For Reviewers

01

Before you accept

Read the title, abstract and the editor’s note, then answer three questions before replying:

• Can you judge it? Take the parts you are qualified to assess and say which parts you are not. A review that covers the clinical reasoning and states that the statistics need a specialist is more useful than one that covers everything thinly.
• Are you independent? Decline where you are a recent co-author, work at the authors’ institution, supervise or are supervised by an author, hold a financial interest in the outcome, or are working on a directly competing paper.
• Can you do it in the time agreed? If not, decline promptly and suggest someone who can. A fast decline is more valuable than a slow acceptance.

Declining costs nothing and is always acceptable; suggesting an alternative reviewer is a real help to the editor.

02

Confidentiality

A manuscript under review is confidential and unpublished. Do not share it, show it to a colleague without the editor’s agreement, discuss it outside the review, or use anything in it in your own work before it is published.

Where you involve a trainee in the review as a learning exercise, tell the editor first, name them, and take responsibility for the report.

03

Reviewing a double-blind manuscript

Single-blind by default; double-blind review is available on request. Some manuscripts therefore reach you anonymized, and most reach you with the authors named.

If you recognize the work anyway, say so to the editor and let them judge whether you should continue.

04

What a review is for

The journal asks you to judge whether the methods answer the question, whether the analysis supports the result, and whether the conclusions stay inside what the evidence shows. That is the whole of it.

Scope eligibility has already been decided before the manuscript reaches you. A study is not weaker because its result is negative, its design is descriptive or observational, its contribution is methodological, or it replicates or validates rather than discovers. Please do not ask for a novelty claim, a mechanism, a clinical endpoint, a particular setting or a particular population as a condition of publication — the journal does not require any of them.

Where you believe the subject sits outside the journal’s remit, say so to the editor as a scope observation rather than folding it into a quality judgment.

05

What to examine

Question and design

Is the question clear, and can this design answer it? Are the inclusion and exclusion criteria, the comparator and the outcome defined before the analysis?

Methods

Enough detail to repeat the work? Assays, instruments, kidney function measures and estimating equations named with versions and sources?

Analysis

Test appropriate to the data, assumptions checked, missing data handled and described, effect sizes with confidence intervals rather than p values alone, post-hoc and subgroup analyses labeled as such?

Results and figures

Do the numbers in the text, the tables and the figures agree? Are the figures legible, and is there any sign of duplication, splicing or selective adjustment?

Conclusions

Supported by the data presented, with the limitations stated rather than argued away?

Reporting and ethics

Reporting guideline followed and checklist attached? Ethics approval, consent, trial registration and the data-availability statement present?

Literature

Does the manuscript engage with the work that bears on it, and are the citations accurate and primary?

06

Writing the report

Write for the authors as well as the editor. Open with a short summary of what the paper does, in your own words, so the authors can see what a careful reader took from it. Then number your points, separate the essential from the optional, and give the reason for each: a point an author can act on changes the paper, and a point without a reason does not.

Keep the tone about the work rather than the people. Where you are uncertain, say you are uncertain. Confidential remarks to the editor go in the confidential field, not in the report the authors will read.

07

Your recommendation

Accept

The work is sound and complete as it stands.

Minor revision

The conclusions hold; clarification, additional detail or presentation changes are needed.

Major revision

The question is worth answering and the data may support an answer, but analysis, additional work or substantial rewriting is needed first.

Reject

The design cannot answer the question, or the conclusions cannot be repaired by revision.

The recommendation is advice to the editor, who weighs it with the other reports and takes the decision. Editorial decisions are based on scope, scientific quality, methodological rigor, ethical compliance, reporting quality, and relevance to the journal.

08

Concerns you should raise, not resolve

Tell the editor privately, and let the journal handle it, where you suspect plagiarism or duplicate publication, see signs of image manipulation or fabricated data, doubt that ethics approval or consent was obtained, or believe an author list is wrong. Say what you observed and what made you doubt it; the journal investigates in confidence following COPE guidance and gives the authors the opportunity to respond. Do not contact the authors yourself.

09

Using AI tools

This applies to every externally hosted tool, not only public ones. The judgment in a review must be yours.

10

Reviewing a revision

Read the authors’ point-by-point response against your original report and against the revised manuscript. Judge whether each point was addressed, and accept a reasoned scientific argument for not making a change as an answer. A revision is assessed on what it now shows, not on how closely it followed the instructions.