How to Write a Scoping Review Protocol: JBI and PRISMA-ScR Guidance

By Dr. Esmaeel Saeedy Robat, Founder of MetaSyn Academy · Published meta-analyst (Nature Human Behaviour, 2026)

A scoping-review protocol has to do two things at the same time: commit the team to a defensible prospective plan and leave room for the kinds of learning that scoping methodology genuinely expects. Too little specification creates post hoc drift. Too much rigidity can make the protocol fight the exploratory purpose of the review. Current JBI methodology resolves that tension by fixing the objectives, question, inclusion logic and planned methods while allowing transparent refinement of searching and data extraction/charting as the team learns more about the evidence base.^1,2

This is why a scoping-review protocol should not be built by taking a conventional intervention-review protocol and changing PICO to PCC. The differences run deeper. Scoping reviews usually map concepts, evidence characteristics, methods, definitions or gaps rather than estimate a pooled effect. Their evidence sources may be broader. Critical appraisal is purpose-dependent. Data extraction is commonly described as charting and may be refined after piloting. Analysis is usually descriptive and categorical rather than inferential. PRISMA-ScR then tells authors how to report the completed review; it does not supply the conduct method used to design the protocol.^2,3

Why the protocol matters in an intentionally iterative review

JBI’s current scoping-review chapter states that an a priori protocol must be developed before undertaking the scoping review. The protocol predefines objectives, methods and reporting, specifies inclusion and exclusion criteria, identifies the data considered relevant, and explains how that data will be extracted and presented.¹ Peters and colleagues make the same point in their protocol-specific guidance: the protocol is the first stage of a JBI scoping review and should be finalized before the ensuing review is conducted.²

That prospective commitment does not mean every search term and every charting column must remain unchanged. Scoping reviews often begin in fields where language is inconsistent or where the structure of the evidence is not fully known. JBI therefore allows the search to be iterative and explicitly permits the draft charting form to be refined during the review.¹ The protocol’s job is to make the difference between planned flexibility and hindsight visible.

This distinction is especially important because the word iterative is easy to misuse. Discovering a new synonym after the initial search is a predictable search refinement. Adding a useful charting field after piloting is a predictable data-collection refinement. Replacing the central Concept after seeing the results, widening the review to a different population because the original search returned few sources, or introducing an effect meta-analysis after discovering convenient quantitative data changes the design problem. Those decisions need explicit amendment and, in the most substantial cases, a new or materially revised protocol.

The current JBI protocol structure is practical, not a historical stage list

The historical development of scoping reviews is useful background, but the protocol should follow current guidance rather than an old stage diagram. Arksey and O’Malley introduced the original framework, and later work refined it. The current 2024 JBI Manual, however, has a dedicated section titled Development of a scoping review protocol in Chapter 10. Its subsections cover title; developing the title and question; introduction; inclusion criteria; search strategy; source-of-evidence selection; data extraction; analysis of the evidence; and presentation of results.¹

Write the rationale as an argument for scoping, not as a generic literature-gap paragraph

Many weak protocols say that a topic is important, that the literature is large or heterogeneous, and that no previous review has been found. None of those facts, by itself, establishes a scoping-review purpose. The rationale needs to explain the methodological job the review will perform. Is the problem inconsistent terminology? An emerging field with unclear evidence categories? A need to map which populations, contexts or outcomes have been studied? A need to understand how a concept has been operationalized? A need to describe available methods before a more focused synthesis can be designed?

Munn and colleagues’ guidance on choosing between systematic and scoping review approaches remains useful here. A scoping review is appropriate when the question is about the extent, characteristics or nature of evidence rather than a focused synthesis intended to answer an effect or other inferential question.⁵ That does not mean all broad questions are scoping questions. An intervention-effect question can remain a systematic-review question even when the intervention is complex or the evidence is heterogeneous.

PCC should narrow the question only as much as the purpose requires

Population, Concept and Context are often taught as three mandatory boxes. JBI is more nuanced. Population should describe relevant participant characteristics when participants matter, but the Manual explicitly notes that participant-level criteria may be irrelevant in some reviews, such as a review focused on research designs in a field.¹ Concept is the core idea being mapped. It can include elements that other review frameworks would call an intervention, phenomenon of interest or outcome. Context can capture geography, culture, health system, institution, setting or another contextual boundary, and it may be left open when all contexts are eligible.

