Beyond ACP Journal Club: how to harness MEDLINE for diagnostic problems
ACP J Club. 1994 Sept-Oct;121:A10. doi:10.7326/ACPJC-1994-121-2-A10
Related Content in the Archives
• Editorial: Beyond ACP Journal Club: how to harness MEDLINE to solve clinical problems
• Editorial: Beyond ACP Journal Club: how to harness MEDLINE for therapy problems
• Editorial: Beyond ACP Journal Club: how to harness MEDLINE for etiology problems
• Editorial: Beyond ACP Journal Club: how to harness MEDLINE for prognosis problems
• Editorial: Beyond ACP Journal Club: how to harness MEDLINE for review articles
The previous editorial in the MEDLINE series, published in the July/August 1994 issue of ACP Journal Club, was on searching for articles addressing therapy questions. This editorial deals with searching for articles on the diagnosis of clinical problems. As with the other editorials in the series, we start with a clinical situation.
One of your long-standing patients, a 50-year-old woman, has made an appointment to ask for your advice. She has come in often over the years with various complaints, including some vague stomach and bowel upsets and weight swings that seem to “run in the family.” She is generally healthy although a bit thin. Her niece has been diagnosed with celiac disease. She asks you if screening certain members of her family for this condition might be warranted. Both you and your patient are not all that happy about the prospect of small-bowel biopsy. She asks you about the availability of other less invasive tests.
You remember reading about the diagnostic value of the gliadin antibody test for celiac disease. You tell her you will check to see if the test could be used as a substitute for the biopsy.
You check your personal reprint file and some general medicine evidence-based texts on diagnosis (1-3) and do not find any references to either the test or the disease, even using alternative names such as sprue and gluten enteropathy. Scientific American Medicine (4) does not mention the gliadin test but suggests using fecal fat analysis with the biopsy. ACP Journal Club has not included a single article on celiac disease. It is now too late to call the local gastroenterologist. MEDLINE is next on the list.
Each of the 7 million citations in MEDLINE includes the title of the article, authors, the abstract if available, and keywords assigned by content experts working for the National Library of Medicine (NLM). When you search, you can use the index terms given to the article (Medical Subject Headings [MeSH]), or words in the titles and abstracts of the original article, or a combination of these 2. In other words, you search by trusting the indexers (MeSH headings), the authors (title and abstract), or combine both. Most often, the indexing is a good first approach for MEDLINE searching. By using the indexing, you avoid some of the difficulties of having synonyms for items (e.g., celiac disease, sprue, steatorrhea, and gluten enteropathy), alternative spellings and word endings (e.g., etiology vs. aetiology; esophagitis vs. oesophagitis; and random vs. randomized), and different meanings for the same word (e.g., shoe polish vs. speaking Polish; and AIDS, the disease vs. hearing or teaching aids).
Using title and abstract searching (called textword searching) can be helpful for new concepts or ideas that the indexers have not yet incorporated. For example, it took some time for MEDLINE to include Legionnaire's disease as a MeSH term and restenosis is still on the waiting list. Furthermore, some concepts are not indexed consistently (e.g., sensitivity, specificity, positive and negative likelihood ratios). 2 sample searches follow that show the different approaches and output for a search to find articles that provide the sensitivity, specificity, and predictive values of gliadin antibody tests for screening for celiac disease.
This first search is a textword search:
sensitivity OR specificity OR likelihood OR predictive
celiac OR sprue OR stretorrhea OR gluten enteropathy
This search retrieves 37 citations from 1991 forward, about half of which are relevant to the topic.
The second search, based on MeSH headings is as follows:
celiac disease (mh)
exp sensitivity (mh)
English language (la)
This search retrieves 22 citations, 19 of which were in the first search plus 3 unique citations. Although there were fewer citations in the second search, a larger proportion of them were “on target” than those citations in the textword search.
