Overview
Learning to conduct an effective and comprehensive search of the literature takes time and practice. There are many resources available to you through UBC Library (including custom workshops and consultation with subject librarians).
See Cochrane Handbook Chapter 4: Searching for and selecting studies
Remember that searching:
- Is an iterative process;
- often requires re-evaluation and testing, from adding or changing keywords and the ways they relate to each other, to choosing new resources to search, and even rewording or refining the research question;
- is a logic puzzle that requires an balance of precision and sensitivity.
Recommended viewing:
Therapeutics Initiative (TI) Methods Series: Systematic review searching: an overview of best practices & real world experiences
This is a recorded webinar presented by Rebecca Rishar, MSLS – Senior Medical Librarian, ECRI.
Presentation that provides a comprehensive overview of the systematic review search process.
- Current best practices in systematic searching
- translating clinical questions into PICO search statements
- database selection
- PRESS peer review
- PRISMA-S search reporting
- citation management
- brief overview of available text-mining tools
Search Tools
Several search techniques are common to a variety of licensed databases - subject headings, truncation, Boolean operators, and limits. Depending on your topic, there may also be search filters available to apply to one or more databases. See these Medline Ovid tutorials for an overview of different techniques.
Techniques to aid the process include using a table in Word or Excel to list the concepts in your research question. This way you can keep track of potentially relevant subject headings and keywords that you discover when exploring scope notes and thesaurus trees. Another is to use the save search history in most databases to record your search and to set up email alerts for when new articles are found on your topic.
Many databases include a system of subject headings (which may also be called descriptors, or controlled vocabulary). These terms are added to articles by human indexers to make it easier to search for all the articles on a particular concept. Most other topical, or subject, databases have a system of subject headings, which can usually be found in the thesaurus section of a database. In Medline these are known as MeSH (Medical Subject Headings), in EMBASE as Emtree, and in CINAHL they are Headings.
For a thorough literature search, you should search each concept with both keywords and subject headings.
Some databases and grey literature sources only allow searching by keyword (aka free text or natural language). Useful techniques include truncation, wildcards, phrase searching, and proximity searching.
|
Database |
Truncation |
Wildcard: 0 or 1 character |
Wildcard: Exactly 1 character |
Phrase |
Proximity |
|---|---|---|---|---|---|
|
OVID databases
|
* or $ or : Examples: pharm$ will find: pharmacy, pharmacist, pharmaceutical…. Child* will find: children, childbirth, child-centred, childhood… |
? Examples: flavo?r P?ediatric |
# Example: wom#n |
Quotation marks only needed if there's a word like "and," "or" or "use" in your phrase: Examples: “Sensitivity and Specificity” “Substance use disorder” |
adjn (adj=adjacent and "n" is the number of words)
Example: environment* adj3 health will find environment, environmental etc. within 3 words of health. |
|
* Note: truncation stops automatic mapping to MeSH |
* Note: there must be 4 characters before first wildcard. Wildcard stops automatic mapping to MeSH |
* Note: there must be 4 characters before first wildcard. Wildcard stops automatic mapping to MeSH |
“your phrase” Note: phrase searching stops automatic mapping to MeSH, and does not always find results |
"search terms"[Field:~n] Available for title and abstract only. Example: "nursing education"[Title/Abstract:~2]
|
|
|
EBSCO databases
|
*
|
# Examples: flavo#r P#ediatric |
? Example: wom?n |
“your phrase” |
Nn or Wn (N= Near, W= Within and "n" is the number of words) Example: |
|
|
* Web of Science allows left-sided truncation as well as right-sided. Example: *statin will find: atorvastatin, simvastatin, pravastatin… |
$ for exactly 0-1 characters * for 0-multiple characters
|
? |
“your phrase” |
NEAR/n You can specify n number of words; or if you just type NEAR, the default range is within 15 words. |
| Scopus |
* Scopus automatically searches for plurals and applies stemming |
? Scopus automatically includes spelling variants |
? Scopus automatically includes spelling variants |
"your phrase" loose phrase - searches for words in same field (title, abstract, or keyword), but not necessarily as an exact phrase. {exact phrase} Use curly brackets to look only for exact phrase. Note that hyphens count - eg, {COVID 19} and {COVID-19} will find different results. Can't use truncation or wildcards with exact phrase searching. |
W/n (words in any order within n words of each other) PRE/n (looks for words only in the order they are entered) |
|
Proquest databases *note: consider changing the drop-down next to the search box to "NOFT" instead of "anywhere" when searching these databases |
* |
* for up to 5 characters in middle of word |
? | "your phrase" |
NEAR/n (if you don't specify n, default is 4) PRE/n (looks for words only in the order they are entered) |
| CABI Digital Library |
|
? |
Use proximity: reproductive near/1 "aging" searching phrase in quotes may find undesired results, eg "reproductive aging" will also find "reproductive age" |
near/n ("n" is the number of words) |
Boolean operators are useful for combining subject headings and keywords.
