Systematic Reviews - Research Guide

After defining your review question using a framework such as PICO to identify the main concepts, consider alternative terms or synonyms for each search concept.

It is important to spend time developing your list of alternative terms or synonyms, as searches for systematic reviews aim to be as extensive as possible to ensure all relevant studies are retrieved. Authors may use different words to describe a search concept, so you may miss appropriate studies if you do not include all relevant alternative terms in the search.

How do I come up with more alternative terms or synonyms?

• Identify subject-specific terms for your topic in textbooks, lecture notes, subject-specific encyclopaedias and dictionaries
• Perform a simple preliminary search on your main concepts - scan titles, abstracts and keywords for alternative terms
• Consider:
  • different spellings (e.g. paediatric / pediatric)
  • different terminology (e.g. physiotherapy / physical therapy)
  • medical terminology vs natural language (e.g. hypertension / high blood pressure)
  • medication brand names vs generic names (e.g. Panadol / paracetamol / acetaminophen)
  • acronyms (e.g. MCI / Mild Cognitive Impairment)
  • abbreviations
  • plural vs singular word forms (e.g. child / children)
  • hyphenated words (e.g. nonsurgical / non-surgical)
• Subject headings should also be searched as keywords - your alternative terms list should grow once you identify relevant subject headings

Tip: Alphabetise search words to easily identify duplication

Check that each search word adds value to your search  i.e. that it returns relevant results.

It is important to thoroughly develop synonyms and to understand what terms you want to capture, before using search techniques such as truncation and wildcards.

Subject headings are a set of predetermined terms that describe specific concepts.

Not every database utilises subject headings, and databases may have their own subject headings that do not translate to other databases. Subject headings are also called controlled vocabularies, index terms, or a thesaurus. 

Examples of the subject headings used by some databases are summarised in the table below:

Advantages of using subject headings:

• controlled vocabulary
• consistently applied – all articles with the same subject heading will have that as a key concept​
• assigned by an indexer/librarian after reading the content of the article​
• compiled into a thesaurus that can be searched.

Combining subject headings with free-text terms will make your search more comprehensive.

To identify subject headings for your search concepts:

• search the subject heading thesaurus of the database to identify relevant terms for each search concept
• choose the most relevant term 
• copy and paste terms into your search planner or other documentation
• add to both concepts a subject heading that combines the two concepts
• combine two separate subject headings if required to appropriately describe a concept
• look at the subject headings used for relevant articles on your topic - do you want to use these as well?
• add your chosen subject headings as free-text terms (keywords).

For more information on how to use MeSH:

• Medical Subject Headings (MeSH) in MEDLINE/PubMed: A Tutorial
• How to use the medical subject headings (MeSH)

Once you have developed your search terms (free-text/keywords and subject headings), apply search techniques to create your search strategy. This section will cover the practical application of techniques; for details on the various techniques, see the Search techniques box below.

There are three approaches to structuring a systematic search:

• line by line (each search term on its own line)
• block by block (each search concept [e.g. searchable PICO element] on its own line)
• single line (all search terms and concepts combined into one line).

For more detail and examples, see the blog post Structure of Search Strategies for Systematic Reviews: Line by Line versus Block by Block versus Single-Line.

In practice, a combination of all three might be used depending on the database being searched, the stage of search development, the searcher's expertise, etc. 

The block by block search structuring approach will be used in this section to demonstrate building a systematic review search strategy, with each search concept (PICO element) on its own line.

Check the Search Help or Search Tips of the platform for its specific details of applying search techniques, and remember to document your search regardless of the method you choose.

Use OR for a search concept, its related synonyms/variations, and relevant subject headings

Search concept 1 OR synonym OR synonym OR subject heading

Add search field codes for free-text terms (database specific)

If using PubMed and searching for your search terms in the Title/Abstract field:

Search concept 1[tiab] OR synonym[tiab] OR synonym[tiab] OR subject heading[MeSH]

Add quotation marks around phrases (i.e. multiple word terms)

PubMed search syntax:

"Search concept 1"[tiab] OR synonym[tiab] OR synonym[tiab] OR subject heading[MeSH]

Apply truncation and wildcards where appropriate and with care

PubMed search syntax:

"Search concept 1"[tiab] OR synonym*[tiab] OR subject heading[MeSH]

