Systematic reviews and meta-analyses.

Properly performed systematic reviews can furnish an accurate summary of the published literature and thereby provide essential information for clinicians, guideline panels and policy makers. These reviews are increasingly used to develop management guidelines, and are now required by many agencies before considering funding for new research. A successful and rigorous review starts with the development of a clear question structured using the PICO format (population, intervention, comparison and outcomes). Prior to initiating the review, a protocol should be written that outlines study selection criteria, based on the same PICO format and study design. The search, also based on the PICO format, should be performed with the assistance of someone experienced in searches and involve multiple electronic databases without language restrictions. Additional sources include hand searches, the grey literature and correspondence with authors. Once a comprehensive list of titles has been made, selection of studies should be made by two reviewers working independently: first based on the titles, then the abstracts, then full text. Analysis is initially descriptive, including assessment of quality using published and validated checklists. Summary estimates from pooling (i.e., formal meta-analysis) are appropriate if the methodological differences between studies are not clinically important and if quantitative estimates of heterogeneity are acceptable. Reporting should follow published guidelines. Reviews that follow these methods can have a substantial impact on practice and policy. These reviews can help identify research priorities as areas where knowledge is uncertain, and also those topics where little additional work is required.