Posting a Medical Preprint: medRxiv Rules and Risks

A medical preprint puts unreviewed findings in front of journals, clinicians and journalists at once, and most of the choices you make when posting cannot be undone.

Guide · 10 min read · Directive Publications Editorial Team

Why a medical preprint needs more thought

In many fields, posting a preprint is a low-stakes way to share work early. In clinical and health research it carries an extra risk: a finding that has not been through peer review can reach clinicians, patients and journalists who read it as settled.

medRxiv is built around that risk. Its submission guide says preprints "should not be relied on to guide clinical practice or health-related behavior and should not be reported in news media as established information." Every posted preprint carries a statement that it has not been certified by peer review, and the PDF is stamped to say so.

The server is free to use and is run by openRxiv, a non-profit. It screens every submission, but screening is not peer review. As senior author, you still own three decisions no screener can make for you: whether your target journal will accept a preprinted paper, whether the work is ready to be public permanently, and how you will handle attention once it is.

Check journal preprint policies before you post

Posting first and checking later is the most expensive way round.

The ICMJE Recommendations state that posting a preprint "may influence a journal's interest in or priority for peer review and publication of that work," and ask journals to describe their preprint policies in their instructions for authors. Many medical journals consider preprinted manuscripts, but the details differ. When you read a policy, look for three things:

• Whether preprints are accepted at all, and whether the server or licence matters to the journal.
• What you may post during review. ICMJE advises authors not to post the published article, or interim versions that include revisions made in response to journal feedback. medRxiv allows revisions until journal publication, so the journal's rule is the one that actually limits you.
• Rules on media contact before publication. These vary between journals. If the policy is silent or unclear, ask the editorial office in writing and keep the reply.

medRxiv's FAQ points authors to Jisc Open Policy Finder and a Wikipedia list of journal policies as starting points, and then to the journal's own website. Treat the journal's site as the authority. If you are still deciding where to submit, add preprint policy to the checks in our guide to choosing the right journal.

Timing matters too. The submission guide says medRxiv will not post a manuscript that has already been accepted for publication, published, or posted elsewhere. Post before acceptance or not at all.

When you do submit to a journal, disclose the preprint. ICMJE asks authors to tell the journal and provide a link, whether the preprint went up before submission or during review, and suggests noting in the manuscript itself that a preprint exists. Give the DOI in your cover letter as well.

What medRxiv screens for, and why it declines submissions

medRxiv describes two screening stages. In-house staff check the submission details, the article type and compliance with its policies. Volunteer clinicians and health professionals then consider whether the manuscript is health research and whether posting it could cause public harm. The medRxiv FAQ says screening typically takes 2 to 4 days, and longer over weekends and holidays or when a submission needs further discussion. medRxiv also reserves the right to decline any preprint for any reason, but most declines trace back to one of the problems below.

The article type is out of scope

medRxiv posts research articles with new data, research design protocols, systematic and scoping reviews with meta-analyses, and bibliometric studies. It does not post case reports or case series, narrative reviews, editorials, letters, opinion pieces, hypotheses without new data, guidelines, policy statements, preclinical work, or protocols for systematic reviews. If your unit writes up interesting cases, route those to a journal that publishes case reports instead.

The ethics statement is missing or comes from the wrong body

medRxiv requires a statement of oversight by an independent body, such as an ethics committee or IRB, whenever human participants, samples or data were used, unless the data are already openly available. That includes anonymized data, surveys, service evaluations and controlled-access datasets. It does not accept decisions from online tools such as the HRA Decision Tool, audit bodies, Caldicott Guardians, or research and development departments. If your study was classed locally as audit or service evaluation, find out before you submit whether an independent committee ever reviewed it.

A trial or interventional study is not registered

Clinical trials and other prospective interventional studies need a trial ID from an ICMJE-recognised registry, whatever the cohort size, and even if the trial has not started. medRxiv defines interventions broadly, including behavioural treatments, educational programmes, quality improvement interventions and process-of-care changes. If a study is unregistered, medRxiv will ask you to register it retrospectively. It also strongly recommends avoiding causal language unless a robust causal inference framework justifies it, and says papers with inappropriate causal language may be declined.

