Development of search strategies

Search strategies

We encourage all authors to cooperate with information scientists or librarians who might be available to them locally and to collaborate closely with the CMED group's Information Specialist (CIS), Maria-Inti Metzendorf (maria-inti.metzendorf@med.uni-duesseldorf.de).

As support to our authors the CIS helps to develop search strategies for new reviews or review updates in order to identify all potentially relevant studies for your topic. If you prefer to develop search strategies on your own, please feel free to do so. Nevertheless they will need to be handed in for peer review by the CIS.

Five resources need to be searched mandatorily: The literature databases Medline, Embase, and the Cochrane Library, as well as the trial registers ICTRP and ClinicalTrials.gov. Depending on the review topic, additional databases should be considered, e.g. LILACS, CINAHL, PsycInfo, AMED or the Science Citation Index (Web of Science).

To help with the development process of the search strategy, please provide relevant and precise terms (text words and MeSH) for

P = the population at interest and

I = the intervention to be evaluated.

It is very important to identify different wordings that are being used in publications relevant to your topic. You might be astonished about the many ways natural language allows to describe a population or intervention, e.g. people with diabetes mellitus type 2 are described as: individuals with T2DM, subjects with type 2 diabetes mellitus, elderly men with type 2 diabetes, diabetic patients, postmenopausal women at risk for type II diabetes, young patients with diabetes, diabetic population...

Hints for identifying relevant terms

1. A good way of identifying terms is checking the wording of title, abstract and keywords of relevant studies already known to you. Search them in PubMed and extract terms that describe population and intervention. Next, click on “Publication Types, MeSH Terms, Substances” found below the article´s abstract and extract the MeSH terms assigned to the article (cave: MeSH terms are only available when the status of the article is [PubMed - indexed for MEDLINE]).

2. If you identify only one or few studies on your topic, try out the “Related citations” link found below the record of your relevant reference in PubMed. This algorithm automatically provides similar references within PubMed. You might identify more studies on your topic by screening the first 50-100 records displayed.

3. Want to try out a cool tool that makes this process even easier? Try out PubReMiner. Insert the PubMed-ID (PMID) of a relevant study already known to you. PubReMiner will display all information on this article. You can also try this with several articles at once: Just insert all the PMIDs of relevant articles known to you, combine them with OR and look at PubReMiner's statistical analysis of terms.

Provision of search results and continuos alerting

The search strategy will be developed between check in of the first protocol draft for editorial approval and protocol publication. It will be sent to you for revision. The mutually agreed search strategy will be run on the databases as soon as the protocol is published. Search results (citations only, not full texts) can be provided to the review's contact person as EndNote-Library, RIS-file or PDF-file. Please indicate your preference. Note: it is the author's responsibility to obtain the full texts of potentially relevant studies identified during the screening process.

The CMED group recently implemented an automatic alert service that runs on Medline (via OVID) using the search strategy developed for the review project. New records added to Medline identified by the strategy will automatically be sent to the email address of the review's contact person.

After supplying the final review draft for editorial approval, the CMED group will perform a complete update search on all databases available at the Editorial Office and will send the results to the review authors. Should review authors identify new studies for inclusion, they will have to evaluate these, incorporate the findings into their review, and resubmit another review draft.