Librarians are uniquely qualified to find, evaluate, organize, and manage the resources of a systematic review.
The question and protocol must be well designed from the beginning. A team should be assembled with the expertise to rigorously search, appraise, synthesize and report the results. A preliminary search of the literature can assess the quantity and quality of the literature on the topic and help define criteria. Existing or in-process reviews on the topic can be found, the feasibility of the review can be evaluated, and criteria can begin to be developed.
The question should be defined using a process such as PICO. A protocol of selection criteria, outcome measures, search strategy, etc. is then developed, using standard guidelines and to minimize risk of bias. The protocol is registered with PROSPERO or other appropriate protocol database to avoid duplication.
To minimize risk of bias, all studies that meet the protocol criteria need to be identified. This requires that multiple databases be identified and searched using the terminology and strategy appropriate to each database. Librarians can bring a great deal of expertise to the search process such as development of a comprehensive search strategy and management of the resulting citations. Searching the grey literature, such as abstracts, conferences and clinical trials, and doing hand searches of cited literature also ensure complete results and reduce publication bias. The search strategy should be well documented.
Using the criteria developed for the protocol, studies are reviewed and included or excluded. At least two members of the team should independently screen and select studies. Once studies are selected, the data needs to be extracted and then checked for accuracy and completeness using standard extraction forms. Critical appraisal will test for validity, results, and relevance when looking at populations, interventions and outcomes. Tools are available to assist with both data extraction and appraisal.
After the data has been extracted, the findings will need to be aggregated and summarized in an understandable manner. Charts, tables, and graphs are often used for this. A meta-analysis is one of the methods available to synthesize the data. The characteristics of the evidence for each outcome in the protocol are systematically assessed. The relevance, strengths, limitations, patterns, and flaws in each study as well as in the review as a whole are described.The synthesis should lead to a comprehensive statement of findings for evidence based practice.
The final report should be detailed and extensive and made available for free public access. Detailed information describing search strategies, methodology, validity and bias assessments, and conflicts of interest are some of the areas to be included so that decision making by the reader will be well informed, and so that the study is reproducible. The PRISM checklist is the minimum standard of what should be reported. Style manuals are available for journal publications and for publishing requirements of agencies.