Add brackets around each search concept

PubMed search syntax:

("Search concept 1"[tiab] OR synonym*[tiab] OR subject heading[MeSH])

Use AND between each search concept

PubMed search syntax:

("Search concept 1"[tiab] OR synonym*[tiab] OR subject heading[MeSH])

AND

("Search concept 2"[tiab] OR "synonym 1"[tiab] OR "synonym 2"[tiab] OR subject heading[MeSH])

If searching for a particular study design, add appropriate study design search filter

PubMed syntax: RCT filter (Cochrane highly sensitive search strategies for identifying randomized trials in PubMed)

("Search concept 1"[tiab] OR synonym*[tiab] OR subject heading[MeSH])

AND

("Search concept 2"[tiab] OR "synonym 1"[tiab] OR "synonym 2"[tiab] OR subject heading[MeSH])

AND

(randomized controlled trial[pt] OR controlled clinical trial[pt] OR randomized[tiab] OR placebo[tiab] OR drug therapy[sh] OR randomly[tiab] OR trial[tiab] OR groups[tiab] NOT (animals [mh] NOT humans [mh]))

Developing a systematic search is an iterative process; you will need to review and refine it before reaching your final, reportable search.

After constructing your search, review the search itself and consider:

• Have you used the correct syntax (e.g. field codes, truncation, etc) for this database?
• Are there any spelling errors or typos?
• Have you been comprehensive with your search terms?
• If the database uses subject headings: have you included relevant subject headings and also added them as keywords?
• Have you combined your terms and concepts correctly with OR & AND?

Next, run the search and review the results. When reviewing your search results, consider whether:

• they can answer the review question
• the results are relevant
• too many or too few rwesults are being retrieved
• key articles are being found.

Adjust your search if it is not retrieving many relevant results, if it is retrieving too many or too few results and if it not is not finding key articles.

The PRESS Checklist details items to check when reviewing search strategies and is a useful reference for where to identify possible adjustments to your search strategy.

For additional help, Ask our Librarians to review your systematic review search strategy; please attach your completed PICO Worksheet, Search Planner, and copy and paste your search. 

The final search strategies for each database must be reported in a systematic review. It is essential to document your search process to ensure correct reporting of your final search strategies.

For each database search you conduct, you should record:

• the date the search was run
• the database searched
• the name of the platform used to search the database (e.g. Ovid)
• your search strategy - copy and paste exactly as run
• any limits used, such as date, language, study design
• any published search filters used (as designed or modified), and filter citation details 
• the number of studies identified.

It may be useful to save your search strategies in the databases or platforms you use (where possible) to refer to later. Resources such as Medline, Scopus, Web of Science and Cochrane provide this feature.

Recording the details of your searches will allow you to correctly report your search strategies in accordance with PRISMA and PRISMA-S. 

Item 7 in the PRISMA 2020 Checklist pertains to reporting of searching for systematic reviews:

Present the full search strategies for all databases, registers, and websites, including any filters and limits used.

The PRISMA-S Extension for searching goes into further detail about preferred reporting of search strategies for systematic reviews.

The Cochrane Handbook recommends (and Cochrane Reviews require) search strategies for databases be copied and pasted in full and exactly as run, with search set numbers and the total number of records retrieved by each. 

The base search strategy will remain the same for each database; however, you will need to make adjustments when you search a new database. Different databases use different search platforms and may have different search options – check the ‘Search help’ or a similar page of a database before you begin searching.

Check:

• Subject headings

Does the database use subject headings? If so, are the subject headings different to those of the previous database? For example, the previous database may have used MeSH, while the new database uses the APA Thesaurus. 

What is the syntax when using subject headings? For example, PubMed and MEDLINE (Ovid) both use MeSH, but the syntax used when searching is different.

• Syntax for truncation, wildcards, phrase searching, Boolean and proximity operators

Some databases may use different characters for truncation and wildcards or different commands for proximity operators. It’s important to be aware of a database's search commands.

• Field codes

Different databases have different codes for field searching (the fields available to search free-text terms in).

• Limits

Databases may have different types of limits available.

Tools for translating searches

The following tools may assist with translating searches from one database to another. However, an understanding of the differences between database syntax is still required as no tool is perfect. 