- Example: Stroke AND balance finds articles with both these concepts
- Example: Stroke OR CVA finds articles that include either or both concepts
NOT is another operator which will find one concept while excluding another. Use with caution because you may exclude relevant articles this way. See the document below for tips.
Most databases include various limiters. These usually qualify human characteristics such as gender, ethnicity or age or publication characteristics such as language, publication date, study design, or type of publication.
Consult with a librarian about using limits in a systematic review search - it's easy to lose relevant articles when adding limits. Usually a search filter (aka hedge) is a better choice than using the default limits in a database. A search filter includes a mix of keywords and limiters.
For further discussion, see: Cochrane Handbook Chapter 4.4.5: Language, data and document format restrictions
Search Filters/Search Hedges
Search filters are pre-tested strategies used to find evidence in databases. They are often used to locate a specific type of study, such as a randomized controlled trial or a qualitative study. Some search filters have been validated against a test set of records.
- ISSG Search Filters ResourceIn addition to search filters for study designs, this resource includes search filters for topics such as adverse effects and quality of life. It also provides information and guidance on how to critically evaluate filters.
- Ovid expert searchesIncludes subset filters from PubMed (for example, systematic reviews[sb]), adjusted for use in Ovid, along with other topical search filters.
- Cochrane RCT and human studies filtersTechnical Supplement PDF
- University of Alberta Health Sciences Search FiltersIncludes filters for Indigenous peoples, mobile health and animal studies.
- Exclusion filters to selectively remove records (e.g. animal studies)Methods for excluding studies indexed as animal only, child only and adult only prepared by McGill University.
Search Automation Tools
- Ovid Search History LauncherThis tool allows you to enter a multi-line Ovid search history and run it in one of UBC's subscribed databases. Note: you must be on a UBC network computer or on VPN for the search to work.
- SR-Accelerator: Polyglot Search TranslatorEnter a PubMed or Ovid MEDLINE search and convert search for use in Web of Science or CINAHL. Note: the tool only changes the commands and formatting, it does not translate subject headings from one database to another, so further editing of the search is needed. Also: the version of PsycINFO in the tool is Ovid PsycINFO - UBC has EBSCO PsycINFO.
- Yale MeSH AnalyzerSpecific to health and biomedicine. Enter the PubMed ID numbers of known relevant articles and create an HTML or Excel file indicating the MeSH headings used to index those articles.
Updating Searches
If a lot of time passes between running your search and completion of your review, you may need to run a search update. The link below has some helpful tips for updating searches in different databases.
Phrase Searching in Ovid Databases
In general, quotation marks are not needed in order to search for words as a phrase in Ovid databases. However, one exception is when a phrase contains the word "use," which is a command in Ovid. You'll need to put a phrase containing use, or the word use by itself, in quotes in order to avoid an error message.
Also, if a phrase in Ovid includes one of these words: and, as, for, from, is, of, that, the, this, to, was, were
Ovid will ignore the word in the phrase. This may cause you to find many more results than intended, in particular if you're searching for an exact phrase. Solution - use truncation on that word in your phrase:
"treatment as* prevention"
"female to* male"
See the link below for more details.
How do I know when I'm done searching?
It is recommended to consult with a subject librarian. Each question is nuanced and there are disciplinary differences.
For general guidelines, please see Cochrane Handbook Chapter 4.4.11 When to stop searching
Highlights: "At a basic level, investigation is needed as to whether a strategy is performing adequately. One simple test is to check whether the search is finding the publications that have been recommended as key publications or that have been included in other similar reviews (EUnetHTA JA3WP6B2-2 Authoring Team 2019)."
*See: Mining seed papers and search development template created by Alix Hayden & Zahra Premji