Spend some time planning your search. This will involve developing a search strategy.
Before starting, it’s a good idea to do some preliminary or background searches using PubMed or Google Scholar to identify relevant studies that answer your question. These studies can be examined to help locate relevant subject headings or MeSH (the latter specific to MEDLINE/PubMed) as well as synonyms you may not have thought to use.
Text mining tools can help with identifying relevant subject headings (MeSH) and keywords. The following use PubMed data.
When developing your strategy, we recommend using a concept or logic grid to help break the question down into its key concepts. It's generally recommended to have at least two concepts, but more can be included depending on your topic or question. Using a concept grid offers another benefit: it aids in brainstorming and recording various synonyms for each concept. This is crucial when conducting a comprehensive search, as it ensures you consider all relevant terms.
Consider singular/plural forms, variations in spelling (British vs American), acronyms and relevant subject headings.
Here's an example PICO question which has been translated into a searchable format:
In patients with cardiovascular disease, how effective is the use of Omega-3 fish oil supplementation (compared to no treatment)?
| Concept 1 | Concept 2 | Concept 3 | Concept 4 |
|
KEYWORDS cardiovascular diseases CVD heart diseases cardiac events |
KEYWORDS omega-3 fish oils cod liver oil |
KEYWORDS N/A |
KEYWORDS N/A |
|
SUBJECT HEADINGS (MeSH) Cardiovascular Diseases/ Heart Diseases/ |
SUBJECT HEADINGS (MeSH) Fatty Acids, Omega-3/ Fish Oils/ Cod Liver Oil/ |
SUBJECT HEADINGS (MeSH) N/A |
SUBJECT HEADINGS (MeSH) N/A |
Boolean operators (AKA combining or linking terms) can be thought of as the building blocks to a good search strategy. Understanding how AND/OR/NOT function will enable you to develop better searches.
| Operator | Explanation and example |
| AND | AND will find records that contain both terms, which reduces the number of results. For example, "cardiovascular diseases" AND "fish oils" |
| OR | OR will find records that contain either term, which increases the number of results (OR = mORe!). For example, "fish oils" OR "omega-3" |
| NOT |
NOT will find records that contain your first term but exclude the second term, which will reduce the number of results, for example: ("sleep hygiene" NOT apnoea) - will find records only referring to sleep hygiene and exclude any that mention apnoea. Use NOT with caution! In most searches the NOT operator is not required. However, it can be used for testing and refining your search strategy. |
In addition to using Boolean operators, there are numerous search techniques that can be utilised to search more efficiently and effectively.
Always check the Help function of the library database you are using as the search techniques may differ.
| Name of technique | Explanation |
Ovid platform databases: MEDLINE, Embase, Emcare, PsycINFO. Example |
| Phrase searching |
Phrase searching forces a library database to search for those terms as an exact phrase.
|
Represented by double quotation (") marks around two or more words. For example, "sleep disorders" or "sleep hygiene" will be searched as exact phrases. Ovid databases will search exact phrases by default unless the phrase includes Ovid stopwords (e.g., and, as, for etc.). Use double quotes if this is the case. |
|
Truncation |
Truncation can be used to find different endings of a term, including plural and singular forms. |
Represented by an asterisk symbol * or a dollar sign $ Unlimited right-hand truncation searches for variations on a word that are formed with different suffixes. For example, a search for gene* finds occurrences of gene, genes, genetics, and generation. |
|
Wildcards |
Wildcards and can be used to replace zero, one or more characters usually in the middle of a word. Useful to pick up variations between British and American spelling. |
Represented by a question mark symbol ? or a hashtag symbol # ? character stands for zero or one alphanumeric characters within a word or at the end of a word (but not between two words). For example, p?ediatric finds occurrences of pediatric and paediatric. # character stands for one alphanumeric character within a word or at the end of a word (but not between two words). For example, organi#ation finds occurrences of organization and organisation. |
| Proximity/Adjacency |
The proximity operator searches for terms near each other and in either direction.
|
To use adjacency, separate your search terms with the ADJ operator and a number from 1 to 99. nature ADJ3 therapy - will find nature-based therapy, nature therapy, nature of therapy, therapy in nature etc. |
| Field codes (AKA dot command searching) | Tells the database where to search for keywords in an article's record. For example, specify only to search in the title and abstract fields. |
.ti = title .ab = abstract .tw = title/abstract .kw = author-provided keyword exact .kf = word in author provided keyword .pt = publication type .mp = multipurpose (default keyword) .id = key concepts, including author keywords |
|
Parentheses (AKA nesting or bracketed search) |
Used to clarify relationships between terms and specifies the order of how a search should be run as parentheses are processed first. | (fish or cod) adj3 (nutrit* or oil* or supplement* or capsule*).ti,ab,kf. |
For a comprehensive search, it's really important to use a combination of subject headings and keywords for each concept.
While you might be familiar with keyword searching, which is something we all do everyday when searching Google or other online tools, you might be less familiar with subject headings. Simply put, these are like tags or labels that help describe what an article is about.
Subject databases such as MEDLINE, Embase, PsycINFO etc. use subject headings from a controlled and hierarchically-organised vocabulary to organise and describe the documents they index (list) in their databases.
Searching with an appropriate subject heading will help find documents on that specific topic regardless of the terminology used by the authors.
It's important to note that each database has its own thesaurus (see table for details). When searching across different databases you will need to modify your search to ensure you are using the most relevant and appropriate headings. Refer to our page on Test & translate for tips on how to translate headings.
| Database | Controlled Vocabulary/Thesaurus |
| MEDLINE/PubMed/Cochrane Library | MeSH or Medical Subject Headings |
| Embase | Emtree |
| Emcare | Emtree |
| PsycINFO | Thesaurus of Psychological Index Terms or APA Thesaurus |
Searches for systematic reviews and other evidence syntheses should aim for high sensitivity. This approach usually yields a larger number of results which require screening for inclusion.
What's the difference between sensitivity and precision?
"Sensitivity is defined as the number of relevant reports identified divided by the total number of relevant reports in existence."
"Precision is defined as the number of relevant reports identified divided by the total number of reports identified."
Reference: Cochrane handbook, Chapter 4, 4.4.3 Sensitivity versus precision