The central conclusions of biomedical preprints rarely change after peer review in a journal.¹according to a study published on the preprint server bioRxiv this month. The research also found that studies that first appeared as preprints are retracted at about half the rate of articles that did not appear online before being in a peer-reviewed journal. The authors say the findings suggest that preprints are a reliable source of information, although some scientists say the finding should be interpreted more cautiously.

Publishing preprints is common practice in science today, but Ruslan Rust, a neuroscientist at the University of Southern California in Los Angeles, says he often hears his scientific colleagues say that they are unreliable. In his experience, peer review does not usually result in major changes to the content of a study. Rust wanted to see if this was true in different fields of biomedical research.

Using a large language model (LLM), Rust and his colleague extracted the main scientific conclusion from the abstracts of 72,644 biomedical manuscripts that were first uploaded to bioRxiv between 2018 and 2025. The model then assessed how much the abstracts had changed compared to their eventual peer-reviewed versions. The study, which has not been peer-reviewed, reports that 39.9% of the main conclusions remained unchanged between the preprint and the abstract published in the journal, another 50% underwent only minor revisions. Just over 10% experienced significant changes.

When conclusions changed, they were more likely to become more cautious than more confident after peer review, the study found. About 8.4% of the main findings adopted more cautious language after peer review, while 4.2% used more confident wording.

The scope of the review also varied across disciplines. Significant changes occurred in only 7.2% of bioinformatics articles compared to 17.5% of microbiology studies.

The authors also found that the frequency of major revisions decreased over time, falling from 17% among articles published in 2019 to 5.7% in 2024.

Julian Sienkiewicz, who studies artificial intelligence and data exploration tools at the Warsaw University of Technology, says the decline in major reviews over time could indicate that peer reviewers are overloaded and may not be reading articles in depth.

Rust suggests that the decline reflects a change in how people use preprints. In the years before bioRxiv was launched, particularly during the COVID-19 pandemic, scientists were under pressure to publish their findings in a short time. This meant that many articles had to undergo major revisions before publication, he added. In recent years, some manuscripts may have already included reviewer comments in the first preprint version that was published online, he adds.

Fewer retractions

Rust and his colleague also found that articles that first appeared as preprints were removed at a rate of 8.1 per 10,000 articles, compared with 18.7 per 10,000 comparable articles that had never been published as preprints.

The authors caution that the comparison is observational, based on relatively few retractions, and does not prove that publishing an article as a preprint reduces the likelihood of a retraction.

Reacting to the findings on the professional networking platform LinkedIn, some researchers noted that preprints are subject to strong selection bias depending on who publishes them and what studies are published.