Tip #23: PubMed's [tiab] vs. [tw]

Search field tags are a quick way to specify the exact fields you would like to search when using keywords. For example, I can tell PubMed to search for my term only in the title field or only in the MeSH Terms field.

You can view a list of available search fields from the drop down menu in PubMed's Advanced Search Builder. Specifying field tags for keywords is also a good practice in PubMed because keywords searched without field tags will be translated through Automatic Term Mapping which may reduce the precision of your results. Here are a few sample searches in my search history, including #4 which wasn't limited to a field (notice how it was translated in the expanded search details):

The fields that are included in the [title/abstract] fields are pretty self-explanatory (words and numbers included in a citation's title, collection title, abstract, other abstract and author keywords), but what is actually included in the [Text Word] (can also be searched as [tw]) fields?

According to the PubMed Help documentation, the Text Word fields include "all words and numbers in the title, abstract, other abstract, MeSH terms, MeSH Subheadings, Publication Types, Substance Names, Personal Name as Subject, Corporate Author, Secondary Source, Comment/Correction Notes, and Other Terms (see Other Term [OT]) typically non-MeSH subject terms (keywords), including NASA Space Flight Mission, assigned by an organization other than NLM."

You can see in the screenshot above, that switching from [tiab] to [tw] dramatically increased the number of results. Does that mean that we should always use the Text Word [tw] field? Not necessarily! 

Note that in the list of fields included in [tw], the MeSH terms and MeSH subheadings are included as keywords, so this means that depending on my topic, I may be pulling in many irrelevant MeSH terms and subheadings into my search. You will need to test your individual terms (using the search example above - #3 NOT #2) to determine if [tw] is better than [tiab]. 

If my research question is focused on breastfeeding and postpartum depression, and I use the synonym "nursing" in my breastfeeding concept, I won't likely want to see articles on nurses and depression (unless it is somehow also related to breastfeeding).

Let's compare these two versions of my (very simplified) search (click on the searches to check them out):

#2 ("Breast Feeding"[Mesh] OR breastfed[tw] OR "breast milk"[tw] OR nurse[tw] OR nursing[tw]) AND "Depression, Postpartum"[Mesh]

#1 ("Breast Feeding"[Mesh] OR breastfed[tw] OR "breast milk"[tw] OR nurse[tiab] OR nursing[tiab]) AND "Depression, Postpartum"[Mesh]

Let's review the additional 371 citations that are added to my results when using the [tw] field (#2 NOT #1 in the screenshot below):

If I were doing this for an actual evidence synthesis project, I would have a much more comprehensive breastfeeding concept search, so you will find a few relevant results in the set of 371, but overall you will see that most of the results primarily focus on the field of nursing or the role of nurses in postpartum depression care (MeSH terms or subheadings that have "nursing" in them). These are not relevant articles, so for this research question, using the less inclusive [tiab] for both (nurse[tiab] OR nursing[tiab]) will improve the precision of my results. This tip won't eliminate articles that talk about the field of nursing because we will still see results that use the keywords "nurse" or "nursing" somewhere in the title or abstract, but using this tip will help eliminate many of the irrelevant results that have been indexed with a nurse-related MeSH term or subheading.

A few irrelevant results are posted below to highlight the "nursing" as keyword in the MeSH terms and subheadings:

Using the [tw] tag as a default for all your keywords will help improve the sensitivity of the search, but be careful with any terms that are also terms in MeSH/subheadings that may not be relevant. We recommend testing the [tw] vs. the [tiab] in those cases.