🧠 Concept

1. What Is a Systematic Review Protocol?

A protocol is a document that specifies, in advance, the rationale, question, and methods for a systematic review. It describes what will be done, how, and why — before the reviewer has seen any results. In this sense, a protocol is the systematic review's analogue of a clinical trial's study protocol: a pre-registered plan that fixes the study's design before data collection begins.

A protocol typically includes:

• Background and rationale for the review.
• The review question (PICO/PECO/PICo).
• Eligibility criteria (inclusion and exclusion).
• Information sources (databases, registers, grey literature).
• Full search strategy (or at least the search concepts).
• Study selection process (screening workflow).
• Data extraction plan and form.
• Risk-of-bias assessment tools.
• Synthesis plan (statistical and/or narrative).
• Assessment of certainty of evidence (GRADE).
• Plans for dissemination and updating.
• Timeline, roles, and any funding or conflicts of interest.

2. Protocol vs Manuscript

The protocol and the manuscript together form a complete, auditable record: the protocol says what was planned; the manuscript says what was done. Deviations between them (documented and justified) are visible to readers.

3. How Protocols Reduce Bias

Systematic reviews are vulnerable to a class of biases collectively known as researcher degrees of freedom — the many small decisions a reviewer makes during a review that can influence its outcome. A protocol constrains these decisions:

Pre-specification ensures that methodological choices reflect prior reasoning about the evidence, not preferences for the outcome.

4. Transparency and Reproducibility

A protocol provides three kinds of transparency:

• Procedural transparency — readers can see exactly what the team intended to do, step by step.
• Analytical transparency — the plan for synthesis, subgroup analysis, and sensitivity analysis is open to scrutiny.
• Historical transparency — the protocol's registration date establishes a public record; readers can see what was changed and when.

Together, these support reproducibility — the ability of an independent team to replicate the review using the same methods and reach comparable conclusions.

5. Consequences of Conducting a Review Without a Protocol

Modern standards — Cochrane, JBI, Campbell, PRISMA 2020, PRISMA-P, and most high-impact journals — assume that a protocol exists and was registered. Reviews without protocols are increasingly treated as methodologically inferior, even when otherwise well-conducted.

6. When Can a Protocol Be Amended?

A protocol is not a straitjacket. Amendments are permitted when:

• Peer review or stakeholder feedback reveals a genuine gap.
• Pilot screening reveals genuinely ambiguous eligibility criteria.
• A relevant new review or study is published mid-conduct.
• Feasibility constraints emerge (e.g., a database becomes unavailable).
• A methodological improvement is justified by new guidance.

Every amendment must be:

• Recorded in a dated changelog within the protocol.
• Justified with a rationale.
• Reflected in the final manuscript as a reported deviation.
• Registered as an update in PROSPERO (or equivalent).

The key principle: deviations are acceptable if transparent; they are unacceptable if concealed.

7. Protocols in Low-Resource Settings

Some reviewers in resource-constrained settings question whether a protocol is worth the effort. The evidence suggests the opposite:

• Protocols reduce rework by forcing clarity on methods before time is spent executing them.
• Registration is free (PROSPERO, OSF, Research Registry).
• Protocols help secure funding, ethics approval, and institutional support.
• They provide a work plan that keeps teams coordinated and focused.
• They establish a published record that is often required for later dissemination.

For teams with limited resources, the protocol is not overhead — it is the most efficient way to allocate those resources.

🧠 Concept

1. Why a Standard Structure Matters

Protocols are structured for the same reason that trial protocols, clinical reports, and grant applications are structured: a well-known format allows reviewers, funders, and registries to evaluate a document quickly. The structure also serves as a checklist — ensuring no essential methodological element is forgotten.

The most widely accepted structure follows PRISMA-P, the 2015 guideline for reporting protocols of systematic reviews. It defines 17 items distributed across 5 sections.

2. The Five Sections of a Protocol

3. Administrative Information

This opening section is short but critical. It includes:

• Title — Should identify the report as a systematic review protocol.
• Registration — Registry name and registration number (once available).
• Authors — Full names, affiliations, and roles.
• Contributions — Who wrote the protocol; who will conduct each stage.
• Amendments — A space to record amendments with dates (initialized empty).
• Funding — Source of funding, or statement that none exists.
• Conflicts of interest — Declarations for every author.

Even where funding is absent, the statement should be explicit — "no funding received" is more transparent than silence.

4. Introduction

Rationale

The rationale explains why the review is needed:

• What is the current state of knowledge on the topic?
• What gaps or uncertainties exist?
• What existing systematic reviews (if any) are available, and why is an update or new review justified?
• Who will benefit from the review's findings?

