The best single term is “explode cohort studies (mh).” To narrow the search further, it is also worthwhile to limit the retrieval to studies of humans and to the language of your choice. The search now looks like this:
explode *optic neuritis (mh)
explode cohort studies (mh)
You retrieve 16 citations. 8 of them confirm that the short-term prognosis for total recovery from optic neuritis is good. Of the remaining citations, 1 is an editorial, 1 is a letter, 1 is a case study, and 1 is a study of children. The other 4 deal directly with your patient's information needs, and the study that seems to have the most valuable information is the cohort study of patients at the Mayo Clinic done by Rodriguez and colleagues (7). 156 patients who had optic neuritis from 1935 to 1991 were studied for a mean of 13.2 years (range, 0 to 51 y). Lifetable analysis showed that MS developed in 39% of patients at 10 years, 49% at 20 years, 54% at 30 years, and 60% at 40 years. The Mayo Clinic study also showed that persons with optic neuritis might actually have a greater 25-year survival than the general population after adjustment for age and sex. Your patient takes a copy of the Mayo Clinic abstract with him and is reassured that MS is not inevitable. (Our abstract for this article appears on page 21 in this issue of ACP Journal Club.)
Table 1 provides some other strategies for maximizing the sensitivity and specificity of searches for prognosis studies. It is of interest that the best search terms are different for 1986 and 1991. This is because indexing has improved over the years. Substituting the terms “prognosis” and “survival analysis” from the table to give maximal specificity backfires in the case of optic neuritis: The search yields 4 articles, most of which were not helpful. This may be because we are not looking for “survival analysis” after optic neuritis. The maximum sensitivity search retrieved 29 articles, including the Mayo Clinic study and publications of the Optic Neuritis Treatment Trial along with 15 of the 16 of the articles from the initial search. (The article describing children with optic neuritis was not retrieved.)
Table 1. Most Effective Searching Strategies for Prognosis (10)
|Time period||Strategy||Sensitivity %||Specificity %|
|Best single term|
|1986 (2 choices)||explode cohort studies||56||89|
|1991||explode cohort studies||60||80|
|To maximize sensitivity|
|1986||prognosis or explode cohort studies or mortality or natural and history (tw) or predict: (tw) or course (tw)||95||78|
|1991||incidence or explode mortality or follow-up studies or mortality (sh) or prognos: (tw) or predict: (tw) or course (tw)||92||73|
|To maximize specificity|
|1986||prognosis or natural and history (tw) or prognostic and factor: (tw)||58||97|
|1991||prognosis or survival analysis||49||97|
Prognostic information is difficult to find in texts; MEDLINE may be one of the better tools to use for retrieving clinically relevant information, particularly if methodology searching terms are used to separate the gold from the slag.
The next scheduled editorial for harnessing MEDLINE will be how to search for high-quality review articles and meta-analyses. Let us know if there are other searching (or filing) issues that you would like to see addressed.
K. Ann McKibbon, MLS
Cindy Walker-Dilks, MLS
R. Brian Haynes, MD, PhD
Nancy Wilczynski, MSc
1. McKibbon KA, Walker CJ.Beyond ACP Journal Club: how to harness MEDLINE for therapy problems ACP J Club. 1994 Jul-Aug;121:A10.
2. McKibbon KA, Walker CJ.Beyond ACP Journal Club: how to harness MEDLINE for diagnostic problems. ACP J Club. 1994 Sep-Oct;121:A10.
3. Walker-Dilks CJ, McKibbon KA, Haynes RB.Beyond ACP Journal Club: how to harness MEDLINE for etiology problems. ACP J Club. 1994 Nov-Dec;121:A10.