Scientific Review Process
A step-by-step account of how a review should check evidence, resolve concerns, and document its scope.
Page updated
- 01Define & trace
- 02Evaluate & resolve
- 03Record & reassess
A review should make the reasoning clearer and preserve uncertainty where the evidence remains incomplete.
This is the intended review workflow. External review remains pending unless a verified reviewer and completed review are recorded on the article.
- Set the scope
Define the question and review scope
A review begins with an identifiable article version and an agreed scope. The question may concern the entire article or a specific mechanism, study interpretation, clinical outcome, or regulatory statement. A limited check should not be presented as approval of material outside that scope.
This page describes the review standard and intended workflow. It does not assert that every published page has completed external review. The status on the individual article determines whether a documented reviewer has examined it.
- Match the expertise
Match expertise and check conflicts
The assigned reviewer should have experience appropriate to the material. Relevant identity, qualifications, and affiliations should be verified before publication of an attribution. Reviewers should disclose relationships that might influence the work.
The editor assesses whether a conflict can be managed through disclosure or requires another reviewer. Scope, expectations, and attribution should be clear before the review begins, so a contributor’s limited involvement does not become an exaggerated endorsement.
- Check the sources
Trace the claims to the original material
The review should connect important statements to the sources that actually support them. Check study populations, research setting, intervention details, comparators, outcomes, and reported results. Ensure a citation is not being used to support a broader claim than the publication made.
Where possible, compare trial reports with their registry entry or protocol. Distinguish a primary study from a synthesis or commentary, and identify whether multiple reports refer to the same participants. Unavailable full text or incomplete reporting should remain an explicit limitation.
- Appraise the findings
Evaluate methods and interpretation
Study design provides a starting point for appraisal. The reviewer examines relevant issues such as comparison groups, allocation methods, missing data, selective reporting, follow-up, precision, and applicability to the question. A design label alone should not settle confidence.
The interpretation should distinguish statistical findings from clinically meaningful outcomes, associations from causal claims, and biomarkers from benefits important to patients. Findings from cells or animals must remain identified as such throughout the summary and illustrations.
- Examine safety & status
Check safety and regulatory context
Safety discussion should accurately represent the available evidence and its gaps. The review should consider whether limitations are visible and whether the text implies reassuring certainty that the available reports cannot support.
Regulatory claims should be traced to official documentation for the relevant product, use, authority, and jurisdiction. A trial registration, published experiment, or discussion of a compound is not interchangeable with authorization for a particular clinical use.
- Resolve the concerns
Return comments and resolve concerns
Review feedback should identify the statement at issue, the reason for concern, and a supporting source or suggested qualification. The editor should address material comments and record the resolution before representing the review as complete.
If the evidence does not resolve a disagreement, the claim may need to be narrowed, qualified, omitted, or held pending further information. The aim is an accurate account of the evidence, rather than a stronger conclusion produced by compromise.
- Record the review
Record completion and publish attribution
Completed review should be tied to the version examined, with the reviewer’s verified name, relevant credentials, date, and scope available to readers. Important conflicts should accompany the attribution. An editorial citation check must retain a distinct label from external clinical review.
Before publication, inspect the headline, summary, tables, captions, and evidence descriptions as well as the body. A careful qualification in one paragraph does not correct an unsupported promise elsewhere on the page.
- Keep it current
Reassess after publication
New findings, source corrections, retractions, or material changes to an article may require another assessment. The existing review attribution should not silently extend to revised clinical claims that the reviewer has not examined.
Reader reports follow the Corrections Policy. The Evidence Methodology explains claim classification, while the Medical Review Policy defines the attribution requirements. These processes support one another, but represent different editorial activities.