Identifying Core Search Concepts
While a research question (e.g., PICO question) and inclusion criteria may be very specific, to build a comprehensive search for research evidence, it is a good idea to structure the search more broadly. A comprehensive search often just includes two or three core concepts (for a PICO question, these concepts are often the population and intervention of interest).
For example, imagine your PICO question is:
"For adolescents with type II diabetes (P) does the use of telehealth consultations (I) compared to in-person consultations (C) improve blood sugar control (O)?
To build a comprehensive search for evidence that answers this question, you could use diabetes and telehealth as your core search concepts, looking for research that brings both these concepts together. You would then retrieve a broad set of results that could potentially fit with your research question, and the other elements of the research question (e.g., the comparison and outcome) would then be used as screening criteria to help you identify studies for inclusion.
With 2-3 core concepts in mind begin to identify keywords that describe these concepts. After doing some initial scoping searches, pay close attention to the articles you find that appear to be most relevant - what language do they use to describe these concepts? In order to perform a comprehensive search, it is important to keep track of these synonyms.
As the search evolves, you will begin to notice that some terms are more fruitful than others; building a concept table is a useful way to track search terms, enabling you to document which terms work well and which terms should be eliminated from the strategy.
| Concept 1: diabetes | Concept 2: telehealth | |
|---|---|---|
|
Keywords/Synonyms |
type II diabetes type 2 diabetes diabetes mellitus |
telehealth mobile app cell phone |
The database will only retrieve a record if that record contain your specified search terms, but for some abstract concepts, there is a wide variety of ways that the concept can be expressed or even implied. By using terms to try to describe those ideas in your searches, you may end up excluding relevant articles simply because their records don't include the exact word that you entered. Some words may even introduce bias into your search.
In particular, try to avoid using:
These are words that words that describe the relationship between two topics (e.g., compare, contrast, correlation, impact, effect, causation, relationship). Instead, consider just using keywords for the concepts you're trying to relate, and then you can evaluation the relationship that exists.
These are value-laden words that imply that imply something is better or worse than something else (e.g., best, worst, pro, con, advantages, disadvantages, benefits, harms). These words can unnecessarily introduce bias into your search; it's better to allow the database to retrieve records that meet your topic criteria overall, and then apply your own judgment as the researcher.
Remember that both synonyms (same meaning) and antonyms (opposite meaning) terms can be useful to include to describe a concept. As you're collecting synonyms, consider including terms for your concepts of interest that may be used more commonly outside of the US (e.g., physiotherapy/physical therapy; barrister/attorney; flat/apartment).
Also, it may be useful to include antonyms for your topics of interest. For example, if you were interested in student retention, not only would you want to include synonymous terms like student persistence, or graduation, you'd also want include terms for the opposite phenomenon (e.g., student dropouts, student attrition)
It's a good idea to search for both abbreviations and the full phrases for an important concept. For example, a researcher interested in cognitive behavioral therapy would also want to use the term CBT.
If the abbreviation for your topic is very common, and you find the abbreviation is generating lots of false hits, it may be useful to pair it with other qualifying words (e.g., intellectual property OR IP law).
Additionally, consider if your term of interest might have an alternative spelling or in British or Australian English (e.g., labor vs. labour). See Oxford International English School's webpage for some common spelling variations to look out for.
Searching for the core concepts as keywords or text words is often a good place to start your search. A text word search will retrieve records where the search term appears anywhere in the database record for that article (eg., the authors’ names, the publication title, or the abstract).
However, you will begin to notice that database records are tagged with a controlled vocabulary to designate the subjects discussed in the full articles. These are referred to as index terms or subject headings. They are considered a controlled vocabulary because they are drawn from standardized thesauri that establish definitions and preferred usages; these terms may differ from the natural language text words you initially choose to describe a concept.
For example, indexing in MEDLINE (PubMed) employs Medical Subject Headings (MeSH). Upon being added to the database, an article about type II diabetes would be tagged with the MeSH term “Diabetes Mellitus, Type 2”.
To identify database-specific subject headings to use in your search, you can look up terms directly in the database's subject thesaurus, or notice what subject headings are used in records for relevant literature.
Most article databases allow you to use field tags to search specifically for records that have been tagged with a controlled subject heading, as opposed to performing a keyword or text word search.
In contrast to a text word search (where the query returns citations if the term appears anywhere in the record), a subject heading search (using a field tag) is much more targeted, only returning results where the search term appears as the subject of that article. Consider the following example from PubMed:
A comprehensive search is usually achieved by searching for a combination of subject headings and text words, so a strategic searcher would be well-advised to keep track of both.
| Search Terms | Concept 1: diabetes | Concept 2: telehealth |
|---|---|---|
|
Keywords |
type II diabetes type 2 diabetes diabetes mellitus |
telehealth mobile app cell phone |
|
Subject Headings (PubMed MeSH Terms) |
“Diabetes Mellitus, Type 2” | Telemedicine “Remote Consultation” Smartphone “Mobile Applications” “Text Messaging” |
Also see:
Boolean operators (also called connectors) allow you to specify how you would like a database search platform to handle the terms of your search.
