Identifying systematic reviews in MEDLINE: developing an objective approach to search strategy design

Background: systematic reviews are of increasing importance to health care professionals seeking to provide evidence-based health care, because they provide systematically prepared summaries of the current state of research knowledge on the effectiveness of health care interventions. To be able to make use of them, both researchers preparing systematic reviews and health care professionals need to be able to identify systematic reviews easily. In the past, systematic reviews have been difficult to identify easily among the mass of literature labelled 'reviews'. Search strategies have been published which were intended to provide easier access to systematic reviews and meta-analyses. These strategies, mostly devised for the MEDLINE database, have been constructed using the authors' knowledge of the subject area, the databases and past experience. Objective: the objective of this study was to design search strategies based on a more objective approach to strategy construction. Methods: word frequencies from the titles, abstracts and subject keywords of a collection of systematic reviews of the effective health care interventions were analysed to derive a highly sensitive search strategy. Results: the proposed strategy offers 98% sensitivity in retrieving systematic reviews, while retaining a low but acceptable level of precision (20%). Other strategies with other levels of sensitivity and precision are also presented. Discussion: this study shows that a frequency analysis approach can be used to generate highly sensitive strategies which retain adequate levels of precision when retrieving systematic reviews.