Searching the Cochrane Library: CENTRAL vs. CDSR, and Building a Trials-Register Search

“The Cochrane Library” is not one database — it is a platform bundling several distinct content types, and “searching the Cochrane Library” means something different depending on which of those you actually want. The two that cause the most confusion are the Cochrane Central Register of Controlled Trials (CENTRAL), a register of trial records used as raw material for a systematic review’s search, and the Cochrane Database of Systematic Reviews (CDSR), the finished reviews themselves. Confusing the two — or assuming a Cochrane Library search box searches everything with one uniform syntax — produces a search strategy that either misses a documentation requirement a peer reviewer will flag, or misses trials a sensitivity-maximizing search is supposed to catch.

What’s actually inside the Cochrane Library

The Cochrane Library, published on Wiley’s platform at cochranelibrary.com, bundles several distinct content types under one search interface:

• Cochrane Database of Systematic Reviews (CDSR) — completed Cochrane Reviews and their protocols, produced under Cochrane Handbook-compliant methodology and a Cochrane Review Group editorial process. This is what most people mean when they say “a Cochrane review.”
• Cochrane Central Register of Controlled Trials (CENTRAL) — a register of trial report records, not reviews. It exists to be searched during a systematic review, not read as a finished output.
• Cochrane Protocols — registered, published plans for a review that is still in progress, searchable and citable before the review itself is complete.
• Cochrane Clinical Answers (CCA) — short, point-of-care summaries of what a Cochrane review found, written for clinical decision-making rather than for a search-strategy appendix.

A default Cochrane Library search box query runs across some or all of these at once depending on which databases are selected. If your goal is a reproducible, reportable search — for a systematic review, a grant methods section, or a protocol appendix — the first decision is which of these you actually need, not just what keywords to type.

CENTRAL vs. CDSR: two different kinds of content

These are built differently and serve opposite ends of the same process:

• CENTRAL is populated by systematic searching of MEDLINE, Embase, CINAHL Plus, the Australasian Medical Index, KoreaMed, and ClinicalTrials.gov, plus WHO International Clinical Trials Registry Platform (ICTRP) records, conference proceedings, and records from other trials registers — including reports in languages other than English and records that aren’t independently available in the major bibliographic databases. It is updated monthly.
• CDSR is the output: the completed systematic reviews that CENTRAL searches (among other sources) feed into.

Practically, this means: CENTRAL feeds a review’s search; CDSR publishes what the review concluded. Running a CDSR-only search to check “has anyone already reviewed this” and running a CENTRAL search to find primary trial records for a new review are different tasks with different search strategies — a title/topic search built for the first will miss most of what a sensitive trials-register search for the second needs to catch, and vice versa.

Searching CENTRAL and CDSR separately

The Cochrane Library’s advanced search lets you restrict which content types a search runs against — CDSR only, CENTRAL only, or a combination — rather than always querying everything at once. For a systematic review’s trials-register search specifically, restrict to CENTRAL: a combined search across CDSR and CENTRAL together mixes trial records in with completed reviews and commentary, which is the wrong shape of result set for a PRISMA-style trial-identification search and complicates deduplication against your MEDLINE/Embase results later.

The search syntax: MeSH, field codes, and proximity

The Cochrane Library’s search runs on controlled-vocabulary and free-text syntax in the same family as other major health databases, not a proprietary one-off:

• MeSH descriptor searching — written as MeSH descriptor: [Diabetes Mellitus] explode all trees. “Explode” pulls in the narrower terms beneath that heading in the MeSH tree automatically, the same exploded-hierarchy logic covered in more depth on this site’s MeSH Terms in PubMed guide — the concept transfers directly even though the bracket syntax is specific to this platform.
• Field qualifiers — appending :ti, :ab, or :kw to a term restricts it to the title, abstract, or keyword field respectively (for example diabetes:ti,ab,kw), instead of the broader default search scope.
• Truncation — an asterisk (*) at the end of a word stem catches variant endings, e.g. therap* for therapy/therapies/therapeutic.
• Proximity operators — NEAR/n finds two terms within n words of each other in either order; NEXT/n requires that order. These matter for a sensitive trials-register search because they catch phrase variants (“randomized controlled trial” vs. “trial, randomized controlled”) that a straight phrase search would miss.
• Boolean operators — AND, OR, and NOT combine terms and field-restricted blocks, with parentheses controlling precedence exactly as in Scopus’s query language or PubMed’s.