Connecting search terms with AND tells the database to return records only if all those terms appear in the record. The operator AND is typically used to connect conceptually distinct terms (such as the P and the I from a PICO question)
Connecting additional terms with AND creates a narrower search, as there will be a smaller number of records that contain the required terms.
| Search Query: | Number of Results Returned: |
|---|---|
| type 2 diabetes | 166,000 |
| type 2 diabetes AND telehealth | 556 |
| type II diabetes AND telehealth AND blood glucose | 195 |
Connecting search terms with OR tells the database to return records if any of the given terms appear in the record. Connecting synonyms and alternate terms with OR expands the search results ("OR retrieves MORE").
| Search Query: | Number of Results Returned: |
|---|---|
| telehealth | 6,100 |
| telehealth OR telemedicine | 33,000 |
| telehealth OR telemedicine OR mobile application | 51,000 |
Connecting search terms with NOT tells the database to exclude any records that contain the specified term. Adding an excluded term with NOT will create a narrower, more specific search.
| Search Query: | Number of Results Returned: |
|---|---|
| telemedicine | 36,000 |
| telemedicine NOT video | 33,000 |
*Caution: use the NOT operator sparingly and carefully; you could accidentally exclude records that are actually relevant to your search. For instance, in the example above, even if you aren't interested in seeing articles that talk about video-based telemedicine interventions, if you restrict that word completely, you will miss articles just because they contain the word 'video' somewhere in the abstract.
A complex search can be built by using the AND and OR operators together. Separate concepts can be connected with AND, while synonyms for those concepts can be connected with OR. It is important to note that synonymous terms connected with OR should be nested within parentheses, so that the database understands to keep those terms together as a set.
| Search Query: | Number of Results Returned: |
|---|---|
| type 2 diabetes | 166,000 |
| type 2 diabetes AND telehealth | 556 |
| (type 2 diabetes OR T2D) AND (telehealth OR telemedicine OR mobile application) | 700 |
Using specific punctuation as wildcard or truncation characters allows you to build a more flexible search by accounting for unknown characters, multiple spellings or various endings.
Many databases use an asterisk * as the wildcard punctuation. For instance, if you're interested in allergies, a search for that exact word will only return records with the word allergies. But searching for allerg* will be much broader, capturing records that contain the words allergy OR allergies OR allergen OR allergic etc.
| Search Query: | Number of Results Returned: |
|---|---|
| allergies | 21,000 |
| allerg* | 33,000 |
Different database platforms allow for different truncation and wildcard punctuation. See these resources for the specifics related to common platforms:
Using specific punctuation or commands to look for exact phrases allows you to build a more specific search.
Many databases use quotation marks as the phrase searching punctuation. For instance, if you were interested in finding records that talk about the glycemic index, you could locate records with those specific words, in that exact order (rather than the words separately).
| Search Query: | Number of Results Returned: |
|---|---|
| glycemic index | 16,000 |
| "glycemic index" | 6,000 |
Different database platforms may allow for different phrase searching functions. See these resources for the specifics related to common platforms:
Most databases allow you to include field codes, (also known as field tags, or search tags) with your search terms to specify in which field in the record you would like the database to look for that term.
Different database platforms have different codes and structures for field tags, but some of the commonly available field tags are title, abstract, author and subject heading.
Some databases also have codes that allow you to search in multiple fields at once. For instance, PubMed's TW (Text Word) tag allows PubMed to search for the given term in fields like the title, abstract, author supplied keywords, and subject headings.
For more specific information about available field tags by database, see:
If you don't specify a field, you are performing an "unqualified search", and the database will default to searching multiple fields. Different databases have different default search fields, but title, abstract and subject heading fields are commonly included. For specific information about what fields are included in unqualified searches by database, see:
![Example search result showing that searching with a MeSH term yields fewer results. Two columns labeled "Query" and "Results" have two entries below: "Search: 'Glucose Intolerace'[TW]" with 16,100 results and "Search: 'Glucose Intolerance']Mesh]" with 8,633 results.](https://lh3.googleusercontent.com/c4gmNqtXV7rVMkj06bc1YtrbgmsBr2TlQJNyzdqDUHHzPj9Envh2aUZNrETzIyi9FEsH8pHX_T7sPxdiiRsmD3rDQQceCz2N_SUA36inKl0YOfgK0RT3TVDicR_nzYV_1xo3iSgT)