A patient could be identified

Photographs that show a face or other identifying features are declined, and blurring or pixelation is not enough. Direct identifiers include names, initials, patient or sample IDs, precise ages, and exact dates or locations linked to individuals. medRxiv also asks for fewer than three indirect identifiers, such as age range, ethnicity, occupation or family details. Signed consent does not solve this, because the server says it cannot verify whether consent was obtained.

The content could endanger public health

Screeners look for dual-use research of concern and for work that challenges or could compromise accepted public health measures on infectious disease transmission, immunization and therapy. Submissions like these can be escalated for discussion before any decision is made. Every manuscript is also checked for plagiarism and for material already online in a journal or on another preprint server, and it must be written in English.

Prepare the statements before you upload

The submission guide asks for more than a manuscript file. Have these ready before anyone starts the upload:

1. Ethics declaration naming the approving body, with confirmation that required consent was obtained
2. Trial registry and registration number, stated in the manuscript
3. Competing interest statement that says whether any author or their institution received payments or services from a third party in the past 36 months that could be perceived to influence the work
4. Funding statement covering all funding that supported the work
5. Data availability statement, with links to any external repository
6. Agreement from every author to the deposit

Do not skip the last item. The submission guide requires that all authors have consented to the deposit, and a trainee who posts without a co-author's agreement creates a permanent public record that person never approved. Make written agreement a routine step in your group.

Decisions that are hard to undo

You cannot delete a preprint. medRxiv preprints receive DOIs and are indexed by Google Scholar, Crossref, Europe PMC and others, and preprints of NIH-funded studies are also indexed in PubMed. If you no longer stand by the findings, you can withdraw. A statement giving the reason appears on the preprint page, the original stays accessible in the article history, and the PDFs are watermarked "Withdrawn." medRxiv says removal has happened only in extremely rare cases, for legal or safety reasons.

Every version stays visible. A revision posts under the same DOI and earlier versions remain in the history tab. A reader who saved version 1 keeps version 1. When a key estimate changes, say so plainly in the revised text rather than letting readers find the difference themselves.

The licence can only loosen. You choose from CC BY, CC BY-NC, CC BY-ND, CC BY-NC-ND, CC0, or no reuse without permission. medRxiv now lets authors update the licence on preprints posted within the last two years without a new version, but only to a less restrictive licence. If your funder or target journal expects a particular licence, choose it at submission. Missing funder information, by contrast, can now be added after posting.

Preprint risks in clinical research

A useful decision rule: if a result could change what a clinician does for a patient, ask whether the benefit of sharing it now outweighs the harm if it changes during review. Posting early and waiting can both be defensible, provided the senior author chooses deliberately.

Handling media coverage of a preprint

medRxiv's published screening procedure says manuscripts cannot be posted at a specific date or time or embargoed to a requested schedule, so you cannot line a press release up with the posting moment. The FAQ urges journalists to emphasise that preprints have yet to be evaluated by the medical community and may contain errors. You can make that easier for them.

• Brief your press office before you submit, not after the preprint appears.
• Make sure any press release, institutional news item or social post says "not peer reviewed" near the top, names medRxiv, and links to the preprint.
• Write one plain limitations sentence a reporter can quote: the design, the sample, and what the study cannot show.
• Decline to turn findings into advice for patients, and point questions about individual care back to clinicians.
• Check the target journal's policy on media contact before you give interviews.
• If coverage overstates the results, contact the outlet with a short, specific correction, and clarify the preprint itself in a revision if its wording invited the misreading.

A posting sequence for senior authors

Pin this order to the lab's submission checklist:

1. Check the journal's preprint policy, including revisions and media contact.
2. Get every author's agreement in writing.
3. Confirm the article type is in scope.
4. Prepare the required statements.
5. Remove identifying information from text, figures and supplements.
6. Choose the licence deliberately.
7. Submit, allow for screening, and tell the journal.
8. Keep the record current after posting.

After the journal publishes

medRxiv says it usually adds a link to the published article within about two weeks and emails the corresponding author to confirm it. If no link appears after two to three weeks, contact medRxiv or leave a comment on the preprint. Do check, because ICMJE places the responsibility for pointing readers from a preprint to later versions on the authors, not on the journal's editors.

Then tidy your own records. List the preprint and the published article as one output on your CV and ORCID record, not two, and cite the published version once it exists. Our guide to preprints, accepted manuscripts and versions of record covers how to cite and share each version.

If you plan to submit to a Directive Publications journal, read the author guidelines before you post, and send any preprint questions to the editorial office through the contact page.