Preprints: a game changer in scientific publications?

Abstract

The landscape of scientific publications has always been a moving field. Dissemination of medical and scientific information has been boosted in the digital era with the advent of Internet which has won the race for priority and exclusivity to traditional journals. Recently, preprints have gained grounds to conventional peer-reviewed journals as means to disseminate scientific advances. A preprint is a complete version of an original manuscript posted by the authors to an open-access server before formal peer-review process. This allows full manuscripts to be made public immediately to stimulate general and scientific discussion and obtain feedback before formal peer-review and final publication in a classical journal.1–7 This phenomenon has represented a major paradigm shift in the scientific process. Of note, the use of preprints has been boosted in the last 3 years because of the challenges generated by the coronavirus disease 2019 (COVID-19) pandemic. Preprints permit claiming provenance of an idea allowing scholars to obtain author’s rights. Preprints offer intellectual protection against scooping because a timestamp of the original manuscript can be demonstrated to claim priority.3 Preprints receive a Digital Object Identifier (DOI) and are actively cited. Most large preprint servers do not permit the authors to remove preprints since they are part of the scholarly record. Standard practice is that when the final version of the paper is published the authors add the full reference of the published paper to the preprint.1–4 Of note, preprints on COVID-19 with US National Institutes of Health (NIH) support have been indexed in PubMed Central (PMC), as a pilot protocol launched by NIH. However, information in preprints lacks the scientific scrutiny and validity of an expert editorial and external peer-review process.1–7 Preprints are considered as preliminary reports; hence they should only be used with the knowledge that the findings have not been validated by peer-review, and the media should acknowledge this limitation when informing the public. Franz Ingelfinger, a former editor of the New England of Medicine, indicated that the Journal would not accept a manuscript if it had previously been published elsewhere. This ‘editorial mantra’ has prevailed until very recently.2 In 1961 the NIH launched a preprint programme aimed to share biological research.1 This caused major scientific and editorial discussions and most high-ranked biomedical journals expressed serious concerns on the consequences of the initiative that, eventually, was discontinued. However, the NIH remained involved in promoting preprints allowing preprints to be cited in grant applications.8,9 In 2018, the International Committee of Medical Journal Editors (ICMJE) issued recommendations on preprints which have recently been updated.10 In the updated Wellcome open-access policy, researchers on public health will be required to post their preprints at the time of journal submission. Preprints may be used by public funders to ensure open-access of the information to the public.3 Preprints can always be obtained openly whereas the final paper may have been published in a journal that is only available to subscribers. In this regard, preprints are more transparent and democratic. Preprints have moved the field to a new model where each work will stand on its own merits rather than the prestige provided by the journal where it is published.9 To survive in this new editorial landscape, journals should select only the best primary information and critically evaluate and weight the novel findings, providing the adequate scientific context and perspective to their readers.1–4 The aftermath of this editorial revolution remains unclear. Main differences between conventional papers published in Journals and preprints are summarized in Table 1. Main differences between scientific journals and preprint servers Two major preprints servers for biomedical studies, bioRxiv and medRxiv, succeeded in being accepted and gradually adopted by the scientific community.8 They provided well-organized servers that quickly established partnerships with respected journals that agreed to accept manuscripts already posted as preprints and allowed simultaneous submission of papers both to the preprint server and to the journal.3 Preprints have a long history in other scientific disciplines.1–6 However, the problem in biomedical research is that poor quality or flawed studies can harm patients as treatment strategies may be changed in response to faulty information.11 The idea that the public should be ‘protected’ from misleading information by the classical journals might be biased: first, by the interest of editors willing to maintain their status as guarantors of scientific information2 and, second, by the relevant economic gain obtained by the publishers and journal owners. Preprint servers initially were supported by scientific and non-governmental organizations.1 Classical journals, initially reluctant to accept studies previously posted as preprints, progressively changed their policies to accommodate to a more transparent and global perception of the scientific communication.3 Journals initially requested that authors notified the presence of a posted preprint; then some journals encouraged authors to include the preprints in the citation list. Subsequently, many journals started offering to post preprints in their publisher-owned preprint servers, simultaneously with the submission of the paper to the journal.1 This drastic change in editorial policies has taken place in just a few years, a journey drastically accelerated by the pandemic. During the worst pandemic in a century the scientific community has been deluged with an unprecedented increase in the amount of scientific information. Rapid release of information has been considered crucial to timely inform general health public policies. Accordingly, non-peer-reviewed pathways to release scientific information flourished.5 The pandemic demonstrated the advantages of preprints for rapid dissemination of information and providing expediency