PROSPERO Registration: What the Record Must Contain and How Journals Check It

Most guidance about "prospective" registration talks loosely about registering "before you start." PROSPERO's actual rule is narrower and worth stating precisely: a review is ineligible if it has already started data extraction at the point of registration. Scoping the literature, running preliminary searches, or reading around the topic before you formally register does not disqualify you. Beginning to pull data out of the studies you have identified does.

That distinction matters if you are used to how ICMJE treats late clinical trial registration, where a rare, disclosed exception exists. PROSPERO draws a harder line for systematic reviews: register early, or register nowhere, at least on this registry.

Registering is done online by a member of the review team who is logged into a PROSPERO account. The form asks for the review's design and methods, along with information about the review team, funder, and timelines. A PROSPERO record is not considered a full protocol, but you can attach a PDF of your complete written protocol as part of the record.

Two mechanics catch teams off guard:

The system checks for similar reviews already registered. If it finds one, you are asked, though not required, to explain why your review is still needed. Whatever you enter, including nothing, becomes part of the public record. This exists to reduce duplicate effort across research teams working on the same question at the same time.

Every named team member has to confirm participation before publication. A named guarantor, usually the senior team member, approves the record's content, and every listed reviewer has to consent to having their name and email published alongside it. Only after every confirmation is in does PROSPERO assign a registration number and publish the record, immediately and without a review queue.

That last point surprises people who assume registries hold a manuscript-style review period before publishing. PROSPERO does not review content beyond automated field checks, does not peer review it, and does not endorse it. A PROSPERO record being live is not PROSPERO vouching for your methods.

One more scenario worth flagging: if you are registering a review as part of a student module or training exercise rather than a review you intend to submit for publication, PROSPERO asks you to use its separate student project option instead of the live registry. Those records are not published and will not turn up in PROSPERO searches, which keeps training exercises out of the pool that journals and other researchers actually search.

A protocol you write once and never revisit is a fiction for most reviews. PROSPERO expects amendments and builds a specific mechanism for them rather than treating every change as a violation.

Changes come in two sizes. A minor amendment covers only a few select fields. Most substantive changes, particularly anything to the review methods or the review team, require a major revision, which opens the full record for editing. Once the changes are made, the revision is submitted for publication like the original record was.

Every amendment, minor or major, gets a dated entry in the record's audit trail, and prior versions of the record stay publicly visible rather than being overwritten. A reader, or an editor, can see exactly what the review team planned originally and what changed later, with a date attached to each version.

This is where registration stops being a formality and starts mattering to your manuscript. PRISMA 2020's reporting checklist has a specific item for it, item 24c under Registration and Protocol, and it asks for three things about any amendment to what was registered: the amendment itself, the reason for it, and the stage of the review process at which it happened.

In practice that means a change to your primary outcome, your eligibility criteria, or your synthesis plan needs to show up in two places that agree with each other: the PROSPERO record's amendment history, and the manuscript's own account of what changed and why. A change that exists in one place but not the other is the exact gap an editor is trained to look for.

That gap does not require bad intent to be a problem. A protocol amendment made for a defensible reason, say the planned outcome measure turned out to be reported in too few of the eligible trials to synthesize, is a normal part of running a review. What makes it read as a normal amendment rather than a quiet reshaping of the question is whether it is dated, explained, and visible in both places.

PROSPERO publishes guidance aimed specifically at journal editors, and it names the failure patterns editors are asked to check for, not in the abstract, but as concrete verification steps.

Source: PROSPERO, Important Information for Journals (accessed September 2026).

PROSPERO's own page is direct about why this matters: falsification of registration to improve a manuscript's chances of acceptance is something the registry says it has observed increasing, which is exactly why it built the audit trail and dated version history into every record rather than leaving amendments untracked. It is also explicit that editors and peer reviewers should not ask or encourage authors to register retrospectively once a manuscript has already been submitted; at that point, PROSPERO's position is that it is simply too late to register.

PROSPERO is equally clear about the limits of its own role here: it does not check record content beyond its automated field validation, has no control over whether a review is ever submitted for publication, and has no control over whether a team completes a review without registering it anywhere at all. The verification work described above is something journals and reviewers have to do themselves, using the public record as the evidence.

A team plans a systematic review of a physiotherapy intervention for post-stroke shoulder pain. Before running their MEDLINE and CINAHL searches, they register on PROSPERO with a PICO question, a planned search strategy, and a primary outcome measured at six weeks. Three months into screening, they notice the six-week measure is reported in less than half of the trials that otherwise meet their eligibility criteria, while a twelve-week measure appears in nearly all of them.

They open a major revision, change the primary outcome to the twelve-week measure, and record the reason in the amendment field: the six-week measure was underreported across the eligible literature and using it as the primary outcome would have meant discarding trials that were otherwise a good fit for the question. The revision is dated and stays attached to the public record next to the original version.

When they write the manuscript, item 24c gives them the structure: what changed, why, and at what stage of the review. They state it in the Methods section using the same account the amendment record already carries. An editor comparing the registered outcome to the reported one sees a documented, dated, reasoned change rather than a swap discovered only by cross-referencing.

Compare that to a review where the primary outcome quietly shifts to something looser, "clinical improvement, however measured," somewhere between registration and submission, with no amendment on the PROSPERO record and no mention of a change in the manuscript. That is precisely the pattern PROSPERO's own guidance asks journals to check for: an undisclosed gap between what was registered and what was reported, regardless of whether the gap was intentional.

In our pre-submission review work, PROSPERO problems rarely look like a missing registration. They look like a record and a manuscript that were clearly written by people who stopped comparing them to each other partway through.

The registration number in the abstract resolves to a real PROSPERO record, but the record's title describes a narrower or differently scoped review than the manuscript actually reports. Nobody updated the record when the review's scope shifted, and nobody flagged in the manuscript that the scope had moved.

The record's registration date sits a few weeks after the stated search start date in the Methods section. That gap is often nothing, a delay in finishing the online form after the team had already agreed on their approach, but it reads exactly like the retrospective-registration pattern PROSPERO's own guidance tells editors to check for, and the manuscript never explains the gap.

The manuscript's reported primary outcome does not match the record's, and there is no amendment on file. Sometimes this is a genuine oversight: the team changed course for a defensible reason mid-review and simply never opened the major revision to log it. An editor cannot distinguish that from an undisclosed change by reading the manuscript alone, and by PROSPERO's own guidance, reviewers are asked to assume the more skeptical reading until it is explained.

Comparing your PROSPERO record against your manuscript, line by line, is not something a generic grammar or formatting check will catch, because it requires reading two separate documents against each other. The free readiness scan and the $39 Full Review can flag an unresolved mismatch between a stated registration number and what the manuscript describes, alongside other methods-and-claims consistency issues, but neither one substitutes for a co-author who was not involved in writing the outcomes section actually opening the live PROSPERO record and reading it against the draft before you submit.