Registering a Systematic Review on PROSPERO

PROSPERO registration is quick and free, but it has a strict scope and a hard deadline. Once data extraction starts, the chance to register has gone.

Guide · 11 min read · Directive Publications Editorial Team

What registration does, and what it does not

PROSPERO is an international register of systematic reviews run by the Centre for Reviews and Dissemination (CRD) at the University of York and funded by the UK's National Institute for Health and Care Research. It is free to register and free to search.

A PROSPERO record is a public, dated statement of what your team planned before it saw the data from included studies. Editors and peer reviewers can compare it with your manuscript, and other teams can see the question is already being worked on.

Three things a record is not:

• Peer review. PROSPERO does not check record content beyond built-in automated checks, and publishing a record does not endorse your methods.
• Your full protocol. You can upload a protocol PDF, but the record fields still have to stand on their own.
• A formal publication. It is a registration that later outputs link back to.

The value lies in the comparison. If your published review changes a primary outcome or the synthesis plan, the record shows readers what changed and your manuscript has to explain why. That is the safeguard against selective reporting, and it is why editors look closely at the registration date.

Check PROSPERO eligibility before you start the record

PROSPERO's eligibility page is short and strict. The main register takes systematic reviews with outcomes of direct relevance to human health. A companion registry on the same platform takes reviews of animal studies that relate to human health, and a review that includes both human and animal studies needs two separate records that refer to each other.

PROSPERO's own example shows the health-outcome test: a review of bullying and mental health is eligible, but a review of bullying and educational attainment is not.

If your aim is to map what evidence exists rather than answer a defined question and appraise the included studies, you probably have a scoping review; PROSPERO's help text notes that scoping reviews do not assess the quality of included studies. If it is a scoping review, your write-up follows the PRISMA extension for scoping reviews rather than PRISMA 2020. Other excluded reviews keep their usual guideline: a systematic review of sporting performance is still reported against PRISMA 2020 even though PROSPERO will not take it. For any of them, PROSPERO points to other registries such as the Open Science Framework.

Every field must be completed in English, although an uploaded protocol or search strategy can be in any language.

Timing: the window between a final protocol and data extraction

Not too early. Your protocol should be finalised, or very nearly, before you submit. Every edit after registration creates a new public version, and a major revision needs the whole team's approval again, so a record with many versions is slower to change and harder for others to follow. PROSPERO also asks you not to register speculatively: have a funding award confirmed, or a firm commitment of author time for unfunded work, before you register.

Not too late. Reviews must be registered before data extraction has started, or before any individual participant data have been received. PROSPERO defines the review start date as any point after the protocol is complete but before data extraction begins.

A workable rule: submit the record as soon as the protocol is signed off, ideally before screening starts. That leaves a buffer for co-author approvals, so nobody is tempted to extract data while the record is still waiting.

The form asks which review stages have started and which are complete: pilot work, formal searching, screening, data extraction, risk of bias assessment and synthesis. Answer exactly. PROSPERO regards knowingly false information about start dates or the stage of the review as research misconduct, and it may withdraw records and replace them with a withdrawal notice.

If you have already missed the window, do not register. PROSPERO's registration terms say you must not create a registration for a completed review even if a journal asks for a registration number, and its information for journals tells editors to check that the registration date is not after the submission date. State in the manuscript that the review was not registered, and say whether a protocol was prepared and where it can be read.

