9.2.6 Search Strategy

The search for an Umbrella Review should aim to identify all research syntheses relevant to the review question. The protocol should provide a detailed strategy for locating research syntheses including the key terms to be used and the resources to be searched. Predefined search filters for reviews for various databases already exist and they are worthwhile investigating when developing the search strategy for the review. An example is the “systematic[sb]” search filter for PubMed. As many databases do not have a predefined search filter for review articles, in these cases, it is preferable to search with key terms such as “systematic” or “meta- analysis” across the title or abstract fields. Most authors will use these terms in the title of their publications to clearly identify the type of publication. Authors of JBI systematic reviews will be familiar with the recommendation to identify the document as a systematic review in the manuscript title to maximize the likelihood that it will be retrieved and read.

The search terms used should be broad enough to capture all relevant reviews. A three- phase search process should be used. First, initial keywords are identified followed by analysis of the text words contained in the title and abstract, and of the index terms to describe relevant reviews. The additional terms i.e., meta-analysis or systematic review need to be included in the key terms for searching. Second, database-specific search filters for each bibliographic citation database stipulated in the protocol are constructed, and finally the reference list of all included reviews should also be searched.

The search for systematic reviews rarely needs to extend prior to 1990 as there were very few systematic reviews published prior to that time (Smith et al, 2012). Essentially searching for the research syntheses conducted within the last five to ten years will yield original/primary research conducted 30+ years prior that has been included in the located reviews and research syntheses. As well as biomedical citation databases such as Medline, PubMed, EMBASE, and CINAHL, other sources to search include the major repositories of systematic reviews such as JBI Evidence Synthesis, the Cochrane Database of Systematic Reviews, DARE and the PROSPERO register. The federated search engine Epistemonikos that specifically targets research syntheses is also worthwhile using, particularly for initial searches. The databases searched for an Umbrella Review will depend on the review questions and objectives, for example, PEDro is a database indexing reviews relevant to physiotherapy, OTseeker, indexing reviews relevant to Occupational Therapy while BEME and the EPPI Centre Evidence Library are repositories of reviews relevant to education. Due to limitations of available resources, most JBI Umbrella Reviews will inevitably focus on including studies published in the English language. Where a review team has capacity, the search should ideally attempt to identify research syntheses published in any language and may expand the search to include databases that index languages other than English.

A comprehensive search for a JBI Umbrella Review should also encompass a search for grey literature or reports that are not commercially published. As decision makers are increasingly required to base their decisions on available evidence, more and more research syntheses are being commissioned by practitioners and health care policy makers in governments globally; as a result many reports available via government or organizational websites are syntheses of research evidence and may be eligible for inclusion in a JBI Umbrella Review. A JBI Umbrella Review should include a search of at least two or three relevant sources for “grey” reports.