LITERATURE REVIEW & EVIDENCE GRADING
Duke Health performs its own literature searches and creates evidence tables. We have dedicated medical librarians from the Duke University Medical Center Library & Archives, and we have defined procedures and policies for evidence collection and review.
This process includes:
- Duke Health AUC multidisciplinary teams perform literature reviews in order to identify high-quality evidence, with searches developed by a medical librarian in cooperation with subject matter specialists who help select and refine search terminology.
- The medical librarian helps translate clinical questions into functional search strategies using relevant databases (including but not limited to MEDLINE). The medical librarian also updates the searches, as needed. Clinical guidelines and consensus statements by professional medical societies are included in the evidence assessment.
- The search strategy is designed to include search terms related to the clinical priority area as well as search terms related to imaging. Search terms will include a mix of database-specific medical subject headings and keywords searched in the title or abstract of the articles. Additionally, the searches are limited by study designs (e.g., guidelines, systematic reviews and meta-analyses, randomized controlled trials, etc.). Searches are limited by date (1990 – present) and to publications written in English.
- The search strategy results are uploaded to Covidence, a web-based screening software to which the Duke Medical Center Library & Archives subscribes, and reviewed by members of the multidisciplinary team in duplicate during two phases of blinded, independent screening (title/abstract review, followed by full-text article review of a smaller subset of articles). Studies are excluded from the review when two independent reviewers from the AUC multidisciplinary team determine to do so. Included titles are reviewed at the abstract and full article levels, with the same methodology applied.
- AUC multidisciplinary team members review the final list of included manuscripts and identify discrete pieces of evidence from each manuscript. A piece of evidence is defined by indications, including any limitations by population, history, prior imaging, signs, symptoms and/or lab values; the associated imaging study, including any particular protocol or modification of the study; and the information that the imaging study is expected to yield. A manuscript may include zero, one or more pieces of evidence.
- Each AUC multidisciplinary team member reviewing the manuscript independently assigns the Level of Evidence using the Oxford Centre for Evidence-Based Medicine (“OCEBM”) system. OCEBM uses levels 1-5, where 1 is the highest Level of Evidence. The members reviewing each manuscript compare their identification of evidence and resolve discrepancies by discussion and consensus; any discrepancies not resolved through this process are resolved by vote of the multidisciplinary team.