ISSG structured abstract for:
van der Weijden T, Ijzermans CJ, Dinant GJ, van Duijn NP, de Vet R, Buntinx F. Identifying relevant diagnostic studies in MEDLINE. The diagnostic value of the erythrocyte sedimentation rate (ESR) and dipstick as an example. Family Practice 1997;14(3):204-8.
Structured abstract prepared by: Cynthia Fraser
Objective:
To examine sensitivity and positive predictive value of MEDLINE searching for diagnostic studies for two test (ESR and dipstick) relevant to primary care.
Methods:
Reference standards for each test were created from appraising studies reported in journal articles in a personal database (for ESR) and from four literature searches (for dipstick). The performance of search strategies comprising only MeSH terms were compared with those including both MeSH and free text terms for both tests. The addition of primary care terms was also tested.
Results:
Combined diagnostic methodology MeSH and free text terms searches had higher sensitivity than MeSH only searches as measured against the reference standard for both tests. Dipstick searches, which used disease terms rather than diagnostic study terms, had higher positive predictive values. Inclusion of primary care terms reduced sensitivity.
Discussion:
Combining MeSH with free text terms without restriction to a primary care setting, produced the best results in terms of sensitivity. Authors concluded that the predictive value of the search seems to depend on the breadth of the disease area.
ISSG commentary:
The authors adopted a pragmatic approach in developing and testing their search filters and these were validated on small data sets, especially for the dipstick search. The ESR search used test terms in combination with diagnostic methodology terms whereas the dipstick terms were combined with disease terms. The authors suggest that the use of disease terms is preferable; however any comparison is problematic because the comparison uses search terms relating to different tests.