ISSG structured abstract for:
Wilczynski NL, Haynes RB. Developing optimal search strategies for detecting clinically sound prognostic studies in MEDLINE: an analytic survey. BMC Medicine 2004;2:23.
Structured abstract prepared by: Anne Fry-Smith
Objective:
The filters are designed to identify methodologically sound literature on the prognosis of health disorders in MEDLINE.
Methods:
The authors identified a gold standard of 1547 records by hand searching 161 core health care journals in 2000. 60% of the gold standard was used to test filters and the remaining 40% was used to validate the filters. The search terms for the filters were collected from relevant records and from consulting widely with experts in the field.
Results:
Several filters are offered. The most sensitive scored 82.3% in the validation test and the most precise scored 3.2% in the validation set. The most specific scored 94.2% in the validation set and the best compromise between sensitivity and specificity scored 73.4% sensitivity and
84.1% specificity (1.8% precision). The authors reported that the differences in performance between the development and validation sets were not statistically significant.
Discussion:
The authors discussed how it might be possible to increase precision and other performance measures of the search strategies, none of which were tested. The authors also compared the performance of these search strategies with those developed by themselves in 1991.
ISSG commentary:
None provided