A search strategy is the documented, reproducible plan a systematic review uses to find every study that might answer its question. It specifies the databases searched, the exact terms used in each, and how those terms are combined, so that another researcher could rerun it and retrieve the same results. The search is the single most consequential methodological step: a flawed search misses eligible studies, and no amount of careful screening or analysis afterward can recover what was never retrieved.
The aim is sensitivity over precision. In ordinary web searching you want a short list of the best hits; in a systematic review you would rather wade through some irrelevant records than miss a relevant one, because a missed study is invisible and unrecoverable.
From concept blocks to a search string
A good search is built, not typed. You take each element of the PICO question and turn it into a concept block, a list of every reasonable way authors might express that idea. The population "older adults" becomes elderly, geriatric, aged, and so on; the intervention "exercise" becomes physical activity, aerobic training, resistance training, and more. Within each block you gather both free-text keywords and the database's controlled vocabulary (such as MeSH terms in PubMed), because relying on either alone leaves gaps.
| Step | What you do | Boolean logic |
|---|---|---|
| Within a concept | Combine all synonyms for one idea | Join with OR |
| Across concepts | Require every concept to appear | Join blocks with AND |
| Refine | Remove a clearly irrelevant sense | Use NOT sparingly |
The ordering matters: you widen each concept with OR, then narrow the result by joining concepts with AND. Used carelessly, NOT can silently delete relevant records, so most experienced searchers avoid it except for unambiguous cases.
Boolean operators are the grammar of the search
The three core Boolean operators, AND, OR, and NOT, are what turn a list of words into a precise instruction. OR broadens by capturing any synonym; AND narrows by demanding that separate concepts co-occur; NOT excludes, at the risk of also excluding relevant studies that merely mention the excluded term. Around these sit field tags, truncation (a wildcard that captures word endings, such as therap* for therapy, therapies, and therapeutic), and phrase searching. Mastering this small grammar is most of what separates a reproducible search from a lucky one.
Choosing and translating across databases
No single database covers all the literature, so a systematic review searches several. Health reviews commonly combine PubMed or MEDLINE, Embase, and the Cochrane CENTRAL register, adding subject databases such as CINAHL or PsycINFO where relevant. Each database uses different controlled vocabulary and syntax, so the strategy must be translated, not copied, from one to the next. Beyond the databases, a thorough search reaches into grey literature sources and uses citation searching to catch studies the database queries miss.
Deciding how many databases is enough is a judgement, not a fixed number, and it should be justified in the protocol against the breadth of the question and the eligibility criteria you have set.
Documenting the search for reproducibility
A search you cannot reproduce is not a systematic search. Best practice, formalised in the PRISMA search-reporting guidance, is to record the full strategy for every database, including the exact terms, the date run, and the number of records retrieved. Our guide to transparent search reporting covers what to include, and the record counts you capture become the first numbers in your flow diagram. The whole search sits within the wider systematic review workflow, and once screening is complete you can turn those counts into a publication-ready diagram with our free PRISMA flow diagram tool.
Frequently Asked Questions
What is a Boolean search operator?
A Boolean search operator is a connecting word, most commonly AND, OR, or NOT, that tells a database how to combine search terms. OR broadens a search by retrieving records containing any of the listed terms, AND narrows it by requiring all terms to appear, and NOT excludes records containing a given term. They are the basic logic that turns a list of keywords into a precise query.
What are the three Boolean search terms?
The three core Boolean operators are AND, OR, and NOT. AND requires that all connected concepts appear in a record and narrows the results; OR requires that any of the connected synonyms appear and broadens the results; NOT removes records containing a specified term. A systematic review search typically uses OR within each concept and AND between concepts.
What are the most common Boolean operators?
By far the most common are AND and OR, which do almost all the work in a systematic review search: OR groups synonyms within a concept, and AND combines the separate concepts. NOT is used much less often because it can accidentally exclude relevant studies that happen to mention the excluded term, so careful searchers reserve it for unambiguous cases.
How do you write a search strategy for a systematic review?
Start from your research question, break it into concepts, and for each concept list every relevant keyword and controlled-vocabulary term. Combine the synonyms within a concept using OR, then combine the concepts using AND. Translate this structure into each database's syntax, search grey literature and reference lists as well, and document the full strategy and record counts so the search can be reproduced.