A strong rationale is typically 2–4 paragraphs. It should not be a comprehensive review of the literature — that is the review itself — but enough background to justify the review's existence.

Objectives

The objectives state the review's aim in a structured, specific form. Typically presented as a single sentence using the PICO/PECO/PICo framework:

"To assess the effectiveness of community health worker-led interventions, compared with usual clinic-based care, for reducing systolic blood pressure in adults with hypertension in low- and middle-income countries."

Where the review has secondary objectives — subgroup analyses, additional outcomes, or parallel qualitative components — they should be listed separately.

5. Methods

The Methods section is the protocol's core. Each subsection pre-specifies a methodological decision.

5.1 Eligibility Criteria

Complete inclusion and exclusion criteria, organized by PICO/PECO/PICo domain:

• Study designs to be included.
• Participant characteristics (age, condition, setting).
• Intervention/exposure definitions, including dose, duration, and delivery mode.
• Comparator definitions.
• Outcome definitions (primary and secondary) with measurement timepoints.
• Language, publication date, and publication status.

5.2 Information Sources

List all databases, registers, and grey literature sources that will be searched:

• Bibliographic databases (PubMed, Embase, CINAHL, Scopus, Web of Science, PsycINFO, etc.).
• Trial registers (ClinicalTrials.gov, WHO ICTRP, ISRCTN).
• Grey literature sources (OpenGrey, ProQuest Dissertations, organizational websites).
• Reference list searching of included studies (backward citation).
• Forward citation searching (via Google Scholar or Web of Science).
• Hand searching of key journals.
• Consultation with experts.

5.3 Search Strategy

The full search strategy for at least one database should be included in an appendix. The protocol should describe:

• The search concepts derived from PICO/PECO.
• MeSH/Emtree terms, free-text terms, and boolean operators.
• Any limits or filters (and their justification).
• Whether the search was peer-reviewed (e.g., PRESS framework).
• Date of the search and whether it will be updated.

5.4 Study Records

This subsection covers data management:

• Reference management software and SR platform used.
• Deduplication method.
• How records will be tracked through the screening process.

5.5 Selection Process

The screening workflow:

• Number of reviewers (dual screening required).
• Pilot/calibration process.
• Conflict resolution (arbitration protocol).
• Full-text retrieval process.
• Handling of multiple reports of the same study.

5.6 Data Collection Process

How data will be extracted:

• Extraction form (included in appendix).
• Number of extractors (dual extraction is standard).
• Discrepancy resolution process.
• Contact with study authors for missing data.

5.7 Data Items

The complete list of data items to be extracted:

• Study identifiers (author, year, country, design).
• Participant characteristics (n, age, sex, condition severity).
• Intervention details (components, dose, duration, provider).
• Comparator details.
• Outcome measures, effect estimates, and confidence intervals.
• Risk-of-bias assessment per domain.
• Funding source and conflicts of interest declared by the study authors.

5.8 Outcomes and Prioritization

The primary and secondary outcomes should be specified with:

• Exact definition (e.g., "systolic blood pressure measured in mmHg").
• Measurement timepoint (e.g., "at 12 months from baseline").
• Ordering (primary vs. secondary).
• Justification for prioritization.

5.9 Risk of Bias in Individual Studies

The tools that will be used, matched to study designs:

• RoB 2 for randomized trials.
• ROBINS-I for non-randomized studies of interventions.
• QUADAS-2 for diagnostic accuracy studies.
• JBI checklists for qualitative, prevalence, and other designs.
• Number of reviewers performing appraisal.
• How disagreements will be resolved.

5.10 Data Synthesis

The plan for combining data:

• Criteria for meta-analysis vs. narrative synthesis.
• Effect measure (OR, RR, MD, SMD, etc.) and rationale.
• Model (fixed-effect vs. random-effects) with justification.
• Heterogeneity assessment (I², τ², prediction intervals).
• Prespecified subgroup analyses.
• Prespecified sensitivity analyses.
• Software to be used.

5.11 Meta-Bias(es)

Plans for assessing bias across studies:

• Publication bias (funnel plot, Egger's test, trim-and-fill).
• Selective reporting.
• Any other meta-bias assessment.

5.12 Confidence in Cumulative Evidence

Approach to assessing certainty:

• GRADE for quantitative findings.
• CERQual for qualitative findings.
• Summary of Findings table format.

