OBJECTIVES: Research is increasingly being published via preprint servers in advance of peer-reviewed publication. Databases such as PubMed and Embase also include publications from preprint servers; as such systematic literature review (SLR) searches are increasingly likely to also capture preprint publications. There are no recommendations on whether it is appropriate to include preprints in SLRs, therefore we sought to understand the conversion rates for clinical trial preprints to peer-reviewed publications to determine the scale of the potential issue.

METHODS: Ten disease indications were randomly selected, including viral infections, oncology, respiratory, and endocrinology. A search was conducted in Embase.com to identify clinical trial preprints published in January 2023 in each indication. A subsequent search was conducted to determine whether the preprints had been published by 31-May-2024 by searching for title matches between preprints and peer-reviewed publications.

RESULTS: Across the ten disease indications, 282 preprints were published in January 2023, ranging from 7 to 128 preprints depending on indication. By 31-May-2024, 22/282 (27.3%) of the preprints had an associated peer-reviewed publication. Hepatitis had both the lowest number of preprints and lowest conversion to peer-reviewed publications (1/7, 14.3%). The greatest conversion rate was observed in breast cancer where 6/16 (37.5%) preprints were peer-review published. In indications with the largest number of preprints (COVID-19 [34/128, 26.6%], diabetes mellitus [9/40, 22.5%], dementia [10/28, 35.7%], HIV [6/24, 25.0%]) conversion rates ranged from 22.5% to 35.7%.

CONCLUSIONS: The use of preprint servers can offer a valuable rapid dissemination of research; however, the apparent low rate of subsequent peer-reviewed publications over a year after preprint dissemination does beg the question of whether it is appropriate to include preprints in systematic literature reviews and evidence synthesis. Further research determining reasons why preprints are not published is required.