ICSR PERSPECTIVES ICSR The Need for Speed How quickly do preprints become published articles? Highlights bioRxiv’s option for authors to submit biology preprints directly to journals typically speeds up publication of articles by nearly 2 weeks—an advantage for authors who are keen to get their work published quickly Authors approach ‘preprinting’ in different ways: while most post a preprint before submitting to a journal, just over a third of preprints were submitted to and accepted by a journal before the preprint appeared on bioRxiv Biology preprints with just one version are published 2 weeks more quickly than those with multiple versions 2019 SEPTEMBER Introduction How quickly do preprints become published The dataset: 8,711 bioRxiv preprints articles? What benefits are there for authors considering posting a preprint and for those people • posted between 2013 and 2017 who determine journal preprint policy? That’s what • all matched in bioRxiv to a published journal article we aim to learn in this study, by discovering a little • publication dates retrieved from CrossRef more about the timing of submission of preprints • journal title information retrieved from Scopus and the relationship between submission and • additional publication data for select journals acceptance of manuscripts. directly retrieved from journal websites Preprints in scholarly with the fact that it launched in the publication date (from CrossRef ). Our communication last 10 years and publishes preprints findings here are consistent with those in subject areas that have only reported in other similar studies: 134 ASAPBio defines preprints as quite recently, but enthusiastically, days (Inglis & Sever, 2016) and, later, “unpublished draft[s] of a research started engaging in this form of 166 days (Abdill & Blekhman, 2019). paper” (Inglis & Sever, 2019) and scholarly communication, makes it a However, the publication date of an in recent years, there has been a fascinating and useful way to look at article is as much linked to a journal’s sharp rise in the number of preprint trends in behaviors around preprints. publishing schedule as it is to the servers, the variety of research areas readiness of an article, so this data they serve (Rawlinson & Bloom, From preprint to publication could be hiding some other trends 2019; OSF Preprints), and the from view. The version history of To consider the basic overall timeline sheer number of preprints posted these preprints, and the relationship first, the median amount of time it (PrePubMed, accessed 2019). This between bioRxiv and journals are takes for a preprint to be published expansion and proliferation of digital also relevant factors influencing is 160 days (Figure 1): that’s from the activity around new scholarly works publication timelines. date a preprint is first published to the has given rise to examinations of the nature of ‘preprinting’—from the subject areas experiencing the most growth in preprint numbers— such as the life sciences, psychology Time elapsed between bioRxiv preprint and journal publication: and the social sciences (Narock & the direct transfer journal eect Goldstein, 2019) — to the positive MEDIAN INTERVAL DAYS correlation found between preprint download activity and the Journal Impact Factor of the journal in which All preprints: rst preprint submission to publication the final published paper appears (Abdill & Blekhman, 2019). For this study, we considered 8,711 Direct Transfer preprints: rst preprint submission to publication preprints on bioRxiv, the preprint server for papers on biology. The preprints were posted between 2013 Non-direct Transfer preprints: rst and 2017 and had to be matched preprint submission to publication in bioRxiv to a published journal article. Additional data on publication Figure 1: median interval between date of bioRxiv preprint and date of publication in a journal, dates were taken from CrossRef and comparing preprints in direct and non-direct transfer journals select individual journals. bioRxiv’s Preprints in direct transfer journals: n=3,438 | Preprints in non-direct transfer journals: n=5,251 advanced functionality, combined Note: 22 preprints were missing the necessary journal-level data for this analysis 2 ICSR PERSPECTIVES ICSR Speeding up submissions Count of versions of preprints Preprint count Share of total count of preprints 1 5,872 67% bioRxiv has a feature that is likely to play a part in the speed of 2 1,991 23% publication: some journals allow 3 611 7% direct manuscript submission via the 4 237 3% platform. When uploading a preprint, TOTAL 8,711 100% these partner (or ‘direct transfer’) Table 1: preprints by the count of versions uploaded to bioRxiv journals enable authors to directly submit a manuscript, without having to visit a different website or fill out Preprints upon preprints When authors replace preprints with new forms (bioRxiv.org, accessed June new versions, it could be a signal that 2019). If authors submitting to these bioRxiv also allows users to update various forms of peer review or other journals were playing Monopoly, their preprints and so authors feedback are occurring. It’s possible they’d be advancing to ‘Go’ and sometimes post multiple versions of that preprints are uploaded and collecting $200 right away. manuscripts. The preprint webpage then edited as the research study or shows the history of each preprint and discussion continues, or as a result Submission to one of the 160+ direct indicates the date that each version of feedback from colleagues or other transfer journals does not guarantee was posted. Most of the preprints in contacts, other bioRxiv users, and/or publication, and publication in our study (67%) had just one version journal peer reviewers. a direct transfer journal does available across the five-year period not necessarily signify that the (Table 1). Of the 2,839 preprints with Whatever drives the need to update corresponding author actually took multiple versions, the majority (70%) preprints, there is a benefit of sorts: advantage of the direct submission have two versions posted and 611 the time from latest available preprint function. Nevertheless, preprints preprints have three versions. to publication is over three weeks that were published in direct transfer shorter than that between first journals were published more Some interesting outliers reveal the preprint submission and publication quickly than those in other journals: range of multiple preprint version (Figure 2). But when we add in time just under 2 weeks more quickly activity: one preprint had no fewer spent working on additional versions (Figure 1). So, when bioRxiv states than 19 versions posted across an (7.7 weeks median average) — overall, that authors who use this option to approximately 9-month period; preprints with just one version on submit a preprint directly to a journal another had an impressive 2 years, bioRxiv are published fastest. They are “save time”… they really mean it. 28 days between the first and latest published about two weeks sooner available versions. than those with multiple versions. Do researchers simply need to get their written work right first time?! Time elapsed between bioRxiv preprint Seemingly, uploading a single MEDIAN INTERVAL DAYS and journal publication version of a preprint is the quickest ‘route’ to journal publication, after All preprints: rst preprint all. Well, maybe in an ideal world. submission to publication But this analysis still doesn’t yet All preprints: latest preprint give us the full picture. We need submission to publication more information about the point of submission to a journal to truly Single version preprints: preprint submission to publication understand the potential advantages of iterating a manuscript on bioRxiv Multiple version preprints: rst preprint submission to publication before publication. Multiple version preprints: latest preprint submission to publication Figure 2: median interval between date of bioRxiv preprint and date of publication in a journal All preprints: n=8,711 | Single version preprints: n=5,872 | Multiple version preprints: n=2,839 3 What do publication dates hide? Journal Count of (preprint) publications bioRxiv direct transfer journal? Scientific Reports 557 For a fuller picture, we need to look beyond article publication dates. As Plos One 388 any submitting author knows, the time eLife 327 taken from submission to publication PNAS 292 can be… lengthy. The availability of Nature Communications 283 reviewers and the speed at which Bioinformatics 262 they work, the number of rounds and PLOS Computational Biology 234 complexity of the reviews, and the journal’s publishing schedule all play Nucleic Acids Research 149 a part in constructing the timeline. So TOTAL 2,492 — if we want to understand the benefits for authors in terms of publication Table 2: the four largest direct transfer and four largest non-direct transfer journals and their count of preprint publications speed, we need more information—in particular, dates that manuscripts were submitted and accepted. In a helpful the time from ‘manuscript received’ There’s no doubt, then, that preprints move toward greater transparency, to ‘article published’ for the two are just that—versions of articles journals now typically publish those groups of papers. released before publication (in line dates on their websites. with bioRxiv’s policy)—but authors As shown in Figure 3, we found vary with respect to how far in advance Eight journals publishing