Either set of citations can help answer the question about the diagnostic properties of gliadin. One of the first citations in each set is abstracted in this issue of ACP Journal Club (see page 51). Bodé and Budmand-Høyer (5) conclude that for adults the sensitivity of the gliadin antibody test was 77%, specificity was 95%, the positive predictive value was 71% to 90%, and the negative predictive value was 97%. The commentator concluded that the gliadin antibody test is an effective noninvasive screening test for patients with suspected celiac disease. A biopsy, however, is still needed to confirm the diagnosis.
Our previous editorial provided the single terms for effective searching for studies of therapy. Table 1 has the best single terms for effective searching for studies of diagnostic tests. We have included 2 searching terms to give you a choice of high sensitivity or high specificity.
Table 1. Sensitivity and Specificity for the Best Single Searching Terms
|Best single searching term||Term sensitivity||Term specificity|
|Before 1990||1990 and on||Before 1990||1990 and on|
*Note that the terms diagnosis& (px) means that all the MeSH subheadings that deal with diagnosis will be included in the search automatically, a feature called “exploding” or, in this case, “pre-exploding.” We discussed exploding in our previous editorial when we grouped all types of headaches together for the therapy scenario.
Table 2. Textwords and MeSH for Searching for Diagnostic Problems
|Textwords for searching for diagnostic problems||MeSH for searching for diagnostic problems|
|sensitivity, specificity, predictive and value, false and reaction, likelihood and ratio||sensitivity and specificity, predictive value of tests, ROC curve, false negative reaction, false positive reaction, diagnosis, differential, mass screening, diagnostic use (sh), diagnosis* (sh)|
*This subheading can be used in various ways by different searching systems. Use your manual or check with a medical librarian for exact uses and searching conventions.
NLM and other vendors produce many databases, and some of these may be better tailored to specific clinical questions than the general MEDLINE file. Most searching systems, including GRATEFUL MED, can access these databases. Others are available only through specific vendors who provide various “subsets” of MEDLINE and journals to suit specialized interests. EMBASE (Excerpta Medica online), a European “equivalent” to MEDLINE, is another general alternative to MEDLINE with an overlapping set of journals.
The following specialized literature databases are available through NLM and other vendors. AIDSLINE contains references on AIDS and related topics. References are taken from MEDLINE, HEALTH, and CANCERLIT databases from 1980 forward, with additional unique citations added. AIDSLINE and other AIDS-related databases such as AIDSTRIALS and AIDSDRUGS are available free of charge for both health care professionals and consumers.
AVLINE is an audiovisual catalogue that contains listings for all audiovisual materials catalogued by NLM since 1975. The materials are intended for health professionals rather than patients.
CANCERLIT provides citations to cancer literature. Books, technical reports, meeting abstracts, meeting papers, and theses are also included. Material dates back to 1963.
CATLINE contains citations to printed books and journals in the collection at NLM.
CHEMLINE is a combined dictionary and database containing data that can be helpful in selecting strategies for other NLM databases. A user can identify a substance, determine its Chemical Abstracts Registry number, and find out which databases would carry more information on the substance.
DIRLINE contains information on over 15 000 organizations that act as information resource centers. DIRLINE references include names, addresses, telephone numbers, and descriptions of each organization, including primary interests and services.
HEALTH Planning and Administration database provides references back to 1975 on nonclinical aspects of health care delivery including administration and planning of health facilities, services, health insurance, and health care aspects of financial management, regulation, personnel, quality assurance, licensure, and accreditation.
SDILINE is a database that contains references from the current month of MEDLINE and permits you to “store” searches for automatic updating each month.
TOXLINE and TOXLINE65 contain references to the pharmacologic, physiologic, biochemical, and toxicologic effects of drugs and other chemicals. TOXLINE has the current files.
TOXLIT and TOXLIT65 contain references covering the pharmacologic, physiologic, biochemical, and toxicologic effects of drugs and other chemicals. These files carry royalty charges and are more expensive to search than TOXLINE and TOXLINE65. The toxicology files have some overlap, but enough citations are unique that both may need to be searched for comprehensive retrievals.
K. Ann McKibbon, MLS
Cynthia J. Walker-Dilks, MLS