Prospero Registration Systematic Review: A Practical Guide

The prospero registration systematic review process gives a review team a public, dated record of its planned methods. It is best understood as an auditable plan—not a full protocol, peer review, or certification that the review is methodologically sound. PROSPERO is a free registry for systematic reviews with health-related outcomes.[1][3][4]

Registering prospectively helps readers compare planned methods with the completed review and identify potentially overlapping work. It does not guarantee that methods are sound or prevent selective reporting or duplication. A useful registration therefore needs enough detail to make the plan understandable, and the final report must still explain any important changes.[2][3][4]

What is PROSPERO registration—and what does it not certify?

PROSPERO registration is a public record summarizing a review’s planned methods; it is not a substitute for a full protocol. The record makes the plan dated and discoverable, but registration alone does not establish the quality of the design or conduct.[1][3][4]

Treat the record as one part of the review’s documentation. PRISMA 2020 distinguishes registration from a full protocol and asks authors to report the registry, registration identifier, and where the protocol can be accessed.[2][3] A publicly accessible protocol can provide details that are difficult to capture in registry fields. An assessment of PROSPERO records against PRISMA-P found substantial reporting gaps, reinforcing the need for sufficiently detailed entries and, where feasible, a full protocol.[6]

When should you register a systematic review?

Register the review before data extraction begins; registering before formal screening is preferable.[5][7] This timing makes it easier to distinguish methods planned in advance from decisions made after examining included studies or extracted data.

A registration made after the review is completed or a manuscript is submitted is retrospective, not prospective.[5][7] If work has already started, describe its actual status accurately rather than presenting the record as if it preceded the work. Check PROSPERO’s current eligibility criteria before submitting, because requirements can change.[1]

What should a useful PROSPERO record include?

A useful record specifies the review question and eligibility criteria, search sources and restrictions, outcomes, screening and extraction procedures, risk-of-bias methods, and planned synthesis.[5][6] The detail should let readers understand what evidence the team intends to include and how it plans to evaluate and combine that evidence.

Be clear about outcomes and planned synthesis decisions, rather than relying on broad labels. Describe the intended approach to screening, extraction, and risk-of-bias assessment in terms that can be compared with the completed work. Attaching a protocol does not replace completing the relevant registration fields.[5][6] Because the registration form and criteria may change, use PROSPERO’s current guidance rather than relying on a static checklist.[1]

How do you keep the registration aligned with the review?

Keep the record current by documenting and dating methodological amendments, updating completion status, and adding publication details when available.[5][7] In the manuscript, independently describe and justify material deviations from the plan; an updated registry entry does not remove the need to report them.[5][7]

A practical workflow is to assign responsibility for maintaining the record and checking it against the manuscript before submission. Compare the registered question, eligibility, outcomes, and planned methods with the methods actually used. If decisions change during the review, preserve a clear account of what changed and why, so readers can distinguish planned methods from later decisions.

What are the benefits and limits of registration?

Registration supports transparency by allowing readers to compare planned and reported methods, and it can help researchers identify ongoing or completed reviews on similar topics.[2][3][4] These functions can make overlap more visible, but they do not guarantee sound methods, eliminate duplication, or prevent every form of selective reporting.[2][3][4]

Some researchers may worry that public registration exposes an idea. In one survey, 47.8% of respondents said they had considered that a registered idea could be stolen, and 5.3% reported personally experiencing idea theft; these are survey responses, not population estimates of actual theft.[8] Teams should weigh that concern against the transparency value of a dated public record, without treating the survey figures as evidence of the frequency of idea theft.

How should teams describe AI-assisted screening or extraction?

Describe AI assistance in the plan and verify AI-assisted extraction with people; current evidence does not support replacing human methods for core evidence-synthesis tasks.[10] The registration and protocol should make the intended role of the tool clear, while the final report should explain what was automated and how outputs were checked.

A 2025 systematic review reported incorrect data extractions ranging from 4% to 31% across included evaluations, with a median of 14%.[9] These heterogeneous findings are not a universal error rate for a particular model, task, or workflow.[9] AI may support structured processes, but it does not replace human decisions about methods or the need to verify extracted data.[10]

Key takeaways

• PROSPERO provides a public, dated summary of planned methods, not a quality certification or full protocol.[1][3][4]
• Register before data extraction starts, and preferably before formal screening.[5][7]
• Specify eligibility, searches, outcomes, review procedures, risk-of-bias methods, and planned synthesis.[5][6]
• Date and explain amendments, update the record, and justify material deviations in the manuscript.[5][7]
• Treat AI extraction evidence as heterogeneous and verify AI-assisted outputs with people.[9][10]

FAQ

How do you register a systematic review on PROSPERO?

Check PROSPERO’s current eligibility criteria and guidance, then complete the relevant registration fields with the planned methods.[1][5] Register before data extraction begins, and preferably before formal screening.[5][7]

Does a PROSPERO record replace a full protocol?

No. PROSPERO’s record is a summary of planned methods, while PRISMA 2020 distinguishes registration from a full protocol.[2][3] Report the registry and identifier in the manuscript and state where the protocol is accessible, if there is one.[2][3]

How should teams handle a change to their registered methods?

Update the record, date and explain the amendment, and describe and justify material deviations in the manuscript.[5][7] This preserves the distinction between the original plan and methods adopted later.

How accurate is LLM data extraction in systematic reviews?

The available figures vary across evaluations: a 2025 review reported incorrect extractions ranging from 4% to 31%, with a median of 14%.[9] Those heterogeneous results do not establish a universal accuracy rate for any specific model or workflow.[9]