6. Appendices

The following documents are typically appended:

• Full search strategy (at least one database; ideally all).
• Data extraction form (blank template).
• Risk-of-bias assessment form.
• Standardized letter template for contacting authors.
• PRISMA-P checklist (completed).

7. Amendments Log

A dated, running record of changes to the protocol. Each amendment entry includes:

• Date.
• Section affected.
• Nature of the change.
• Rationale.
• Approver.

The amendments log makes the review's methodological evolution transparent — a feature that reviewers and readers increasingly expect.

8. Common Weaknesses in Protocols

🧠 Concept

1. Why Register?

Protocol registration establishes a public, time-stamped record of the review's methods before results are known. Its purposes are:

• Prevent duplication. Registers allow reviewers and funders to see which reviews are underway — avoiding wasted effort on questions already being addressed.
• Reduce selective reporting. Once methods are public, deviations become visible.
• Increase transparency. Readers can compare the protocol to the published review.
• Improve credibility. Registered reviews are more likely to be trusted by journals, funders, and policymakers.
• Enable coordination. Registries help networks of reviewers avoid duplicating each other's work.

Since the 2013 update of the Declaration of Helsinki and the 2015 PRISMA-P statement, prospective registration has been increasingly expected by journals and funders.

2. Major Registries

3. PROSPERO — The Primary Registry

PROSPERO (International Prospective Register of Systematic Reviews) is the largest and most recognized registry for health-related reviews. It is maintained by the Centre for Reviews and Dissemination at the University of York.

Key features:

• Free to register.
• Covers systematic reviews of health interventions, diagnostic accuracy, prognosis, and qualitative evidence.
• Requires the review to have a primary outcome and at least one health-related outcome.
• Reviews must be at the protocol stage — PROSPERO does not accept completed reviews.
• Typical turnaround: 2–4 weeks from submission to publication.

PROSPERO fields (main entries):

• Review title and timescale.
• Review team details (with contact person).
• Funding sources and conflicts of interest.
• Review question and objectives.
• Search strategy summary (search terms and databases).
• URL to search strategy (if hosted externally).
• Condition or domain being studied.
• Participants/population.
• Intervention(s)/exposure(s).
• Comparator(s)/control.
• Types of study to be included.
• Primary and secondary outcomes.
• Data extraction, risk of bias, data synthesis, subgroup and sensitivity analyses.
• Dissemination plans.

What PROSPERO does not accept:

• Scoping reviews.
• Literature (narrative) reviews.
• Rapid reviews (unless systematic).
• Reviews without a health outcome.
• Completed reviews submitted retrospectively.

4. OSF Registries

The Open Science Framework provides a general-purpose registry suitable for any discipline. It is maintained by the Center for Open Science.

Key features:

• Free; no strict subject-matter restrictions.
• Customizable templates — reviewers can create their own registration structure.
• Immediate DOI assignment on submission.
• Supports version-controlled updates.
• Integrates with OSF Projects for storage of data, code, and materials.

OSF is preferred when:

• The review is not health-related (e.g., education, criminal justice, environmental policy).
• The review is a scoping review or other review type PROSPERO doesn't accept.
• The team wants tighter integration with project files (data, code, forms).
• The team prefers immediate public availability (no 2–4 week wait).

5. Cochrane Library

Cochrane reviews are registered within the Cochrane Library itself. Registration requires:

• Affiliation with a Cochrane Review Group or Network.
• Approval of the review proposal by the relevant group's editorial team.
• Training in Cochrane methodology (either via the Cochrane Handbook or an approved training program).

Cochrane protocols are published as citable documents. The published protocol precedes the review and provides the reference point against which the final review is assessed.

6. Comparing Registries

7. Choosing the Right Registry

A decision framework:

1. Is the review a systematic review of a health-related question? → PROSPERO (or Cochrane if the team is affiliated).
2. Is it a scoping review or a review outside health? → OSF (or a discipline-specific registry such as Campbell for social sciences).
3. Is the review a JBI-affiliated nursing or allied health review? → JBI registry.
4. Is it a Cochrane review? → Cochrane Library (via Cochrane Review Group).
5. Is registration speed critical and PROSPERO's backlog is prohibitive? → INPLASY as an alternative; but note it is less widely recognized.

8. When to Register

The protocol should be registered:

• After protocol development — before any formal systematic search is run.
• Before data collection begins — screening, extraction, and appraisal should all occur after registration.
• Not retrospectively — registered protocols must be submitted prospectively; PROSPERO and OSF reject retrospectively-registered reviews.

In practice, teams sometimes begin scoping searches (informal, exploratory) before registration. This is acceptable — the scoping search informs the protocol but is not part of the formal review. Only the formal systematic search must wait for registration.