Building a trials-register search the Handbook expects to see documented

The Cochrane Handbook’s guidance for a systematic review’s search strategy is explicit that sensitivity, not precision, is the goal: combine controlled vocabulary (MeSH, or Embase’s Emtree when searching Embase directly) with free-text terms that account for spelling variants, synonyms, acronyms, and truncation, and explode MeSH terms rather than searching only the specific heading — a narrow, precision-tuned search is the wrong tool here, because a missed eligible trial is a much more serious defect in a systematic review search than a handful of extra records to screen out.

Two things trip up a search strategy built only in the Cochrane Library search box:

• CENTRAL does not replace searching MEDLINE and Embase directly. Even though CENTRAL draws from both, it is not a complete substitute for them — coverage lag and indexing differences mean a review’s search methods section documents CENTRAL, MEDLINE, and Embase (via their own native platforms, e.g. Ovid or PubMed) as three separate, individually reported searches, not one Cochrane Library search standing in for all three.
• A trials-register search often needs to go beyond CENTRAL too. For genuinely comprehensive trial identification — including ongoing and unpublished trials — reviewers commonly search ClinicalTrials.gov and the WHO ICTRP directly as well, since CENTRAL’s monthly update cycle and its dependence on registries being indexed elsewhere first means the newest registrations can lag behind a direct registry search.

Documenting all of this — which databases were searched, which syntax translation was used for each, and the date each search was run — is exactly the kind of reproducible search-strategy record a PRISMA-style methods section and a peer reviewer both expect to see, the same reproducibility standard covered on this site’s Systematic Review vs. Scoping Review: PRISMA Guide.

Frequently asked questions

Is CENTRAL the same thing as ClinicalTrials.gov?

No. ClinicalTrials.gov is one of the source registries CENTRAL draws records from (alongside WHO ICTRP and others), not a synonym for it. CENTRAL is a secondary compilation aggregating and deduplicating trial records found across MEDLINE, Embase, ClinicalTrials.gov, and other registries into one searchable register.

If I’ve already searched CENTRAL, do I still need to search MEDLINE and Embase separately?

Yes, for a systematic review search intended to be comprehensive. CENTRAL is built from those sources but is not a complete real-time mirror of them, and Cochrane Handbook guidance treats CENTRAL, MEDLINE, and Embase as three separate searches to be run and reported individually, not one search that substitutes for the other two.

What’s the difference between a Cochrane Protocol and a Cochrane Review?

A Cochrane Protocol is the registered, published plan for a review — background, objectives, and planned methods — made public before the review is conducted. The Cochrane Review is the completed report of that same project once the search, screening, and synthesis are finished. Both are searchable in CDSR, but they represent different stages of the same piece of work, not different topics.

Does a Cochrane Library search automatically search everything at once?

Only if you let it. The advanced search interface lets you select which content type(s) — CDSR, CENTRAL, or others — a given search runs against. For a systematic review’s trials-register search, restricting to CENTRAL specifically (rather than searching CDSR and CENTRAL together) keeps the result set to trial records instead of mixing in completed reviews.

Related CASRAI pages

• Cochrane Handbook for Systematic Reviews of Interventions — the methodology standard CENTRAL and CDSR searches are built to satisfy.
• MeSH Terms in PubMed — the exploded-hierarchy concept behind MeSH descriptor searching, in more depth.
• PICO Framework — turning a review question into the search concepts a CENTRAL search string is built from.
• Systematic Review vs. Scoping Review: PRISMA Guide — where a documented search strategy fits into PRISMA reporting.
• WHO ICTRP — one of CENTRAL’s source registries, and a direct search target in its own right.
• ClinicalTrials.gov — the other major trial registry a comprehensive trials-register search typically checks directly.
• Covidence vs. Rayyan vs. DistillerSR — where CENTRAL/CDSR search results get screened once the search itself is done.