The asterisk * [Shift and 8 on a standard UK keyboard] is used by many databases - Scopus, Web of Science, all databases on the Ovid platform (Medline, Embase, PyscInfo etc), on the EBSCO platform (ERIC, CINAHL etc), on the NHS HDAS databases and the PROQUEST platform as a truncation command. You may also see $ used as a truncation symbol in databases on the Ovid platform but please note that $ is used as a different command in other databases.
Truncation instructs the database that when you are searching for a free-text keyword search that it should search for the root of the word you have typed in and then retrieve any alternate endings.
This is excellent for searching for plurals without having to type out both the singular and plural in your search, but will find also find any other alternative endings (some of which may not be relevant to your topic).
A keyword search for dentist* would retrieve any article which has the word dentist or dentists or dentistry somewhere in the title, abstract or other field. A keyword search for therap* would retrieve any articles where the word therapy or therapies appeared, but would also retrieve articles which included the word therapeutic (likely to be relevant) and also therapist(s) (perhaps less likely to be relevant if you were initially wanting to search for therapies).
|Search term using truncation||Keywords searched for|
In Web of Science the asterisk (*) represents any group of characters, including no character and can also be used within a word, e.g. s*food matches seafood and soyfood.
In those databases which use Subject Headings it is recommended that you search initially for your term in full without using truncation because if truncation is used then the database may fail to suggest appropriate subject headings even if one or more exist. Once you have located a relevant subject heading or headings for the concept then you can start to type in your keywords for that concept using truncation. If you receive an error message in Ovid that the search cannot be mapped to a subject heading (sometimes happens when using truncation) then simply untick the map box to subject heading before repeating the search. Remember to re-tick the box when you come to search for a new concept and you wish to see what subject headings are available.
PubMed Truncation: truncation, or finding all terms that begin with a given string of text, is generally not a recommended search technique for PubMed as truncation bypasses Automatic Term Mapping [to Subject Headings] and automatic explosion.
Subheadings are used to further describe a particular aspect of a subject heading in databases such as Medline, Embase and Cinahl.
An example might be (for Medline) 'thromboembolism/prevention and control' or the two-letter subheading abbreviation -- thromboembolism/pc. 'Thromboembolism' is the subject heading and 'prevention and control' one of its subheadings (this will retrieve a subset of the results which have been tagged with the thromboembolism subject heading, limiting the results to only those articles discussing the prevention and control aspect).
However, it is possible in these databases to search for the subheading independent of any subject heading. This allows you to retrieve all articles which have been tagged with a particular subheading but without having to specify what subject headings it is attached to. A list of subheadings and abbreviations is available.
This can be very useful as seen in the example below where the search aims to retrieve any article in Medline tagged with the floating subheadings of Adverse Effects (ae), Complications (co) or Drug Effects (de) [search line 43] . In the Ovid databases to search using floating subheadings you can type the two letter subheading abbreviation followed by .fs. Keyword searches for safety, side effects, toxicity, adverse effects, etc were also undertaken in the search strategy [search lines 44-45] and the results brought together using OR [search line 46] before being combined with other elements of the search, e.g. the specific patient population and drug intervention which was being reviewed [for the full search strategy see the Appendix of the Cochrane Review linked below].
43. (ae or co or de).fs.
44. (safe or safety or (side adj1 effect*) or (undesirable adj1 effect*) or (treatment adj1 emergent) or tolerability or tolerance or tolerate or toxicity or toxic or adrs or adr or harm or harms or harmful or complication* or risk or risks or (unintended adj1 event*) or (unintended adj1 effect*)).ti,ab.
45. (adverse adj2 (effect or effects or reaction or reactions or event or events or outcome or outcomes)).ti,ab.
46. 43 or 44 or 45
Extract of Medline (Ovid) search strategy from: Storebø, O., Ramstad, E., Krogh, H., Nilausen, T., Skoog, M., Holmskov, M., Rosendal, S., Groth, C., Magnusson, F.L., Moreira-Maia, C.R., Gillies, D., Buch Rasmussen, K., Gauci, D., Zwi, M., Kirubakaran, R., Forsbøl, B., Simonsen, E., Gluud, C. (2015), 'Methylphenidate for children and adolescents with attention deficit hyperactivity disorder (ADHD)', Cochrane Database of Systematic Reviews, Issue 11. Art. No.: CD009885. DOI: 10.1002/14651858.CD009885.pub2
Performing a high quality electronic search of information resources ensures the accuracy and completeness of the evidence used in your review. However, errors have been found in search strategies of systematic reviews (even Cochrane ones!). PRESS EBC is an evidence-based checklist that has been developed to guide and inform the peer review of search strategies for database searching and can also be used to check your own search strategy.
The Yale MeSH Analyzer allows you to enter PMIDs* for different articles (up to 20 at a time) and generates a MeSH analysis grid presenting the ways these articles are indexed in the MEDLINE database (i.e. which subject headings have been assigned to each article) in an easy-to-scan tabular format.
This can provide you with a means to generate useful MeSH (Medline Subject Headings) from articles on your topic which you know are relevant. You can also use it to help identify the problems in your search strategy as you can easily scan the grid and identify appropriate MesH terms, term variants, indexing consistency, and the reasons why some articles are retrieved and others are not. This inevitably leads to fresh iterations of the search strategy to include missing important terms.
*PMIDs are a unique identifying number which are assigned to each article in the Medline database (appearing in the records of articles in both PubMed and Medline on OvidSP).
Proximity or adjacency searching using keywords allows you to search for two words or phrases that appear within a set number of words of each other (in any order). This is less precise than a phrase search (see the box on this page) but ensures it is more likely that the words/phrases will be related than a simple AND search. Different databases require you to type in different operators/commands in order to undertake a proximity search. Check the help pages for the database platform you are searching if the commands are not listed below.