The Polyglot Search Translator 

Polyglot is designed to translate search strings across databases to aid with running searches for systematic reviews. It does not translate subject headings across databases, so it is important to understand how databases work before using Polyglot.

MEDLINE Transpose

MEDLINE Transpose is a tool designed to convert search syntax between the PubMed and Ovid MEDLINE interfaces.

CC BY-NC-ND

Boolean operators connect phrases or keywords to improve your search results.
These determine the relationship between concepts:

When searching OR is used to combine search terms describing one search concept.

AND is used to combine different search concepts.

NOT should be used with care when searching as it could remove relevant results. It is recommended to manually screen by applying eligibility criteria instead. 

Phrase searching instructs the database to search for an exact match on your search words. It is especially useful when individual search words are common.

Most databases require your search terms to be in double quotation marks for phrase searching.

For example, "public health" finds the words as a phrase and in that order.

Tip: turn off smart quotes (curly quotation marks) in your Word document, as the formatting corrupts search strings - for example, "public health" rather than “public health”

Proximity limiters, also called adjacency, are used to search for words within a specified number of words from each other.

Using adj3 between two search words would tell the database to find the words in any order and with up to two words between them, i.e. word 2 can be up to three words away from word 1.

For example, cognitive adj3 therapy will find the words cognitive and therapy within two words of each other in either direction. This would retrieve cognitive behavioural therapy, cognitive behavioral therapy, cognitive processing therapy, cognitive group therapy, etc.

Check the help guide to each database before using the proximity limiter, as the syntax of this operator varies a lot between databases.

Truncation is used to find various word endings and will retrieve variations of a word, such as plurals and nouns, in a database.
The symbol commonly used is the asterisk *
For example, develop* will retrieve develop, development, developmental, developmentally, developments, etc.

Use truncation carefully.
For example, cat* will retrieve catalogue, catastrophic, catatonic, catfish, catholic …​

You can use a truncation symbol in conjunction with phrase searching. 
For example,  "tandem repeat*" will return the same results for ("tandem repeat" OR "tandem repeats" OR "tandem repeating")

A wildcard replaces a character or characters in a word and is used to retrieve different spellings (e.g. American vs British word forms).

Common wildcard symbols include ? and #

For example: behavio?r will retrieve behaviour and behavior; organi?ation will retrieve: organisation and organization; p?ediatric, will search paediatric or pediatric.

Wildcard and truncation symbols vary from database to database.
Check the database help section to identify the correct wildcard and truncation symbols.

Nesting is a technique that groups search words. Nesting is useful to organise your search concepts into the order you want the database to process them, such as combining synonyms for each search concept.

Most database use brackets or parentheses for nesting ( )

Database records have information listed in certain fields, such as author, title, abstract, keywords, etc. You can direct a database to retrieve articles with your search words in specific fields. This increases the precision of your search and reduces the number of irrelevant results.

To direct the database to find your search words in specific fields, you should apply the appropriate field codes to your search words using the syntax required by that database. Usually you should use a field code that will capture at least 'Title' and 'Abstract.' 

For example: 

• To search PubMed for Title and Abstract, add [tiab] to the end of each search term
• To search MEDLINE (Ovid) for Title, Abstract, and Keyword Heading Word, add .ti,kf,ti after each search term

Some platforms (search interfaces) allow you to select search fields within the search builder via a dropdown menu. 

Databases allow you to limit your search results by, for example, demographic details such as age and sex, and by year of publication.

Care should be taken when using limits, such as language, date and study design. In order to prevent bias, it is preferable to include all available studies in the initial search results.

It is recommended to refine your search strategy by using a designed search filter, rather than database limits.

When appraising your results, the criteria from your review protocol can be used to exclude publications that are not relevant to your systematic review.

Filters are pre-designed search strategies to find specific types of records, such as a particular study design. You are recommended to use a published validated search filter rather than applying a database filter.

For example, if you were wishing to find randomised controlled trials, a validated search filter such as the Cochrane highly sensitive search strategies for identifying randomised trials in PubMed would be advisable over applying a database filter. 

Please note that any published validated search filters that you use in your review must be cited.

The ISSG Search Filters Resource lists known filters and provides checklists of assessing the quality of search filters. 

For more information on search filters, see section 3.6 in the Technical Supplement to Chapter 4: Searching for and selecting studies (Cochrane Handbook).