How to register a review in PROSPERO

What PROSPERO is

PROSPERO, the International Prospective Register of Systematic Reviews, is a free database in which review authors can record the key features of a systematic review protocol before the work is done. It was established to promote transparency in a field where, until recently, there was no way to know which reviews were under way or whether a published review had followed its original plan. It is hosted by the Centre for Reviews and Dissemination at the University of York, with support from the National Institute for Health and Care Research in the United Kingdom, and it accepts registrations from anywhere in the world.

A registration creates a public record with a unique number. The record contains a structured summary of the protocol, including the question, the eligibility criteria, the search plan, the outcomes and the planned methods of appraisal and synthesis. It is not a full protocol and is not peer reviewed in the way a journal protocol is. Staff check that the record is complete and falls within scope, but they do not assess the quality of the proposed review. See the page on protocol development and registration for the preparation of the protocol itself.

Why register

Registration serves several purposes. It makes the plan public before the results are known, which allows readers to compare the published review with the plan and to see whether outcomes, criteria or analyses changed, and why. This protects against selective reporting, in which only favorable results are published. It helps researchers to see whether a review on the same question is already under way, and so to avoid duplicating the work or to join forces. It supports the authority of the review, since a registered protocol shows that the methods were planned. And many journals now ask for a registration number at submission, and some will not consider a review without one. A study of reviews published in high-impact journals found that many had not been registered, which suggests that the practice is still not universal, and the PRISMA 2020 guideline asks authors to report the registration.

Registration does not guarantee quality, and a registered review can still be conducted poorly. What it provides is the means to detect departures from plan, and a record of what was intended.

Which reviews can be registered

PROSPERO accepts registrations for systematic reviews of studies with a direct health-related outcome. Its remit covers a range of review types within health, including reviews of interventions, diagnostic accuracy, prognosis, qualitative evidence with a health focus and some others, subject to criteria that have evolved over time. It does not accept scoping reviews, literature reviews, mapping reviews or reviews of methodological studies, which fall outside its definition, and it has had different rules for reviews of animal studies and for reviews with no health outcome. Because its eligibility rules have changed and may change again, authors should read the current guidance on the PROSPERO website before preparing a record.

Reviews that do not qualify can be registered elsewhere, as described below. A review that is already complete cannot normally be registered, since registration is meant to be prospective.

When to register

Registration should happen before the review's key decisions are made. The ideal is before the search is run, and at the latest before screening of records against the eligibility criteria has begun. PROSPERO has set specific timing requirements for registration, such as that formal screening of search results should not have started or, in some versions of the rules, that data extraction should not have begun, and it may decline records that fall outside them. The current rule should be checked at the time of registration, since it has been modified. Registering earlier is always preferable, because it fixes the plan before there is any opportunity for decisions to be influenced by what the studies show.

If a review has already started, authors can still consider whether the registry will accept it, and should say honestly at what stage it began. A late registration cannot offer the protection that a prospective one does, but it is better than none if allowed.

What the record asks for

The record is structured as a set of fields, drawn from a minimum dataset developed through an international consultation. The main items, in outline, are listed below.

• Title and review team. The title identifies the review, and the review team and contact details are given.
• Timing and status. The planned start and end dates, and the stage the review has reached.
• Review question and rationale. The question, and the reasons for doing the review.
• Searches. The databases and other sources, the dates, any language limits and the plan for grey literature.
• Condition, population, intervention or exposure, comparators and outcomes. The criteria for inclusion, with the primary and secondary outcomes.
• Study designs. The types of studies that will be included.
• Data extraction and risk of bias. The data items, the process of selection and extraction, and the tools for assessing risk of bias.
• Strategy for data synthesis. Whether a meta-analysis is planned, how heterogeneity will be examined, and the planned subgroup and sensitivity analyses.
• Funding, conflicts and dissemination plans.

The details of the form change, so authors should use the current version. Preparing the content before filling in the form is wise, since the protocol is the substance and the record is a summary of it.