and transparency to clinical research. However, the pandemic also disclosed the dangers of disseminating false information and articles on COVID-19 have been more frequently retracted than other types of articles.12 It is important to keep in mind that, despite a thorough peer-review process, classical journals occasionally also publish information that subsequently must be retracted. The scientific community embraced preprints and harnessed social media tools to improve dissemination of novel information. The pandemic has been a major crisis but has also provided a unique opportunity to revolutionize the dissemination of scientific research.13 The requirements and quality standards performed by the different preprint servers vary widely.1–5 General editorial recommendations are limited and only a minority of servers suggests following the recommendations of the ICMJE10 or Committee on Publication Ethics (COPE).1 MedRxiv includes the following warning on its home page: ‘preprints are preliminary reports of work that have not been certified by peer-review. They should not be relied on to guide clinical practice or health-related behaviors and should not be reported in news media as established information’.1 The Lancet proposed a list of items to systematically assess the potential problems of preprints.6 More recently the ICMJE proposed that the authors should choose preprint servers that have the following characteristics: (i) clearly identify preprints as work that is not peer-reviewed; (ii) require authors to document disclosures of interest; (iii) require authors to indicate funding source(s); (iv) have a clear process for preprint archive users to notify archive administrators about concerns related to posted preprints—a public commenting feature is desirable for this purpose; (v) maintain metadata for preprints that are withdrawn from posting and post-withdrawal notices indicating the timing and reason for withdrawal of a preprint; and (vi) have a mechanism for authors to indicate when the preprint article has been subsequently published in a peer-reviewed journal.10 Although initially the medical community was very cautious preprints progressively gained medical attention and some even got more citations than the corresponding peer-review articles.6 However, the credibility of preprints that never are published in reputable peer-reviewed medical journals has been questioned. In addition, preprints may cause outside investigators to double count the results, which is of concern in meta-analyses.4 The lack of quality control may favour ‘salami slicing’ initiatives to obtain a higher number of publications.3 The ‘comments’ generated by preprints have also generated attention.1–6 Preprints allow authors to receive feedback from a wide range of scientists but also from a broader and diverse audience.3 However, the amount of peer-review style comments found in the comment section of preprints servers is usually limited.1,14 Preprint servers still face many technical challenges that prevent them achieving their full potential. There is no clear way to track all the updates and changes performed in the original version of the manuscript.1–6 It is obvious that keeping a record of the changes remains critical to understand the evolution of the scientific process.3 The final journal should be able to identify the preprint and consider this as part of the prepublication scholarly record.12–17 Retractions of preprints do not leave a record disclosing that the study findings have been challenged or disputed,14 although the Retraction Watch Database and PubMed have been used to explore the retraction of COVID-19 and non-COVID-19 articles and preprints. Withdrawal of preprints remains a major concern in relation to the increasing use of preprints.15 New mechanisms should be developed to link the final journal publications unambiguously and consistently to previous versions of the work in preprint servers.13–17 Regarding concordance, most studies suggest that there are few meaningful differences between preprints and their corollary journal-published papers.16 Changes may be the result of journal peer-review but also of comments directly generated by the preprint. The marginal difference between preprints and peer-reviewed journal publications supports the idea that preprints should be considered as scientifically valid contributions. The ICMJE recommend that the authors should: (i) inform a journal if the work submitted has been posted on a preprint server; (ii) provide a link to the preprint to indicate in the text of the manuscript that a preprint is available and how reviewers can access that preprint; (iii) ensure that preprints are amended to point readers to subsequent versions of the work, including the published article. In addition, authors should not post in the preprint archive the published article nor interim versions that are produced during the peer-review process that incorporate revisions based on journal feedback. Finally, when a preprint article has been subsequently published in a peer-reviewed journal, authors should cite the subsequent published article rather than the preprint article whenever appropriate.10 Preprints are here to stay. However, preprints can only accelerate scientific discovery if they are credible. We are facing a critical moment where editorial efforts should focus in benefiting from the unique opportunities offered by preprints but, at the same time, in improving their quality to overcome their current limitations. A consensus is required on the recommendations and quality standards required for the reporting and posting of preprints in a server. Many advocate for editors and ethical groups to develop quality control, editorial and ethic recommendations to equally prevent misconduct and misinformation in preprints and biomedical journals. Internationally recognized stakeholders, including the World Health Organization, should issue recommendations for preprints, formulating principles and policies to maintain the integrity of the scientific process.18

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 14 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.