What are preprints?
- Preprints are research manuscripts posted prior to peer review. Depending on the preprint server, manuscripts may be screened for privacy concerns or potential harm. For instance, “All manuscripts uploaded to medRxiv undergo a basic screening process for offensive and/or non-scientific content and for material that might pose a health risk”.
- Preprints are not peer reviewed, and they are not final research products. The NIH offers guidance about “Preprints and Other Interim Research Products”, such as datasets and code.
- Preprints can enable timely discussion of research, especially in rapidly evolving fields. The publication of research via journal articles is often delayed by the formal peer review process.
Where can I find preprints?
- Preprints can be found on preprint servers, including medRxiv, bioRxiv, ResearchSquare, and SSRN (including Preprints with The Lancet). Searching each server individually can be time-consuming. Search across the contents of many servers on EuropePMC, which includes preprints.
- There are also several tools to help researchers locate COVID-19 specific preprints, including a preprint filter in The COVID-19 Portfolio.
- Preprints are also included, on a limited trial basis, in PubMed, and can be located using keyword searches and adding AND "Preprint" [Publication Type].
- For more information on finding preprints, see the new Himmelfarb preprint guide
How do I evaluate preprints?
- Because preprints have not undergone peer review, they require more critical analysis for potential issues that might be caught during peer review.
- Members of the scientific community engage with preprints in a variety of ways. Some platforms, such as F1000Research, post reviewer comments alongside the article. However, preprint reviews and critiques are rarely so easily located. Often, further searching will be necessary to find cross linking to social media discussions about a preprint.
- Researchers can find reviews and critiques of COVID-19 related preprints on review sites and overlay journals, including NCRC from Johns Hopkins, and Rapid Reviews: COVID-19 from MIT. Generally, while review sites link to preprints, preprint servers do not link to reviews.
- Outbreak Science Rapid PREreview offers a browser plug-in that alerts readers to reviews of preprints: for instance, this September preprint has a review.
- SciScore uses AI to detect indicators of rigor and reproducibility in a manuscript. Follow @SciScoreReports for updates on COVID-19 related preprints evaluated with this tool.
I found a preprint and reviewed the discussion around it. Now, how do I cite it?
- Since the rapid dissemination of COVID-19 research via preprints, the need for clarity in citing preprints in reference lists has grown more urgent. Where possible, citations should include the version and preprint status; this is specified in Vancouver Style. Some preprint servers provide no version control, though others, for instance those using OSF Preprints or Jupyter Notebooks, do include version control.
- The DOI should also be included, as well as the server name, to assist readers in identifying and locating the preprint cited. Note, medRxiv provides versioning and DOIs. If you cannot find the preprint‘s DOI, search the preprint title in crossref.org or preprint servers.
- For more information on citing preprints and potential problems, see this Scholarly Kitchen blog post.
What should I be aware of when reading and citing preprints?
- Across the research landscape, there is an increasing focus on transparency and reproducibility, encouraging authors to share interim research products, including preprints, datasets, and computer code. (Note: Data sharing is mandated for federal grant recipients.) Locating and analyzing research data and code may be especially important when appraising preprints, which have not undergone peer review. Preprint servers have varying abilities and infrastructure to link to research data. Read the full text of the manuscript to find mention of, or, better yet, links to, datasets and code on repositories, such as figshare, dryad, GitHub, and Zenodo. You can look for subject data repositories on the website re3data.org. Often, datasets link to preprints, but preprints do not reliably link out to the data.
- Before citing a preprint, check if the research has been published in a peer-reviewed article since the preprint’s posting. While some servers link to published articles, there may be a time lag or other technical reasons that prevent automatic linking. Authors are encouraged to ensure links are posted from preprints to articles, but this is not 100% reliable. Try searching the preprint title (though titles may change between posting and publishing) in PubMed, Google Scholar, or CrossRef to check for a published, peer-reviewed article.
- And remember, any changes to the standard of care need to be based on authoritative evidence. Do not change guidelines based on evidence found only in a preprint. https://blogs.gwu.edu/himmelfarb/2020/06/17/revisiting-medrxiv-in-the-age-of-covid-19/
For more information about preprints in general and to learn how and where to upload a preprint of your own: see the Himmelfarb preprint guide.
