Usually, no—systematic reviews that use published, non-identifiable data don’t need ethics approval.
If your review stays within published studies and public datasets with no identifiers, most institutions treat it as outside human-subjects review. Once you introduce contact with people, private records, or individual-level identifiers, a review board may be needed. The guide below maps common cases and the paperwork that journals and funders tend to ask for.
Ethical Approval For Systematic Reviews — Common Rules
In the United States, the federal definition of human-subjects work centers on interaction with living people or access to identifiable private information. A desk-based synthesis that extracts numbers and text from published papers generally sits outside that definition. In the UK, the Health Research Authority (HRA) sets similar boundaries for Research Ethics Committee (REC) review and provides decision tools to reach the same call.
| Scenario | Typical Requirement | Notes |
|---|---|---|
| Review of published studies with aggregate data only | No IRB/REC review in many settings | Desk-based synthesis; no contact with participants; no identifiers. |
| Meta-analysis using published group-level outcomes | Usually no ethics approval | Still cite sources, manage conflicts, and pre-register when suitable. |
| Use of public, de-identified datasets | Often not human-subjects research | Confirm the dataset is public and de-identified per local rules. |
| Contacting authors for extra, non-identifiable numbers | Frequently no formal review | Keep communications professional; store correspondence securely. |
| Access to individual patient data with identifiers or re-identification risk | IRB/REC review usually required | Data sharing agreements and approval pathways apply. |
| Extracting data from medical records or non-public registries | IRB/REC review expected | Check privacy law, data controller rules, and approvals. |
| Embedding a new survey or interview within the review | IRB/REC review required | Now you’re collecting fresh human data, which changes the status. |
| Classroom or training exercise that won’t be published | Often outside IRB scope | Pedagogic activity; no plan to generalize. |
Why Many Reviews Don’t Need An Ethics File
Reviews synthesize evidence that already passed ethics checks at the study level. You’re not intervening, surveying, or handling identifiers. Under the U.S. Common Rule, work that doesn’t involve living people or identifiable private data isn’t human-subjects research. In the UK, REC review centers on projects that involve participants or confidential patient data.
What Counts As Human-Subjects Work
Three flags change the answer fast: interaction with people, access to non-public records, and receipt of identifiable data. If any appear in your plan, speak with your local board. Even when your team records data without identifiers, the source might still be private or coded in a way that carries re-id risk.
Cases That Usually Need Review
Some designs cross the line from synthesis into human-subjects territory. The items below are the most common triggers.
Individual Participant Data (IPD) Meta-analysis
IPD work uses row-level datasets. If files include names, full dates, or linkable codes, expect to go through your IRB or REC. Even when partners strip direct identifiers, linkage across fields can carry risk.
Access To Non-Public Records
Hospital charts, registry back-ends, and private claims files are classic triggers. A review that mines such sources moves beyond public information, so most boards will want a submission.
New Data Collection Embedded In The Review
Teams sometimes add a small survey to fill gaps, such as reaching authors for unpublished harms. Once you collect answers for research use, you’re interacting with people. That typically requires review, consent language, and a storage plan.
Journal, Funder, And Registry Expectations
Editors and sponsors don’t all ask for the same paperwork, yet many want to see that you checked the ethics path. A clean way to show due care is to register the review, document your decision, and cite the rules you followed.
Plan, Register, And Show Your Method
Pre-registering on PROSPERO or an institutional registry lays out eligibility, outcomes, and analysis up front. Keep extraction forms and decisions in a versioned repository. When you write the methods, state whether human-subjects review applied and on which rule you based that call.
Document The Determination
Many institutions provide a self-cert form for “not human-subjects research,” or a quick route for an exemption letter. Store the PDF with your protocol and add a one-line note so reviewers see it.
Country-Specific Quick Guides
United States. Teams often rely on the federal decision charts to decide if a project involves human subjects and whether it needs IRB review. These charts sit under the Common Rule (45 CFR 46). They include pathways for projects that use public data only and for projects that record data without identifiers. Read the HHS OHRP decision charts for the exact branches.
United Kingdom. The HRA provides two online tools: one for deciding if a project is research, and one for deciding if it needs REC review. A desk-only synthesis with no participant involvement usually doesn’t go through REC review, while projects that use confidential patient data or recruit people do. The HRA’s REC decision tool walks you through that call.
Grey Areas And How To Handle Them
Real projects sit between clear boxes. When in doubt, map your plan to the three flags: interaction, non-public sources, and identifiable data. Then remove or reduce each one.
Emailing Authors For Clarification
Asking a corresponding author to confirm a number or share a missing standard deviation doesn’t usually involve people as subjects. You’re not collecting personal data about the author for research use.
Data From Public Portals
If the data are public and stripped of identifiers, teams in the U.S. often file a self-cert and move on. If the portal requires an application, data use agreement, or contains fields that could re-identify people, stop and seek advice before download.
Linkage Risk
Even without names, rare diseases, small hospitals, and full dates can allow linkage. Drop or bin such fields, or ask the provider for a safer cut.
Step-By-Step: Build A Clean Ethics Trail
- Define the design. Synthesis only, IPD meta-analysis, or use of non-public sources?
- Run the right tool. U.S. federal charts or UK HRA tools; save a PDF of the last branch.
- Log the decision. Complete self-cert or request an exemption letter if offered.
- Register. Post the protocol and keep your search and screening logs.
- Harden data practice. Store files in controlled folders and keep an audit trail.
- Write it plainly. Add one line on the review path in the methods.
Second Table: Quick Reference For Editors And Reviewers
| Project Feature | Usual Status | Proof To Provide |
|---|---|---|
| Desk-only synthesis of published group data | Outside IRB/REC in many settings | Method statement; registry link; citation rules followed. |
| IPD meta-analysis with identifiers or linkable codes | IRB/REC path expected | Approval letter; data use agreement; security plan. |
| Use of non-public health records or registry back-ends | IRB/REC path expected | Approval letter; controller sign-off; access logs. |
| Author contact for missing numbers only | Often outside review | Email log; no personal data collected for research. |
| Class or training assignment with no publication | Often outside review | Course brief; no intent to generalize beyond class. |
One-Page Checklist You Can Reuse
Scope
State review type, sources, and whether any new data will be collected.
People
Confirm no interaction with participants unless a reviewed sub-study exists.
Data
Note whether data are public and de-identified; remove linkable fields where possible.
Governance
Run the national tool, complete self-cert or exemption steps if offered, and file proof.
Reporting
Register, share your search and extraction forms, and add a clear line in the methods about the ethics path.
Bottom Line For Authors
A desk-only synthesis rarely needs ethics review. Projects that touch people or private records do. Run the national tool once, file proof, and add a plain one-line statement in your methods.