A search in PubMed follows the general process described on the Search Strategy page of this guide. Briefly, those steps include: topic, searchable questions, in/exclusion criteria, terms and operators, iterations and refinements, documentation, and citation management.
Use the following link to access PubMed. Our proxied link will grant you access to our library holdings and full text articles:
On this page, we'll discuss how to break your question down into searchable concepts, come up with search terms, execute the search, and document/manage the citations. Use the worksheet below as you work through this guide!
By now, you've established your topic: nursing orientation programs in acute care settings for new hires. And you've created your searchable question: For newly hired nurses in a hospital system (P), do structured orientation programs (I) improve retention and performance (O) when compared to the current informal staff onboarding process (C)?
The next step is to break that question down into searchable concepts. You won't always search every part of your PICO question. For example:
In this example, your searchable concepts may include:
Once you've broken your question into the searchable concepts, the next step is to come up with search terms. In PubMed, this means identifying both keywords and subject headings (Need a refresher on subject headings? Check out the Subject Headings page of the guide!).
Start by turning each concept into a list of searchable words or phrases. Then, brainstorm synonyms and alternate spellings (e.g., color vs. colour). Following the new hire orientation example, you might include onboarding, new hire training, and transition to practice in your search. Keep an eye on your search results: keywords like orientation can have multiple meanings and may pull in irrelevant results. If that happens, try using quotation marks for an exact phrase (e.g., "staff orientation") or swap in a synonym.
Note: PubMed automatically applies truncation and automatic term mapping to your keywords unless you use quotation marks. While you can add an asterisk to manually truncate your terms, it is often unnecessary.
Once you've tackled keywords, it's considered best practice to also include subject headings in your search. PubMed uses MeSH (Medical Subject Headings) terms, which are standardized by the National Library of Medicine.
You can search for MeSH terms using the MeSH Database. Just enter a keyword and browse the results for relevant subject headings. If there's only one match, you'll be taken directly to the term's information page after you hit search. If there are multiple, you'll see a list of terms to choose from, with checkboxes to the left and links to the term's entry.
If your keyword doesn't return results, try synonyms, using broader/narrower terms, or rephrasing the concept. If you still don't find a match, it's possible that there isn't a MeSH term for that concept. In that case, be extra diligent about coming up with robust keywords!
To add a term to your search:
For more help with MeSH Terms, visit the MeSH help guide.
Once you've identified your concepts and selected your terms, you're ready to execute the search. Unlike many databases, PubMed only provides one search line at a time - even on the advanced search page. However, it does allow you to build complex searches using the "Add to query box" function, or by combining previous searches from your history.
One of the most effective ways to connect multiple concepts in a search is by combining search lines in your history using the "Advanced Search" feature. This approach is similar to databases that offer multi-line search interfaces.
Once you've followed these steps, your query box might look something like this:
After running your search, you'll land on the results page. The initial results may be unwieldy, but you can bring them under control by applying filters, adjusting the view, and exporting/saving them where you need them.
Depending on your search terms, you may find that your results aren't quite what you expected. The problem may be obvious with some terms, but with others, you may need to investigate what's going on. As may be the case with the keyword orientation, you may get unrelated results (spatial orientation, etc.). Revisit the search history and see how PubMed is breaking down that term.
After review, you find that orientation is mapping to the MeSH terms for orientation and spatial orientation, neither of which are relevant to your search. At this point, you may choose whether to eliminate orientation, make it into an exact phrase (e.g., "staff orientation" with quotes), or use the NOT Boolean operator. You might also find that retention wasn't the best keyword to leave on it's own either! Think about the different kinds of retention you might see in healthcare, and think of ways you can account for this in your choice of keywords.
Formal documentation is often unnecessary for quick searches, but is necessary for many searches that contribute to a research project. Many journals require researchers to document and present their search strategy in their writing. Properly documenting your strategy supports reproducibility and transparency, as well as makes it easier to revise your search for future iterations.
Begin formally documenting your process from the beginning instead of trying to remember and replicate later (spoiler alert: you will not be able to replicate what you did later). It's also helpful to write down your reasoning for the terms you choose, filters you apply, etc. If you are conducting a systematic review, follow the applicable standards in your discipline. At a minimum, you should document the following items from your search:
Some databases, like PubMed, allow you to export your search history as a .csv file. This file contains all of the information you'll need for documentation, so it's a good trick to become familiar with. Be sure that all aspects of the search carry over to the .csv file including filters/limiters - sometimes, those do not export with the search history and must be manually added to the documentation.
When you've finished a search session, you export your search history from the database you're working in. You make sure that the history includes date/time and filters/limiters - if not, you add them to the exported document. If the database does not allow you to export your history, you copy and paste it into a document and add all additional relevant information like the date/time, limits/filters.
After exporting, you also document your reasoning for the choices you made throughout the search. You explained the study type filters you chose (because you wanted high level evidence), eliminating orientation as a keyword on its own (because it was giving too many false hits), etc.
Once you've found the articles, you'll need to keep track of all of them. A great way to do this is to create a folder, collection, or project (different names, same function) in your database of choice to save articles. As you work through your search, click check boxes next to relevant articles to save for later. Once you've completed the search, you can review your saved articles to determine which full-texts you'd like to retrieve. You can also export your results into a file compatible with citation management software, a .csv file, or even a Word document.
For more information and guidance on citation management, visit the Citations guide!
You are doing your search in PubMed. You've sent relevant articles to your clipboard and are now ready to export the citations to EndNote. You open the clipboard, select the articles of interest, and send them to a citation manager. You also want the full texts of some of the articles, so you also track those down using the full-text links in each article record, or you request an interlibrary loan of the best articles.