You have deleted the author block from page one, saved the file as manuscript_final.docx, and you are about to attach it to an email. Before you do, open File > Info in Word on Windows, or File > Properties on a Mac.
The Author field almost certainly still says your name. Word wrote it there the day the document was created, and nothing you did to the visible text changed it. That is an ordinary way a double-blind submission stops being blind — a field nobody thinks to look at.
Our policies tell you to submit an anonymised manuscript, and then stop
The requirement is on our Peer Review page: our journals operate a double-blind model, and authors should anonymise their manuscript files accordingly. Author Guidelines tells you how to prepare and submit. What neither page contains is a procedure.
Blinding is a word-processing task, not a scholarly one, and nothing in your training covers it. What follows is the missing procedure. It is not policy.
What the editorial office actually checks before your manuscript reaches a reviewer
Our peer review page states what happens first: the editorial office checks the submission for completeness, scope, and adherence to ethical and formatting requirements. That is what the page commits to.
That is a completeness and formatting screen, not a de-anonymisation audit. Nobody opens your document properties, and nobody reads the manuscript hunting for identifiers. A visible problem may be returned to you for correction, but the screen is not designed to find one.
Similarity screening does not help either: it compares text against text, not identity. A manuscript can be wholly original and still name your hospital in the Methods — see how similarity screening works.
What we do not do: strip your metadata, rewrite your sentences, or anonymise the hospital in your ethics approval
We do not clean your files. No one runs Document Inspector for you, edits your PDF properties, or renames your file.
We do not rewrite identifying sentences. If your discussion says "in our previous trial at the Royal Infirmary", it stays as written.
We do not anonymise your ethics or registration statements. Those name a real committee and a real registry record, which is what makes them verifiable.
Split the submission into two files before you edit anything else
File one — the title page. Full title, author names in the agreed order, affiliations, ORCID iDs, corresponding author contact details, funding and grant numbers, competing interests, acknowledgements and author contributions.
File two — the blinded manuscript. Title, abstract, keywords, full text, references, tables, figure legends and supplementary material. No names, no affiliations, no acknowledgements, no grant numbers, no letterhead anywhere.
With two files you never choose between blinding a declaration and losing it. The corresponding author owns the split: see what the corresponding author is responsible for.
Strip the document properties: the exact click paths in Word, LibreOffice and Acrobat
Word: run Document Inspector after your final save
On Windows: File > Info > Check for Issues > Inspect Document. Tick Document Properties and Personal Information, run Inspect, then Remove All. Word for Mac has no Document Inspector — tick Remove personal information from this file on save under Word > Preferences > Security (Security & Privacy in current versions), then check File > Properties by hand. Do this after your final edit — Word rewrites Last Modified By on every save.
LibreOffice Writer: reset the properties, then clear the stored user data
Clear Tools > Options > LibreOffice > User Data first, then untick Apply user data in File > Properties > General and use Reset Properties on that same tab. The order matters: Reset Properties re-stamps the document with whatever name User Data still holds.
PDF: the Author field is inherited from the source file
Exporting sanitises nothing. In Acrobat — the paid application, not the free Reader — check File > Properties > Description and correct the Author field; Remove Hidden Information, an Acrobat Pro feature, goes further.
These tools reduce your exposure. No step here proves a file is clean or guarantees anonymity.
Tracked changes, comments and field codes carry your name after you have deleted it from the page
Revision marks store authorship per edit. Turning off Track Changes only stops new edits being recorded, and switching the review display to No Markup only hides what is already there — neither deletes the record. Accept or reject everything, delete every comment, then run the inspector.
Reference-manager field codes are the same problem: they embed library or account identifiers and a travelling copy of your reference records. In EndNote use Convert to Plain Text; in Zotero, Unlink Citations. Neither conversion can be reversed in the file that ends up without fields, so keep a copy with the fields intact — Zotero changes the open document, EndNote produces a new unsaved one.
A filename like Smith_KCL_trial_v7_final.docx defeats every other step.