OvidSP platform databases and HDAS (NHS) databases, e.g. Medline, Embase, PsycInfo
The ADJ operators finds two terms next to each other in the specified order. The ADJ1 operators finds two terms next to each other in any order. The ADJ2 operator finds terms in any order and with one word (or none) between them. The ADJ3 operator finds terms in any order with two words (or fewer) between them and so on.
ADJn - where n represents the number of words that could appear between your keywords/phrase, e.g. middle ear adj4 infect* would search for the phrase "middle ear" within 3 words (or fewer) of the word infection, infectious, etc
middle ear infection
infected middle ear
infection of the middle ear
middle-ear derived infections, and so on.
CINAHL (EBSCO platform)
Use Nn - where n represents the number of words that could appear between your keywords/phrase, e.g. "middle ear" N3 infect*
Note that whilst the N proximity searching will find terms regardless of the order in which they appear, the Within operator (W) will find only those articles where the terms appear in the order they were entered. For example, typing kidney W3 failure will retrieve articles which include the phrases 'kidney failure'/'kidney transplant failure'/'kidney graft failure' but not 'failure of the kidneys'.
Web of Science
Use NEAR/n - where n represents the number of words that could appear between your keywords/phrase, e.g. "middle ear" NEAR/3 infect*
Using W/n restricts to n words between the two words; the word order is not set, e.g. pain W/5 morphine will retrieve 'pain controlled using morphine' as well as 'morphine to control pain'
Pre/n restricts to n words between the two words, but the word order is as set, e.g. newborn PRE/3 screening will retrieve 'newborn hearing screening' but not 'screening of the newborn'
Subject heading search
It is important when searching databases which have a thesaurus and which tag articles with subject headings (Medline, Embase, PsycInfo, Cinahl, etc) that your search strategy combines (with OR) both relevant subject headings and keyword/free-text searches on a particular concept. For full details see the Drawing up your search strategy tab.
In databases on the Ovid platform a subject heading search is shown with a / after the term:
If the subject heading has been exploded to include narrower more specific terms then this will show with exp before it:
exp drug hypersensitivity/
If the subject heading has been focussed (limiting to articles where the selected subject heading is a major concept of the article) then this will show as:
exp *drug hypersensitivity/.
In CINAHL on the EBSCO platform you will see MH used to indicate that a subject heading has been searched; a + sign to indicate the subject heading has been exploded; and MM is used to indicate that the subject heading has been limited to results where this is a focussed/major concept of the article.
(MH "Wound Care") - subject heading search
(MH "Wound Care+") - exploded subject heading search
(MM "Wound Care+") - exploded and focusssed/major concept search.
The databases on the NICE HDAS platform use the same symbols as the Ovid platform databases above.
The default keyword search on databases on the Ovid platform is a multi-purpose search across several fields including title, abstract, original title, name of substance word, subject heading word, keyword heading word, protocol supplementary concept word, rare disease supplementary concept word, unique identifier. This is shown as .mp search
In Cinahl the default keyword search is of the Title, Abstract and Subject headings fields.
In the NICE HDAS databases the default keyword search is of Title and Abstract.
It is possible to select a more specific keyword search on those databases that have a broader multi-purpose search as default - see the 'Searching in the Title and Abstract fields' box on this page.
In Cinahl on the EBSCO platform you need to search for your terms in the Title and in the Abstract and OR these together:
As well as truncation other wildcards are available to use on some databases. These wildcards differ from database to database so it is worth checking (via their help pages) if you are looking for a particular function on a database platform.
OvidSP databases (e.g. Medline, Embase, PsycINFO)
Web of Science
The wildcard is represented by a question mark ? or a hash [pound] sign #.
# wildcard replaces 0 or 1 character.
For example, type colo#r to find all citations containing color or colour.
When developing your search strategy you may wish to search using specific phrases rather than simply undertaking a search on individual keywords combined with OR. For example searching for "physical therapy" as a phrase in the title or abstract of articles will limit your search significantly compared to searching for 'physical OR therapy'.
Most database platforms use double quotation marks "..." to ensure that keywords are searched as a phrase. (NB phrase searching is the default in databases on the Ovid platform).
Examples of phrase searching include:
"cognitive behavioural therapy"
Phrase searching can sometimes be too restrictive so do bear in mind that some databases also allow you to use proximity searching (see the box on this page).
The screenshots below show the difference in the number of results when using phrase searching compared to OR keyword searching.
Web of Science
Use NOT in a search to narrow your search and exclude keywords or subject headings from your search.
1) Your combined search terms
2) exp animals/ not humans.sh.
3) 1 not 2
Search line 2 limits results to animal only studies and search line 3 then excludes these from the results when combined with your search terms.
In the search above the results have been limited to articles which have been tagged with the focussed subject heading of Pregnancy (and narrower more specific terms as it has been exploded) and then animal only studies have been excluded. This double use of NOT (in search line 2 and 3) as opposed to just using the human limit (a tick box under the limits option) is to ensure articles which may have been tagged with both animal and human are returned as well as human only studies.
NOT searching can also be used to exclude particular publication types, e.g. letters or editorials. See the Using Filters tab for more information.
Use NOT with care as used incorrectly it may exclude results that you are interested in. For example, if you were interested in retrieving research on the use of antidepressants in treating depression and excluded the terms CBT OR cognitive behavioural therapy using NOT then you would also exclude any results which directly compared the two methods. If you do wish to use NOT to exclude specific keywords or subject headings then consider adapting the Cochrane method for excluding animal only studies using the double use of NOT.