Can Preprints and Published Versions Accidentally Be Counted Twice?

Could a Preprint and Journal Article Be the Same Study?

You find a preprint during your search. Later, you retrieve a peer-reviewed journal article with a similar research question, overlapping authors, and familiar-looking results. The title has changed, the sample size is slightly different, and several analyses have been revised. Do you have two studies?

Possibly not. PRISMA 2020 explicitly recognizes preprints and journal articles as types of reports that can provide information about an underlying study. A study can have multiple reports, so the existence of two separately accessible documents does not establish two independent investigations.

This matters in systematic reviews because accidentally treating a preprint and its subsequent journal publication as independent studies can duplicate the underlying participants and give one investigation more influence than intended.

Yes, Unless You Link the Versions at the Study Level

Why Preprints Create a Particular Duplicate-Study Problem

A Preprint and Published Article Can Be Different Reports of One Study

PRISMA 2020 distinguishes a study from the reports that communicate information about it. Its definition of a report explicitly includes both journal articles and preprints, alongside conference abstracts, study-register entries, dissertations, clinical study reports, and other documents.

This distinction is useful because a preprint and a journal publication are genuinely separate documents. They may have different titles, dates, digital object identifiers, formatting, and even substantive content. Yet they can still describe one underlying investigation.

Ordinary reference deduplication is designed primarily for the first problem. Study-level linkage is needed for the second.

Do Not Require an Identical Title

A manuscript can acquire a different title between preprint posting and journal publication. Authors may sharpen the framing, change terminology, or revise the emphasis during peer review and editorial processing.

Exact-title matching is therefore insufficient. Cochrane recommends using multiple study characteristics when determining whether reports belong to the same study, including study identifiers, authors, locations, interventions, sample sizes and baseline characteristics, and study dates.

Authorship Can Change Between Versions

The journal article may add or remove authors or change their order. A matching author list is helpful evidence, but it is not a necessary condition for linking reports.

Instead, look for a broader pattern. Shared investigators combined with the same participant population, recruitment period, intervention or exposure, outcome definitions, and study identifier provides much stronger evidence than authorship alone.

Sample Sizes May Change Without Creating a New Study

A preprint might report 850 participants and the journal article 812. That difference does not automatically mean the papers describe separate studies.

Changes may result from data cleaning, revised eligibility decisions, additional exclusions, updated follow-up, correction of errors, or a change in the analysis population. Conversely, recruitment may have continued after an early preprint, producing a larger final sample.

Cochrane specifically cautions that multiple reports of the same study can contain different participant numbers. Sample size is therefore one part of the study fingerprint, not a unique identifier.

Use Persistent Study Identifiers Whenever Available

Trial registration numbers, protocol identifiers, cohort names, study acronyms, ethics identifiers, grant numbers, and other study-specific identifiers can provide strong linkage evidence.

A shared trial registration number is particularly useful when titles and author lists have changed. Still, inspect the actual analysis because one version may report a subgroup, later follow-up, or other modification of the underlying study.

Check Whether the Preprint Server Links to the Published Article

Some preprint platforms display links to subsequently published versions when those relationships are known. Likewise, the journal article may cite the preprint or use a title, registration number, or protocol that connects the documents.

These explicit links can make identification straightforward. Their absence does not prove independence, however. Reports from the same study do not always cross-reference one another.

Do Not Assume the Published Version Is Numerically Identical

Peer review and revision can change analyses, tables, interpretations, participant counts, outcome definitions, and conclusions. A journal article should therefore not merely replace a preprint in your database without checking whether information used in your review has changed.

Compare the versions for features that matter to your synthesis:

The Journal Article Will Often Be the Main Source, but Not by Automatic Rule

A later peer-reviewed article will often contain fuller and more stable reporting than an earlier preprint. That makes it a natural candidate for the main report in many reviews.

Still, publication status alone should not become an inflexible hierarchy. A preprint might contain an analysis, appendix, outcome, or methodological detail absent from the final article. Cochrane's broader principle for multiple reports is to collate them and preserve useful secondary information rather than simply discarding every source except one.

If you need to designate one main source, apply the same principles used to identify the primary report of a study.

Version Replacement and Study Deduplication Are Different Decisions

You may decide that the journal article supersedes the preprint for a particular result. That is a source-selection decision. It does not mean the preprint never existed or that its relationship to the study should disappear from your audit trail.

Keeping the reports linked lets you document why an earlier value was replaced and helps explain discrepancies encountered during extraction.

Be Particularly Careful When the Preprint and Published Article Both Enter Screening

Searches conducted across bibliographic databases, preprint servers, citation indexes, and general scholarly search platforms can retrieve both versions. If screening is performed independently at the report level without subsequent study linkage, both documents can progress through eligibility assessment as though they represented different studies.

Cochrane's MECIR standards require multiple reports of the same study to be collated so that the study, rather than each report, remains the unit of interest.

Publication Does Not Create New Participants

If a preprint reports 600 participants and its journal version reports those same 600 participants, publication in a journal does not convert the sample into another 600-person study. Even if the analysis changes, the underlying participant provenance remains relevant.

This is the same conceptual problem encountered with conference abstracts and subsequent journal publications.

When a Preprint and Journal Article Look Similar but Not Identical

Common Mistakes With Preprints and Published Versions

Track the Version Relationship Explicitly

When a preprint and journal article appear related, compare them before final extraction. Assign both reports to the same study record when the evidence supports that relationship and document which version supplies each result used in the review.

The goal is not to privilege one publication format reflexively. It is to ensure that multiple reports from one study remain one underlying study while preserving whatever information the reports legitimately contribute.

How to Check a Preprint Against a Published Article

Questions About Preprints and Published Versions

Publication Version Is Not Study Identity

How to Cite This Guide

This guide is intended to be read, shared, and used in research, teaching, and academic work. If you draw on its ideas, explanations, or other content, please acknowledge the source by citing the guide. Doing so gives appropriate credit and helps your readers locate the original resource.