The common error is not that PCC is too broad. It is that the three elements are disconnected from the question. Consider a review that aims to map how academic institutions define and evaluate responsible generative-AI use in doctoral education. Population might be unnecessary or could refer to doctoral learners and supervisors, depending on the objective. Concept could include definitions, policies and evaluation approaches. Context could include universities internationally or a defined region. If the protocol instead inserts “AI intervention” as Concept, “no AI” as a hidden comparator and “academic performance” as a primary outcome, the team has changed the question into something closer to comparative effectiveness.

Keep the title, question, objective and criteria congruent

JBI recommends that a scoping-review title clearly identify the work as a scoping review and that protocols are identified as protocols. The question should incorporate PCC elements, and there should be clear congruence among the title, review question and inclusion criteria.¹ This sounds elementary, but it prevents several downstream problems. A title can promise a global review while the criteria quietly restrict to one country. An objective can promise mapping of implementation approaches while the charting form collects only outcomes. A question can mention clinicians while eligibility admits policy documents with no clinician-level information.

A useful protocol test is to trace every noun in the objective into an inclusion criterion or charting field. If the objective says the review will map definitions, the charting form needs a place to record definitions. If it says the review will compare how a concept is measured, the eligibility and charting strategy need to capture measurement instruments or operationalizations. If it says the review will identify gaps, the planned analysis should explain how gaps will be recognized rather than leaving “identify gaps” as a rhetorical conclusion.

A broad evidence base still needs a source policy

JBI allows a wide range of evidence sources in scoping reviews, including primary research, systematic reviews, meta-analyses, letters, guidelines, websites and other literature when appropriate to the question.¹ That flexibility is one of the strengths of the method, especially for topics that cross research, policy and practice. It also creates a protocol risk: the phrase “all types of evidence will be included” can become an excuse for an incoherent evidence set.

Source types should be justified by function. A review of implementation definitions might need empirical studies, frameworks and policy documents. A review of trial outcome measures might need primary studies and perhaps prior reviews for contextual mapping, but opinion pieces may add little. A review of national policy approaches may depend heavily on government documents and organizational reports. The protocol should therefore state both what source classes are eligible and what each class contributes to the mapping objective.

If primary studies and systematic reviews are both eligible, the protocol should also consider duplicate representation. A systematic review may summarize primary studies that are separately included. The issue is not necessarily “double-counting” in the statistical sense, because a scoping review may simply map source characteristics, but readers should be able to understand whether the same underlying evidence appears at multiple source levels.

The three-step search is rigorous and can still be iterative

Current JBI guidance says a scoping-review search should ideally be as comprehensive as possible within time and resource constraints. The protocol should describe limitations and justify them. JBI recommends a three-step strategy: an initial limited search of at least two appropriate online databases; analysis of text words in titles and abstracts and of index terms, followed by a comprehensive search using the identified terms across all included databases; and reference-list searching for additional sources.¹

The initial step is not the final evidence search. It is a term-discovery exercise that tests how the Concept is described in the literature. This is one reason scoping searches are often iterative. New terms or source types can appear as reviewers become more familiar with the field. JBI explicitly permits this form of learning as long as the search strategy and results remain transparent and auditable.¹

Grey or difficult-to-locate material deserves similarly careful wording. JBI aims to identify published and unpublished evidence and says a grey-literature search might be necessary. That is not the same as saying every scoping review must search every possible grey-literature source. The source policy should follow the question. For a policy-mapping review, omitting government and agency documents may fatally distort the evidence map. For a narrowly methodological question centered on indexed research reports, a smaller grey-literature component may be defensible if the rationale is explicit.

JBI also describes research-librarian or information-scientist input as invaluable and cites PRESS for search peer review. This is strong methodological support for involving search expertise, but it should not be rewritten as a universal staffing requirement. The protocol should say who designed the search, whether it will be peer reviewed, and how platform-specific strategies will be preserved. PRISMA-S can be used as a supplementary search-reporting standard when greater reporting detail is helpful, but it does not replace JBI conduct guidance or PRISMA-ScR.⁸

Broad eligibility makes calibration more important, not less

JBI’s current guidance is direct on source selection: title/abstract screening and full-text screening are performed independently by two or more reviewers; disagreements are resolved by consensus or a third reviewer.¹ The Manual also gives a practical pilot framework: a random sample of 25 titles and abstracts can be screened by the team, discrepancies discussed, eligibility definitions refined, and full screening started once the team reaches the recommended agreement level. These details are implementation guidance rather than a universal measurement of protocol quality, but they illustrate the role of calibration.