How to register a systematic review, step by step

1. Confirm the review is in scope. Use the table above. If the team is split between systematic review and scoping review, settle it now: the answer decides where you register and which reporting guideline you follow.
2. Finalise the protocol. The PRISMA-P 2015 checklist is a sound structure, and it includes items on registration and on documenting amendments. Agree the review question, eligibility criteria, outcomes and synthesis plan with every co-author now, since later method changes need everyone's approval again, and confirm the funding or author time is in place.
3. Prepare accounts and team details. The person creating the record needs a PROSPERO account and an ORCID; our guide to ORCID and affiliations covers setting one up properly. Collect each team member's name, institutional email and conflict of interest details. Only the record owner can edit the record, so choose that person deliberately. The guarantor, usually the senior team member, approves the content and is notified of substantial changes.
4. Complete every field. Expect to describe the review question, sources searched with dates and restrictions, population, intervention and comparator, study designs, outcomes, screening and extraction methods, risk of bias assessment, synthesis plan, timelines, funding and conflicts, within word limits. PROSPERO does not accept records that refer readers to an uploaded protocol instead of completing the fields, and it asks for a synthesis description specific to your review, not generic text. The form also warns that screening or data extraction by a single person is not recommended.
5. Check similar registered reviews. PROSPERO uses your title, review question and tags to find similar registered reviews. It does not stop you proceeding, but it asks how each one differs, for example in population, outcomes or methods, and shows your answers in the published record. If a near-identical review is already under way, contacting that team may serve you better than starting a parallel one.
6. Submit and collect approvals. On submission, every other team member receives an email link to approve publication. If anyone does not approve, the record returns to editing and everyone must approve the revised version, so check each email address before you submit. PROSPERO sends one automatic reminder; after that, contact slow approvers yourself. When all approvals are in, the record publishes with its registration number, which begins CRD42.
7. Cite and maintain the record. Put the registration number into the protocol document, the submission system and the manuscript at the same time, so the three never drift apart. Set a calendar reminder at the registered end date: PROSPERO emails the record owner then, and that is the moment to either revise the dates with a reason or mark the review completed and add the citation.

Amending the record when plans change

Changing a plan is not the problem; an unrecorded change is. PROSPERO gives you two routes.

Either route publishes a new version. Earlier versions and a dated audit trail stay publicly visible. If you start a minor amendment and realise you need a major revision, delete the draft amendment (the published version is unaffected) and start the major revision instead.

Three habits keep amendments from becoming a problem at peer review:

• Record the change when you make it. If piloting the extraction form shows an outcome is almost never reported, revise the record with the reason at that point, not after the analysis.
• Keep the dates honest. Start and end dates can be changed if the schedule slips. PROSPERO collects the reason and shows it in the record, and emails the record owner when the end date passes.
• Close the loop. When the review is completed and published, update the record's status and add the citation. PROSPERO sends automated reminders, but the guarantor should not rely on them.

Citing your PROSPERO registration in the paper

The PRISMA 2020 checklist sets out what to report:

• Abstract. Give the register name and registration number (item 12 of PRISMA 2020 for Abstracts).
• Registration and protocol (item 24a). Give the register name and registration number, or state that the review was not registered.
• Protocol access (item 24b). Say where the protocol can be accessed, or state that no protocol was prepared.
• Amendments (item 24c). Describe and explain any changes to the information provided at registration or in the protocol.

A sentence you can adapt for the methods or other information section:

If nothing changed, say so in one sentence. Keep the number to hand for the submission system and cover letter, alongside the other items in your manuscript submission checklist. Then open the record and confirm that its title, review question and team match the manuscript, because editors are advised to check that the record accurately reflects the review they receive.

What the record looks like once it is live

The published record is a public web page at a CRD42 URL, and it is indexed, so a search on your review question will surface it. Before you circulate the number, read the page as a stranger would.

Three things routinely look wrong in a live record and are worth fixing straight away:

• Placeholder or boilerplate text in the synthesis field. Text copied from a previous review, or a generic line about narrative synthesis, is visible to every editor who checks the record. Replace it with the plan for this review through a minor amendment or major revision, whichever route the field falls under.
• A team list that no longer matches the paper. People join and leave reviews. The record's contributor list is the version editors compare with your byline, so changes to the review team go in as a major revision when they happen, not at submission.
• An end date that has quietly passed. An expired date on a live record suggests an abandoned review. Either extend it with the reason or mark the review completed.

Keep the record's public URL with the number rather than the number alone. Co-authors, the submission system and your institutional repository can all use the link, and it saves anyone having to search for the record later.

The guarantor's job does not end at registration

For a senior author, agreeing to be guarantor means answering for a public record that editors, reviewers and other teams will rely on for the life of the review. It belongs with the wider research integrity responsibilities of a lab lead: check the stage declaration yourself, make sure changes reach the record as they happen, and read the manuscript's registration and amendments text before it goes out.

When your review is ready, Directive Publications' For Authors pages cover manuscript preparation, article types and research ethics, so you can check what to prepare before you submit your manuscript.