A roadmap for searching literature in PubMed: Evaluating

You have given diligent thought to your search strategy and its execution. Generally speaking, though, that’s not the end of the road. Take a good look at your search results and adjust your search parameters: add search terms to your building blocks, adjust the building blocks, add filters or remove them. Take another good look at your research question. Will you be able to use the articles you’ve found to answer your question satisfactorily? Also, take another good look at your building blocks. Are your Boolean operators and brackets constructed properly? Have you proofread for typos? Ask yourself the following questions:

• Has my search returned too many results?
• Has my search returned too few results?
• Why do my search results contain articles that are of no use to me?
• Have I used all possible search terms to find all relevant articles?
• Do my search results contain relevant articles?

• Take a good look at your building blocks: is your search strategy too broad or do you have to approach your topic from a different angle?
• Use more specific search terms. For example, go to the MeSH database and see if a narrower term is more suited to your research question.
• Add an extra aspect and combine this building block using AND.
• Use a filter: for example for publication date or study type. See the Filters tab for more information.

• Try using a broader MeSH term.
• Add more search terms to your building blocks. Read for example abstracts of relevant articles to see if these contain other words.
• Use truncation to get more variations in phrasing/spelling to the right of the truncation symbol.
• Remove an aspect of your search. For example, the Outcome in a PICO strategy is often not convenient, because you do not know what will be the outcome.
• Remove search filter(s)
• Check the spelling of your terms (e.g. British or American spelling), synonyms, singular/plural. Check the brackets and the right position of AND's and OR's.

A search will always produce some irrelevant articles (noise). If you find that you have far too many irrelevant results,

• Take another look at your research question to see whether it has been broken down into the right aspects.
• Remove search terms that result in irrelevant hits.
• Check to see whether your Boolean operators (AND, OR) and brackets are constructed properly.
• Check to see whether you have used [MeSH] or [tiab] and the truncating asterisk properly.

• Check the titles and abstracts of relevant articles to see if these contain keywords not included in your building blocks. Also look at the MeSH terms for these articles.
• Consult another database such as Embase, PsycINFO or CINAHL to find other relevant articles that are not part of the PubMed index. You might also wish to browse the keyword indexes of these databases for inspiration. The Embase Emtree database, for example, often contains a greater number of synonyms.

Retrieving key articles is a convenient way to see if your search has produced reliable results. Use an author search or PMID (the unique PubMed number for each article) to find the article and combine this with your end set using the Boolean operator AND in the Advanced Search Builder.

If the results do not contain the article, check the keywords in the title and abstract and compare these to your building blocks. Also check the article’s associated MeSH terms.