Calibration is particularly useful when the Concept is abstract. If two reviewers disagree repeatedly over whether a paper really addresses “responsible AI governance,” the problem may not be reviewer performance; the Concept may not be operational enough. Pilot screening therefore tests the inclusion criteria as much as it tests the reviewers.

When calibration leads to a wording change, preserve the history. Clarifying a definition before formal screening begins is different from changing the central Concept after half the evidence has been screened. Both can be documented, but they have different implications for transparency.

Data extraction and data charting: the important difference is the purpose

The terminology itself should not become a distraction. Current JBI uses the heading Data extraction and explicitly states that, in scoping reviews, the process may be referred to as data charting.¹ The methodological distinction is more useful than the vocabulary: the form is built to describe and map the evidence in relation to the review question, and JBI expects a draft form to be developed and piloted at protocol stage.

JBI lists common information that reviewers might chart, such as author, year, country, aims, population and sample size where relevant, methodology, intervention or comparator details where applicable, outcomes where relevant, and findings that relate directly to the scoping question.¹ That list is not a universal extraction sheet. Pollock and colleagues’ later guidance reinforces the purpose-driven nature of extraction, analysis and presentation.⁶ A review mapping definitions should chart definitions. A review mapping implementation models should chart model components and contexts. A review mapping outcome measurement should chart outcome domains and instruments.

JBI explicitly permits the charting table to be updated as the team becomes familiar with sources. It recommends pilot testing on a small number of sources by at least two review-team members; where that is not feasible, one reviewer extracting and another verifying can be considered, provided the method is transparent.¹ This is a useful example of legitimate iteration. Adding a field called “implementation level” after the first pilot reveals that the evidence needs a better descriptive category. Replacing “definitions of implementation” with “effectiveness of implementation strategies” changes the review question.

Critical appraisal is a decision, not a defining feature

Scoping reviews are often described as reviews that “do not appraise quality.” That is too absolute. The more accurate formulation is that critical appraisal is not routinely required because the principal purpose is usually to map evidence rather than make an inferential judgment about an effect. Peters and colleagues therefore treat risk-of-bias or critical-appraisal items differently from conventional systematic-review protocol items and advise authors to describe appraisal when it is actually planned.²

The protocol should answer three questions if appraisal is included. First, what purpose does appraisal serve? A review might map how robust the existing methodological base is, or use appraisal to contextualize evidence characteristics. Second, which tool fits each eligible evidence type? A review that includes qualitative studies, surveys and policy documents cannot responsibly apply one generic quality score. Third, how will appraisal affect the review? If methodological quality will influence inclusion, categorization or interpretation, that decision should be made prospectively rather than after the appraisal results are known.

Appraisal also should not be confused with certainty of evidence. There is no standard scoping-review equivalent of GRADE that authors can simply attach to a descriptive evidence map. Peters et al. explicitly noted the absence of a scoping-review GRADE approach and uncertainty about whether such an adaptation would be useful.² If the review’s real goal is to produce a certainty-rated answer to a focused clinical question, that is another reason to reconsider the design.

Scoping reviews analyze evidence; they usually do not make the same synthesis claims as systematic reviews

The statement “scoping reviews do not synthesize” is unhelpful because it collapses very different forms of analysis. JBI describes simple frequency counts as sufficient for many reviews but also permits more in-depth descriptive analysis, including descriptive qualitative content analysis and coding, depending on the purpose.¹ Pollock and colleagues expand this guidance for extraction, analysis and presentation.⁶

The boundary appears when the analysis changes the kind of conclusion being sought. Describing how many studies used a particular framework, which countries are represented, how definitions vary, which outcome domains have been measured, or which combinations of population and context are missing is consistent with a scoping purpose. Basic qualitative content analysis can organize concepts or categories. More advanced descriptive quantitative summaries can also be appropriate.

By contrast, interpretive qualitative synthesis that aims to generate a new explanatory model may belong in a qualitative evidence-synthesis methodology. Statistical pooling intended to estimate an intervention effect usually belongs in a systematic review designed for that inferential purpose. Current JBI wording is appropriately cautious: meta-analysis is unlikely to be required in scoping reviews rather than being declared logically impossible.¹ A protocol that anticipates pooled effect estimates should therefore trigger a design review, not an automatic software prohibition.