9. Updating a Registered Protocol

When amendments are made, the registration should be updated:

• PROSPERO: Log in and use the "Submit a revision" function. Changes are dated and stored; the public record retains both the original and revised entries.
• OSF: Create a new registration version, preserving the original as history.
• Cochrane: Contact the Review Group editorial team; the protocol amendment goes through editorial review.

Every update should be accompanied by a rationale, and the final manuscript should report deviations against the registered protocol.

10. What If the Review Cannot Be Registered?

Occasionally a review may not fit any major registry — e.g., a review of internal organizational data, or a narrative synthesis that doesn't meet PROSPERO's criteria. Options include:

• Use OSF with a self-defined structure.
• Deposit the protocol as a preprint (medRxiv, bioRxiv, Research Square) with a DOI — this provides a time-stamped public record even without a formal registry.
• Publish the protocol in an open-access journal (e.g., Systematic Reviews, BMJ Open).

Even in the absence of a formal registry, having a version-controlled, publicly accessible protocol is far better than having none.

🧠 Concept

1. Why Amendments Happen

Even with the most careful planning, real-world reviews encounter situations that require adjustments:

• Feasibility surprises. A database becomes unavailable; a paywall blocks full-text access; a country's registry is offline.
• Methodological clarification. Peer review of the search reveals missing synonyms; pilot screening exposes ambiguous criteria.
• New evidence. A relevant systematic review or landmark trial is published mid-review.
• Stakeholder input. Patient or policy advisors request consideration of additional outcomes.
• Team changes. A reviewer leaves; capacity for a prespecified analysis is reduced.
• Peer review feedback. Journal reviewers request an additional sensitivity analysis.

Amendments are a normal part of rigorous review conduct. The problem is not that amendments happen — it is when they happen without documentation or justification.

2. Legitimate Amendment vs. Methodological Drift

The key test: would you be comfortable if readers saw exactly what you changed and why? If yes, it's likely a legitimate amendment. If no, it's drift.

3. Categories of Amendments

Amendments can be classified by impact and by stage at which they occur.

By impact

• Clarifying amendments — sharpen existing criteria or specify an analytical detail. Low impact on results.
• Substantive amendments — expand or narrow eligibility, add or remove outcomes, change synthesis approach. Moderate to high impact.
• Structural amendments — change the review question or the review type. High impact — often warrant restarting or reframing the review.

By stage

• Pre-search amendments — before the formal search. Lowest risk; easiest to justify.
• Mid-search amendments — during screening, extraction, or appraisal. Must be transparent; often trigger rework.
• Post-synthesis amendments — after initial synthesis. Must be clearly justified; typically arise from peer review or stakeholder input.

4. Documenting Amendments

Every amendment requires a dated entry in three places:

The three records should be consistent with one another. A reader who compares them should see the same story.

5. Reporting Amendments in the Manuscript

PRISMA 2020 requires reporting of:

• The protocol registration details (item 24a).
• Any changes to the protocol with reasons (item 24a).
• Any changes to eligibility criteria, outcomes, or analysis methods (also item 24a).
• Whether amendments were made before or after data analysis (item 24b).

A typical amendment statement in a manuscript's Methods section:

"The original protocol registered with PROSPERO (CRD42025xxxxxx) specified a primary outcome of systolic blood pressure at 12 months. During screening, we identified that many trials reported only 6-month outcomes. With the agreement of all co-authors, and prior to data analysis, we amended the protocol to include both 6-month and 12-month outcomes, designating 12 months as primary and 6 months as secondary. The amendment was recorded in PROSPERO on [date] (revision 2)."

This level of detail is increasingly expected by journals and valued by readers.

6. Impact Assessment

Amendments should be evaluated for their potential to affect conclusions. Questions to ask:

• Does the amendment alter the scope of the review? (E.g., expanded population — results may be less applicable to original subgroup.)
• Does it change the primary outcome? (Higher risk of selective reporting; must be carefully justified.)
• Does it affect the search? (E.g., adding a database mid-conduct — increases comprehensiveness but may introduce temporal asymmetry.)
• Does it affect the analysis plan? (E.g., changing the model — must be justified by methodological reasoning, not by the results.)

Where an amendment has potential impact, consider reporting both the original and amended analyses as sensitivity analyses.

7. Amendments and Stakeholder Involvement

Stakeholders — patients, clinicians, policymakers — can be valuable in deciding whether an amendment is appropriate. Their input is particularly useful when:

• An outcome needs to be reframed to better reflect pati