Should You Include Preprints in a Literature Review?

Should research count before it has passed journal peer review?

You find a study that directly answers your research question. It has recognizable authors, a complete methods section, results, tables, references, and perhaps even underlying data. There is one important difference from the other papers in your review: the manuscript is a preprint.

Should you include it?

Preprints complicate a familiar shortcut in literature reviewing. Researchers often use journal publication as a convenient boundary around the literature, but relevant research can now circulate publicly before conventional peer review. This can make preprints particularly valuable for emerging topics, where waiting for journal publication may leave a review several months behind current research.

The trade-off is equally important. A preprint is not the version of record. It may be revised, replaced by a later version, subsequently published in a journal, or never reach journal publication at all. The question is therefore not simply whether preprints are "reliable." It is whether preprints are appropriate evidence for your particular review and how their provisional status should affect searching, appraisal, citation, and interpretation.

Preprints can be included, but their status needs to remain visible

Preprints provide early evidence, not necessarily final evidence

What exactly is a preprint?

A preprint is a scholarly manuscript made publicly available before formal journal peer review and publication. Authors commonly deposit manuscripts on preprint servers so that research can be disseminated rapidly, discussed, cited, and discovered without waiting for the complete journal publication process.

Preprints have long been established in some disciplines and have become increasingly visible in others. Different research communities use different servers, conventions, and terminology, so the role of preprints in scholarly communication varies considerably among fields.

A preprint can look remarkably similar to a journal article. It may contain a title, abstract, introduction, methods, results, discussion, references, tables, and figures. What distinguishes it is primarily its publication status rather than its visual appearance.

That distinction should remain explicit whenever preprints contribute evidence to your review.

Why would you include preprints in a literature review?

The clearest advantage is speed. Journal peer review, revision, production, and publication take time. Preprint dissemination can make findings available much earlier.

This matters most when the evidence base is changing rapidly. Cochrane guidance identifies preprints as a potentially relevant source of study evidence, particularly for emerging topics where relatively little evidence exists. It also emphasizes the importance of identifying multiple reports of eligible studies because study information may be reported differently across sources.

Preprints can therefore help you:

• identify recently completed research that has not yet reached journal publication;
• reduce the time lag between study completion and evidence synthesis;
• identify additional reports of studies already known to the review;
• observe earlier versions of analyses or outcomes;
• broaden a search beyond conventionally published literature;
• follow rapidly developing research areas more closely.

The value is not that preprints are inherently better or more complete than journal articles. It is that publication delay should not automatically determine whether potentially relevant research is visible to you.

Preprints are especially relevant in rapidly developing fields

The faster a research area changes, the greater the possible gap between the work researchers are currently doing and what has completed journal publication.

This became particularly visible during the COVID-19 pandemic, when preprints were used extensively to communicate emerging biomedical evidence. The same principle can apply more broadly to fast-moving topics such as artificial intelligence, emerging technologies, new infectious diseases, rapidly changing policy environments, and other areas where a review limited to formally published literature may become outdated quickly.

That does not mean every fast-moving field requires preprints. It means the cost of excluding them may be greater when timeliness is central to the review's purpose.

Not every literature review needs preprints

A literature review can legitimately restrict eligibility to peer-reviewed publications when that boundary fits its purpose. For example, a review may specifically examine the peer-reviewed evidence base, or an assignment, journal, institution, or protocol may impose such a restriction.

Other reviews may have little to gain from preprints because the topic is mature and the relevant evidence is largely stable and conventionally published.

The important point is to make the boundary deliberate. "I did not search preprints because I did not think about them" is methodologically different from "The review was intentionally restricted to peer-reviewed journal evidence for a stated reason."

Preprint does not mean unreviewed by anyone

It is common to describe preprints as "not peer reviewed," which is useful shorthand for saying that they have not completed conventional journal peer review. But that wording should not be stretched too far.

A preprint may have received comments from colleagues, undergone internal review, been discussed publicly, received community feedback, or been revised after posting. Some preprint platforms also conduct screening before posting, although screening should not be confused with journal peer review.

Conversely, public availability does not guarantee that anyone with relevant expertise has critically examined the work.

The safest description is therefore specific: the manuscript has not yet completed conventional journal peer review unless evidence shows otherwise.

A preprint is not automatically unreliable

The absence of journal peer review creates uncertainty, but empirical comparisons do not support a simple rule that preprints are generally contradicted once they reach journal publication.

For example, a study comparing 547 medRxiv preprint-journal article pairs describing clinical trials, observational studies, or meta-analyses found that 81.1% had numerically concordant primary endpoint results and 96.2% had concordant study interpretations. The authors concluded that most clinical studies subsequently published in journals had consistent study characteristics, results, and interpretations across versions.

Other studies have similarly found that many changes between preprints and published papers are relatively minor and do not alter the main conclusion.

That evidence is reassuring, but it needs careful interpretation. Such comparisons generally concern preprints that subsequently became journal articles. They cannot establish the reliability of every preprint, including manuscripts that never proceed to publication. The findings also come disproportionately from particular disciplines and periods, including intensive research on COVID-19.

Preprints can change, sometimes substantially

Although many preprint-publication pairs remain broadly consistent, changes do occur.

A meta-research study of COVID-19 intervention studies found that among studies represented by multiple preprint versions or both a preprint and journal article, 45% had a change in at least one examined evidence component. Changes included numerical results and abstract conclusions, and 21% had changes classified by the researchers as important according to their predefined criteria.

Another systematic review of 119 COVID-19 randomized trial preprints with subsequent journal publications found differences in outcomes, analyses, results, or conclusions in 55% of pairs, although the main conclusion changed in only two studies.

