You want to know:
Does the use of a mobile application improve medication adherence in older adults?
Next, craft the search using FOUR separate searches:
Then, build your search one concept at a time and then use 'search history; to combine separate concepts. This allows you to test different combinations of the concepts to find the best set of articles for your needs, maybe (S1 AND S3) would give you a better set of results compared to (S2 AND S3).
Something to consider as you build your search is phrase searching. Phrase searching limits your results to articles containing an exact phrase, as opposed to keywords scattered throughout the title, abstract, author, or journal fields. Adding double quotes around a set of words will turn them into a phrase.
1. Your search (not a phrase): patient compliance (~133,491 results)
Article title: Regulatory landscape with U.S. patient requirements and Clinical Trial Diversity expectations
Author: Melissa Vozza (CDCS Regulatory Compliance & Quality Assurance, Fortrea, USA.)
2. Your search (a phrase): "patient compliance" (~71,277 results)
Article title: Patient Compliance With Bowel Preparation for Elective Colorectal Surgery
The next step is to add synonyms for each concept, using phrase searching when appropriate. Synonyms are important because a relevant article might use different words than your original concept terms. Think back to our question from the first step:
Does the use of a mobile application improve medication adherence in older adults?
If you search in PubMed for "mobile application" AND "medication adherence" AND "older adults" [click to run the search] you will get relevant results, but there are likely many more relevant articles that use slightly different words for each concept. To find those additional results, you will need to include synonyms. For example:
Boolean operators such as AND and OR are used to connect synonyms and concepts. Imagine that all of the colored dots or dot clusters below represent articles about different topics. To get the database to return only relevant results, you will use different combinations of our relevant keywords and ANDs and ORs.
For example:
Controlled vocabulary is a standardized set of terms used by databases to describe and categorize articles based on their content. It works similarly to a unique hashtag for social media posts (e.g. #APHA2024 for the American Public Health Association 2024 meeting) or navigating through an online store's categories to get to a desired item (Clothing > Women's Clothing > Tops > Sweaters) rather than using the search box.
Using controlled vocabulary terms can return more relevant and specific results, and can help overcome variance in terminology, but you have to know where to find it and how to add it to your search. Different databases have their own controlled vocabularies:
Not all databases use controlled vocabulary; for example, Google Scholar and Scopus do not have their own controlled vocabulary. Searching with only controlled vocabulary may limit your results to just the articles that have been indexed - you may miss new articles or articles that are older than the controlled vocabulary term.
Here is a sample search in PubMed using just a few relevant MeSH terms for our research question:
Does the use of a mobile application improve medication adherence in older adults?
The search process is iterative! After running an initial search and skimming through the results, you may discover new keywords to include or see that one of your keywords is pulling in irrelevant results. You may also realize that you need to break out an additional concept from your research question or find that your concepts need to be combined in a different way.
It is helpful to keep track of all of the search terms that you have used during your search process in order to create the most comprehensive search and to allow the search to be run again at a later time. If you plan to publish the results of your literature review, you may also need to include a section on your search methodology.
Choosing the "right" search for the job depends on how comprehensive the search needs to be and how much time you have to review results. Here are a few sample iterations for our research question [click on the links to view search results]:
1. "Mobile Applications"[MeSH] AND "Medication Adherence"[MeSH] AND "Aged"[MeSH]