Peer-review revision support

What the service is for

A decision of major revision is not a rejection, but it is a test, and many papers that deserve publication fail it because the response is mishandled. Authors may answer defensively, ignore a comment that seemed minor, run an analysis that creates new problems, or send a response that is hard for the editor to check against the revised manuscript. The task is partly scientific, because comments often require new analyses, and partly editorial, because the answer must be organized and courteous and must show what has changed.

This service supports that work. It helps authors understand what each comment is asking, decide how to address it, carry out any additional analysis, and present the response and the revision clearly. The authors remain responsible for their paper and for every statement in the response, as set out in the research integrity and authorship statement.

What the service includes

Understanding the decision

The decision letter comes first. It states whether the editor invites revision, what kind, and sometimes which comments matter most. Minor revision usually means the paper will be accepted if the points are handled satisfactorily. Major revision means that substantial changes, often new analyses or additional text, are needed, and the paper may still be rejected after the revision. Rejection with an invitation to resubmit differs from outright rejection, and some journals offer a transfer to a sister journal. The deadline for the revision matters, and extensions can usually be requested for good reason.

The editor's own comments take priority over those of any individual reviewer. When reviewers disagree, the editor's letter often says which view the journal favors, and where it does not, the authors can ask. The response should be written for the editor, who decides, as much as for the reviewers.

Analyzing the comments

Before drafting anything, each comment is separated and classified. Many reviews contain several points in one paragraph, and each needs its own answer. A comment may be a request for a new analysis, a question about the methods, a criticism of the interpretation, a request to add or remove material, a point of presentation, or a misunderstanding of the paper. The response differs accordingly. A request for an analysis needs the analysis, and the result should be reported whatever it shows. A question needs a clear answer and, usually, a change in the text so that the next reader does not ask it. A misunderstanding is not a reason to dismiss the comment, because it shows that the paper was unclear, and the fix is to make it clearer.

Comments are also sorted by importance. A few concern validity, such as the choice of model or the handling of missing data, and they decide the outcome. Many concern presentation. The plan should show that the first group has been taken seriously and the second dealt with efficiently.

Common statistical and methodological queries

Reviewers of evidence-synthesis papers raise a familiar set of points, and the answers can often be anticipated.

Choice of model

Why a fixed-effect or a random-effects model was used, which estimator, and whether the conclusions depend on it. The answer is a justification and a sensitivity analysis.

Heterogeneity

Why it exists, how it was examined and what the prediction interval shows. Subgroup analysis or meta-regression may be requested, and the answer should state how many studies support it and that it is exploratory when it is.

Publication bias

A request for a funnel plot or a test. The answer gives the result with the caveat that fewer than about ten studies limit it, or explains why it was not done.

Risk of bias and certainty

Requests for details of the tool, for sensitivity analyses excluding studies at high risk, or for a GRADE assessment.

Search and selection

Questions about databases, dates, languages and grey literature, or a request to update the search. An updated search is a substantial piece of work and is planned as such.

Multiplicity and post hoc analyses

Concern about many outcomes or analyses. The answer separates planned from exploratory analyses and does not present the latter as confirmatory.

Reporting

Missing PRISMA items, an absent registration number, unclear flow diagram numbers, and unreported data sources.

The guide on responding to reviewers covers typical queries and how to structure answers.

Additional analyses, done honestly

Reviewers often ask for analyses that were not in the protocol. These can be done, and should be, where they are reasonable, but their status must be stated. An analysis requested after seeing the results is post hoc, and the response and the manuscript should say so, and report the result whichever way it goes. It is not acceptable to run an analysis, find that it weakens the conclusion, and omit it, or to run several variants and report the best. Where a requested analysis cannot be done well, for example a meta-regression with too few studies, the response says why, and offers an alternative such as a descriptive comparison.

Every analysis added in revision is documented and its code is kept, so that the revised results can be reproduced. If the re-analysis changes a main result, that is reported and the abstract, the text, the tables and the conclusions are all updated, since a mismatch between the revised analysis and the old abstract is a frequent and avoidable error.

The response letter

The response letter is read by the editor and the reviewers, who want to see quickly that each point has been addressed. The usual form is a point-by-point reply. The letter opens with a short thanks and a summary of the main changes. Then, for each comment, it quotes the comment, gives the response and states what was changed in the manuscript, with the location, for instance the page and paragraph or the line numbers. Responses should be specific, courteous and short. A reviewer who is wrong is answered with evidence and without sarcasm, and with an acknowledgment of what in the paper may have caused the misunderstanding. A disagreement is stated respectfully, with reasons, and, if it concerns a matter of interpretation, the manuscript may be changed so as to acknowledge the alternative view.

Changes in the manuscript are tracked, and a clean version is supplied as the journal requires. The letter and the manuscript must agree, because an editor who finds that a promised change is absent loses confidence in the whole response.

Tone and conduct

Peer review is imperfect, and some comments are frustrating. A response should still be professional. Authors should not contact reviewers directly, or ask the editor to remove a reviewer without substantive grounds. Where a reviewer's comments appear biased, unprofessional or based on a factual error that the authors cannot resolve, the right course is to explain this to the editor courteously. COPE has guidance on the conduct of peer review for journals, authors and reviewers. We do not offer to influence a decision, to suggest false reviewers or to write replies that misrepresent what was done.

When the paper is rejected

Rejection is common and not final. The reviewers' comments, where there are any, are valuable, and addressing the valid ones before submitting elsewhere improves the next submission. The manuscript is reformatted for the next journal, and the choice of that journal follows the checks described under publication support. Where a decision seems to be a misunderstanding or a procedural error, many journals have an appeal process, which should be used only when there are real grounds, expressed courteously and with evidence. An appeal is rarely successful, and moving to a better-suited journal is usually the more productive course.

Deliverables

Optional extension: beyond the response letter

Limitations and boundaries

No revision is guaranteed to lead to acceptance. The decision lies with the editor, and reviewers may raise points that cannot be satisfied, for example a request for data that do not exist. We do not alter data or results to meet a reviewer's wishes, and we advise against promising analyses that cannot be done. If a comment exposes a real problem in the study, the honest response is to acknowledge it and to report it as a limitation, even if that makes the paper weaker. This service provides research and publication support and does not provide clinical advice.

How long does it take?

The time depends on how many comments there are, how many require new analysis, whether the data and code are available and organized, and the revision deadline set by the journal. A set of comments about presentation and reporting is a quick job, and a request for new analyses, an updated search or additional data collection takes much longer. Send the decision and the deadline as early as possible so that the plan can fit the time.

Frequently asked questions