Cite your own prior work in the third person without pretending the work does not exist
Deleting self-citations misleads reviewers about what is already established. Our guidelines prescribe no redaction convention — there is no instruction to write [Reference removed for blind review] — so third-person phrasing, with the reference kept in the list, is the safer choice.
Before: As we showed in our earlier cohort study (Author et al., 2022), the effect persists at 12 months.
After: An earlier cohort study reported that the effect persists at 12 months (Author et al., 2022).
Before: This paper extends our group's previously published protocol, used at our centre since 2019.
After: This paper extends a previously published protocol (Author et al., 2019) in use at a single tertiary centre since 2019.
The reference list is not the leak. A sentence that says "our group" in the Methods is.
Three leaks you cannot close — the ethics committee, the registry entry and the single-centre cohort
Some identifiers have to stay: removing them removes the reader's ability to verify the work.
| Identifier | Why it has to stay | What to do |
|---|---|---|
| Named ethics committee or IRB approval | Verifiable only if the committee and its reference number are named | Keep it. Note in the cover letter that it identifies your institution |
| Trial registry number | The ICMJE requires registration in a public registry before the first participant is enrolled, and the registry record itself is public | Keep it. Note that the entry names the sponsor and the site |
| Data or code repository link | Repository pages carry author names | Keep the link; use an anonymised review link where supported |
A fourth case no procedure fixes: a single-centre cohort in a narrow specialty can identify the group by subject matter alone. Identities are withheld under a stated policy, not enforced by software, and reviewers are not prevented from searching. Say so in the cover letter: see suggesting reviewers and how to write a cover letter to the editor.
Run the twelve-point checklist in the ten minutes before you send the file, not after it is returned to you
Run it on the final file. The last column is the honest one.
| # | The leak | What closes it | Checked by the office? |
|---|---|---|---|
| 1 | Names, affiliations, ORCID iDs, funding and competing interests inside the manuscript | Move to a separate title page file | Only that both files are present |
| 2 | Names, initials or letterhead in the body, headers or footers | Delete from the text and from every header and footer | May be noticed and returned |
| 3 | Acknowledgements naming colleagues, a department or a unit | Move the section to the title page file | Not checked; may be noticed on the page |
| 4 | Grant and funding identifiers that map to an institution | Move to the title page file with the funder names | Not checked; may be noticed on the page |
| 5 | Word document properties still holding your name | Windows: File > Info > Check for Issues > Inspect Document > Document Properties and Personal Information > Remove All. Mac: Word > Preferences > Security, tick Remove personal information from this file on save | No — not checked |
| 6 | LibreOffice properties and stored user data | Clear Tools > Options > LibreOffice > User Data, untick Apply user data, then Reset Properties in File > Properties > General | No — not checked |
| 7 | PDF properties inherited from the source file | Acrobat, not the free Reader: File > Properties > Description; correct the Author field | No — not checked |
| 8 | Tracked changes and comments, which store authorship per edit | Accept or reject all changes, delete all comments, re-inspect | Not checked; visible markup may be noticed |
| 9 | Reference-manager field codes, which embed library or account identifiers and a travelling copy of your reference records | EndNote: Convert to Plain Text. Zotero: Unlink Citations. Keep a copy that still has the fields | No — not checked |
| 10 | A filename containing a surname, institution or drive path | Rename to something neutral before attaching | No — not checked |
| 11 | Figures and supplementary files carrying institutional watermarks, patient-record headers, scale bars stamped with a department name, or an ORCID or email inside an image | Open every image at full size, re-export without the stamp, and strip the image file's own metadata — EXIF and equivalent fields can hold a camera or scanner identifier, an account name, or a device or institution name | No — not checked |
| 12 | Self-citations written in the first person | Rephrase in the third person, keeping the reference | Not checked; may be noticed on the page |
Ten minutes, once. The alternative is finding out from a reviewer.
The requirement and the screen that follows it both sit on Peer Review; preparation and submission are under Author Guidelines. See also writing a cover letter to the editor, how similarity screening works, and the corresponding author's responsibilities.