The appropriate takeaway is neither "preprints usually change" nor "preprints never change." Rather, a preprint is a versioned research object whose details may evolve. That possibility should influence how you manage it in a review.

Always check which version of the preprint you are reading

Preprint servers commonly allow authors to upload revised versions. A manuscript may therefore have a first version, one or more revisions, and eventually a journal publication.

Versions may differ because authors:

• correct errors;
• add participants or follow-up;
• revise analyses;
• clarify methods;
• add or remove outcomes;
• respond to feedback;
• change interpretation;
• update figures, tables, or references.

When using a preprint, record the version and date accessed where appropriate. If the server provides a persistent identifier or DOI, use the citation information associated with the specific record rather than constructing your own reference from incomplete metadata.

Search for a peer-reviewed version before including or citing the preprint

Before treating a preprint as the current report of a study, determine whether it has subsequently been published.

Start with the preprint record itself. Many servers link a preprint to a published article when the relationship becomes known. Then search:

• the preprint title;
• distinctive title phrases;
• author names combined with topic terms;
• study acronyms or registration numbers;
• scholarly databases and search engines;
• the authors' publication profiles.

Do not assume the journal article will have exactly the same title. As with conference abstracts and working papers, titles and author order can change between versions.

If a peer-reviewed version exists, the next question is usually which version you should cite and use, rather than whether you should count both as separate studies.

A preprint and its journal article are usually two reports of one study

This point is crucial for evidence synthesis. If the same study appears first as a preprint and later as a journal article, you normally have two reports of one underlying study, not two independent studies.

Counting both separately could duplicate participants and distort a narrative or quantitative synthesis.

Link versions using:

• authors;
• study design;
• participants and setting;
• sample size;
• intervention or exposure;
• outcomes;
• study dates;
• registration numbers;
• funding information.

Where versions differ, preserve those differences rather than silently treating the documents as interchangeable.

What if the preprint and published paper disagree?

Most differences will not necessarily alter the overall interpretation, but some can matter. A later article may contain a different sample size, revised outcome definition, corrected analysis, changed statistical significance, or modified conclusion.

If you discover a consequential discrepancy, investigate what changed rather than applying an automatic rule such as "peer-reviewed always wins." The later journal article will generally be the version of record, but understanding the change can matter for assessing the study and explaining why earlier evidence differed.

When the preprint and published paper report different results, compare the versions systematically before deciding what information belongs in your synthesis.

Peer review does not eliminate the need for critical appraisal

A common mistake is to frame the choice as "unreliable preprint versus reliable peer-reviewed article." Peer review is valuable scrutiny, but it does not guarantee correctness.

Whether a study is a preprint or journal article, examine the research design, sampling, measurement, analysis, missing data, reporting, limitations, and potential conflicts using appraisal criteria appropriate to the study.

For a preprint, publication status adds another layer of uncertainty. It should not replace methodological appraisal.

You need answers to both.

Preprints can complicate risk-of-bias assessment when information changes

Because manuscripts can be revised, methodological information available in one version may differ from later reports. A later version might clarify randomization, add missing-data information, correct an analysis, or provide additional outcomes.

If a preprint is included in a systematic review and a journal version appears later, reassess extraction and appraisal rather than simply changing the citation.

The evidence has potentially changed, not merely its bibliographic packaging.

Preprints can help reduce publication delay, but they do not eliminate publication bias

Preprints make some research visible before journal publication, which can reduce dependence on the conventional publication timeline. They may therefore broaden the accessible evidence base.

But not every study is posted as a preprint. Researchers choose whether to use preprint servers, and those choices may vary by discipline, institution, result, career stage, or study type. Some studies remain completely undisclosed.

Searching preprints is therefore one possible component of efforts to identify research that has not yet reached conventional publication. It is not a complete solution to missing evidence or publication bias.

Consider the stakes of the decision

How cautiously you use preprints should partly depend on what your review will be used for.

A scoping review mapping emerging research may reasonably include preprints because identifying what research exists is central to the question. A rapidly updated systematic review may also need them to remain current.

By contrast, a review supporting a high-stakes clinical, regulatory, or policy decision may require especially careful treatment of evidence that has not completed peer review. That does not necessarily mean automatic exclusion. It means the uncertainty associated with publication status should be explicit and weighed alongside the methodological strength of the study.

The same preprint can therefore be appropriate evidence in one review and inappropriate in another without either review being inconsistent.

Decide your preprint policy before you see the findings

For systematic reviews, eligibility rules should be specified prospectively whenever possible.

Your protocol can state:

• whether preprints are eligible;
• which preprint sources will be searched;
• how different versions will be handled;
• whether a peer-reviewed version will replace or supplement preprint information;
• how preprint status will affect appraisal or certainty assessment;
• when publication status will be rechecked.

This reduces the temptation to include a preprint because its result is interesting while excluding another because its result is inconvenient. Evidence-synthesis methods tend to work better when the rules arrive before the p-values.

How to handle a preprint in an emerging research area

Common misconceptions about using preprints

Make preprint inclusion a methodological choice, not an instinctive one

The appropriate decision depends on what your review is trying to accomplish, how quickly the evidence changes, and what claims you intend to make.

Whatever boundary you choose, apply it consistently. A preprint should not become eligible because its result supports your argument, nor should it become inadmissible because its result complicates an otherwise tidy narrative.

When preprints are included, make their status visible enough that readers can distinguish provisional manuscripts from formally published versions without having to conduct bibliographic archaeology themselves.

Before including a preprint in your literature review

Frequently asked questions about preprints in literature reviews

Include preprints when they fit the review, but preserve their provisional status

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.