The registration process

1. Prepare the protocol

   The question, criteria, methods and analysis plan are developed, ideally using the PRISMA-P guideline, and discussed by the team.

2. Check eligibility and timing

   The authors confirm that the review is within the registry's scope and that its stage allows registration.

3. Create the record

   An account is created and the form is completed. Fields cover each of the items listed above.

4. Submit and respond to queries

   The record is checked by registry staff for completeness and scope, and questions may be raised. After acceptance, the record is published with its registration number.

5. Cite and update

   The number is cited in the protocol, in any grant or journal submission, and in the final paper. The record is updated as the review progresses, and changes are made as amendments.

Amendments and reporting changes

Protocols change, and that is acceptable if the changes are recorded. PROSPERO retains the history of a record, so the original version and every amendment remain visible with dates. When a record is amended, the authors should update the relevant field and note what was changed and why. In the final paper, any departure from the registered protocol is reported, with its reason, so that readers can judge whether it could have been influenced by the results. A change made before data were seen is less concerning than one made afterwards. The aim of the system is not to prevent change but to expose it, and the review that records and explains its amendments is more credible than one that reports no changes where there plainly were some.

If PROSPERO does not accept your review

Several alternatives exist. Open science platforms, such as the Open Science Framework and similar services, allow time-stamped registration of review protocols of any kind, and are widely used for scoping reviews and for reviews outside health. Some disciplines have their own registries, and some systematic review organizations maintain registers for their members. Journals that publish protocols, including several that specialize in evidence synthesis, provide a peer-reviewed, citable record, and a published protocol can complement or substitute for registration where the registry does not apply. Whatever route is used, the aim is the same: a public, dated statement of the plan. The registration should be cited in the review in the same way as a PROSPERO number. See scoping review for the case of reviews that PROSPERO does not accept.

What a good registration looks like

Registrations differ greatly in quality, and readers who open one look for specific things. The question and the eligibility criteria are precise enough that another team could apply them. The primary outcome is named, with the measure and the time point, and not described only as the main outcomes of interest. The search plan names the databases and the date range, and does not say only that major databases will be searched. The synthesis section says whether a meta-analysis is planned, under what conditions, and which subgroup and sensitivity analyses are intended, with a rationale. Vague phrases such as if appropriate, where possible or as needed leave room for decisions after the fact and weaken the protection that registration offers. A record that is specific is also the easiest to follow, and keeping it so saves arguments with reviewers later.

Common mistakes in registering

• Registering after screening is well under way without saying so, or after the review is finished.
• Leaving the primary outcome or the synthesis plan open.
• Registering one question and publishing another, without recording the change.
• Never updating the record, so that the amendments do not appear.
• Failing to cite the registration number in the paper, or citing a number that does not match the review.
• Registering a review that falls outside the registry's scope and then having it rejected late, which loses time.
• Treating registration as a formality and not as the planning document it is meant to be.

Trial registries and other kinds of registration

PROSPERO is a register of reviews. Other registries cover other kinds of research. Clinical trials are registered in trial registries, such as ClinicalTrials.gov and the registries linked through the World Health Organization's International Clinical Trials Registry Platform, and these records matter to reviewers, because they allow a review to compare the outcomes planned for a trial with those reported, a key check for selective reporting. Observational studies and other designs can be registered on open science platforms, where preregistration is increasingly common. A systematic review of trials will therefore make use of trial registers as a source of studies and of protocols, and its own registration sits alongside them. The two kinds of registration have a shared purpose: making the plan visible before the results are known.

What registration does not do

Registration records a plan. It does not certify that the plan was good or that it was followed, and it does not by itself make a review publishable. The registry checks a record for completeness and scope, not for the soundness of the methods. A registration number is also not a guarantee that the review will be completed, and some registered reviews are never finished. Readers who see a registration number should still check whether the published review matches the record. And registration of a protocol does not protect against problems with the studies themselves, such as publication bias in the primary literature.

Support

Frequently asked questions