Preprints versus peer-reviewed papers: a reader guide

This guide helps a reader tell a preprint from a peer-reviewed paper. It matters when a claim about an emerging technology rests on a preprint, because the claim may not have passed formal review.

What a preprint is

We use medRxiv, a preprint server for health sciences, as the example in this guide. Its FAQ says that "Manuscripts submitted to medRxiv are not certified by scientific peer review, edited, or typeset before being posted online" (medRxiv FAQ). Other preprint servers set their own policies, so check the rules of the server you are reading.

Five checks

1. Is there a journal version? The medRxiv FAQ says that medRxiv "will usually automatically add a link to the published version within approximately 2 weeks of journal publication" (medRxiv FAQ). Look for that link on the preprint page. If there is none, search by title before deciding that no journal version exists.

2. Does it say it is peer reviewed? The medRxiv FAQ says its manuscripts "are not certified by scientific peer review" (medRxiv FAQ). If a news story calls a preprint peer reviewed, look for the journal version and read that instead.

3. Read the DOI record, not just the prefix. A DOI's metadata records what kind of work it is. Crossref's record for one DOI starting 10.22541 types it as "posted-content" with a "preprint" subtype, and names Wiley as publisher (DOI 10.22541/au.174183685.53366196/v1). A journal article record names its journal and article number. Crossref's record for one Science Advances paper lists "Science Advances" as the journal and article number eadu6915 (DOI 10.1126/sciadv.adu6915).

4. Has it been updated? The medRxiv FAQ says "a manuscript posted on medRxiv can be revised at any time until it is published in a journal" (medRxiv FAQ). It also says a revision "will be posted under the same DOI", while substantially changed content submitted as a new manuscript "will be assigned a new DOI" (medRxiv FAQ). Compare the version you read with the latest one, and note which version you used.

5. Has it been withdrawn? The medRxiv FAQ says "medRxiv's policy is that preprints cannot be removed." For a withdrawal, the FAQ says a statement explaining the reason is posted on the manuscript page, and a "Withdrawn" watermark is added to the PDF of all posted versions (medRxiv FAQ). Check both the PDF and the page.

How Vrulum labels them

Vrulum's methodology page says "Preprints are labelled as preprints." It also says "Every significant factual statement links to a specific source" (Vrulum methodology). We suggest the same test for any article: a claim that rests on a preprint should say so and link to it.

A practical rule

Treat a preprint as a claim under review. Use it as a lead, not as the basis for a statement of fact, until a journal version exists or the full text has been checked against the claim.