Tip #10: Testing for Key Article Inclusion in PubMed

This tip was submitted by Carrie Price (Towson University)

Do you want to test your search for key article inclusion? Perhaps the review team supplied you with a list of key articles during the reference interview. Or, maybe you did some searching together and identified relevant articles. Typically, these key articles are articles that should appear in the search, and can be a useful building block for term harvesting. 

Once the key articles have been identified, grab the PubMed ID number (PMID) from the item record and paste them into a document. For an example, I'll use a set of articles on pulmonary rehabilitation in cystic fibrosis patients. I've chosen five. Give each PMID a field tag of [uid], "unique identifier."

33618051 [uid] OR 30827470 [uid] OR 29045949 [uid] OR 27320420 [uid] OR 26522923 [uid]

Run them to make sure you get the correct results. It's really easy to miss part of a PMID!

Now, keep this key article set in your search documentation while you develop the search.
When you have completed your initial PubMed search strategy, run it AND'd with the key article set.

Oh no! I missed four out of five articles here. Only one was included.

 

To test why some didn't appear, run the original key article set NOT'd with the test search. 

 

Examine the missed articles to see what terms should be revised. In this case, I had four articles to review and examine.
Revise the search and try again. I added some additional terminology that should capture the key articles.

 

When all the key articles appear, you're in good shape! All five came through for me this time.

 

 In rare cases, there may be a situation where a key article will not be returned with an appropriate search for the topic. In this case, contact the review team to let them know, and to ask how to proceed.

Key article testing is a very useful technique to ensure that the search is capturing relevant articles.

The downside to this method is that the articles must be indexed in PubMed, although it is possible to use this same approach in other platforms if you can uniquely identify articles.