Consultation has evolved toward knowledge-user engagement

The original scoping-review framework described consultation as optional. More recent JBI guidance develops the idea into knowledge-user involvement and co-creation. Pollock and colleagues describe ways knowledge users can contribute to question development, eligibility, search terminology, extraction, interpretation and dissemination.⁷ That does not make engagement mandatory in every protocol.

The protocol should include an engagement section when people outside the review team will shape methodological or interpretive decisions. State who will be involved, at which stages, what kind of influence they will have and how that contribution will be reported. Do not confuse this role with treating stakeholder interviews or consultation statements as primary evidence. If the project is generating new research data, that activity has a different methodological and ethical status.

Protocol development, public availability and registration are different things

A protocol can exist without being publicly registered. A protocol can be publicly deposited without being registered in a formal registry. A protocol can be published in a peer-reviewed journal. These routes overlap in their transparency goals but should not be described as synonyms.

For JBI scoping reviews, the prospective protocol itself is methodologically important. Peters et al. describe protocols as documents that should be finalized before conduct and note that they can be submitted for journal publication or hosted elsewhere, including Open Science Framework and Figshare.² Current JBI material likewise points to OSF, Figshare or journal publication.¹ Public availability makes the prospective record easier for readers to inspect, but peer-reviewed publication is not a universal prerequisite for conducting the review.

PROSPERO requires a separate warning. Its current official eligibility guidance states that scoping reviews are not accepted.⁹ The presence of the word “scoping review” inside some PROSPERO records does not contradict that policy; an eligible systematic review may include published scoping reviews as evidence sources. Mislabeling your own scoping review as a systematic review simply to obtain a PROSPERO identifier damages rather than improves transparency.

PRISMA-ScR should influence reporting readiness, not control protocol conduct

The PRISMA extension for scoping reviews was published in 2018 and contains 20 essential reporting items plus 2 optional items. It covers the completed review’s title, summary, rationale, objectives, methods, results, discussion, conclusions and funding.³ Several of those reporting items can be anticipated prospectively. A protocol can pre-specify eligibility, information sources, search methods, source selection, charting, data items, appraisal and the intended synthesis of results. Other items, such as the number and characteristics of included sources, review limitations and the summary of evidence, cannot be completed until the review exists.

This is why a protocol-to-reporting crosswalk is methodologically cleaner than a claim of “PRISMA-ScR compliance.” The protocol can make future reporting easier by preserving the information that PRISMA-ScR will later require. It cannot satisfy results items prospectively, and PRISMA-ScR cannot guarantee that the conduct methods were appropriate.

The reporting guideline is also being updated. The 2026 scoping review supporting that update screened 8,265 records, included 43 documents and identified 37 potential reporting items for subsequent Delphi evaluation.⁴ Those are candidate items, not a new checklist. As of 9 August 2026, the official PRISMA site still lists the 2018 PRISMA-ScR as the published guideline and the update among extensions under development.^3,4

If AI or automation is planned, the protocol should make the human control visible

AI is not a scoping-specific protocol domain, so it should remain compact. It matters when automation is actually planned. A 2025 joint position statement across Cochrane, Campbell, JBI and the Collaboration for Environmental Evidence supports responsible AI use in evidence synthesis when methodological rigor is preserved, human oversight is maintained and consequential use is transparently reported.¹⁰ The associated RAISE recommendations provide more detailed role- and tool-focused guidance.

For a protocol, a short disclosure is enough: name the tool and version where possible, identify the task, describe whether it makes or suggests judgments, and state how humans will verify those outputs. This is especially important when AI assists eligibility screening or extracts conceptual information. It is less useful to force an “AI section” into a protocol when the only automation is ordinary deduplication already built into reference-management software.

Five protocol mistakes that change the meaning of the review

Frequently asked questions

A strong protocol makes later iteration interpretable

The most defensible scoping-review protocol is neither a rigid contract that denies how the method actually works nor a loose plan that can be rewritten after every discovery. It fixes the reason for the review, the central question, PCC logic, evidence-source policy and planned methods. It also acknowledges where scoping methods learn from the evidence base: search terminology, charting fields and sometimes the practical articulation of criteria.

The protocol therefore becomes an audit trail of reasoning. It tells readers what the team believed before the evidence was mapped, what changed as the review developed, why those changes were made and whether they remained inside the original purpose. PRISMA-ScR can then report that process clearly in the completed review. Used in this order, JBI conduct guidance, scoping-specific protocol guidance and PRISMA-ScR complement one another rather than compete.