AMWA-EMWA-ISMPP joint position statement on medical publications, preprints, and peer review

The public could lose trust in science if scientific and medical researchers choose to bypass the traditional high standards of peer-reviewed medical journals in the rush to get research data released, particularly during crises such as the COVID-19 pandemic. The statement recognizes the benefit of rapid publication, but alerts against the potential harms associated with an accelerated process. Recommendations are provided for key stakeholders, including a Reviewers’ Checklist to provide a minimum standard of pre-publication vetting to enhance preprint publication processes.


Background
In medical publications, just as in research and development, quality depends on the expertise and integrity of researchers/authors as well as qualified peer reviewers and journal editors. However, the laborious and time-consuming process of the traditional peer review 1 can be compromised by the pressure to publish quicklyparticularly during a health crisis, when the timely distribution of credible medical information can make a substantial difference 2 . Recent examples of negative consequences are two articles on COVID-19 that were hastily published in high-profile medical journals and subsequently retracted 3,4 .
Traditional peer review, although not perfect, remains the most frequently used process for vetting scientific publications. However, it has become more common for manuscripts to be released without prior review, which raises new concerns.
The potential value of rapid publication should be weighed against the potential harm of inadequate validation of the final output. There is a danger that lowering the threshold of publication oversight sets a precedent that cannot be easily reversed, potentially eroding standards and public trust in medical science 2 .
We have joined in a multi-party consortium among three eminent professional organizations for medical communication professionals -AMWA, EMWA, and ISMPPto advocate for the adoption of standards by all stakeholders to better ensure the integrity of published scientific and medical information. Thus, the following Joint Position Statement has been developed to provide practical and implementable suggestions to uphold data integrity and quality, and the transparency of medical publications.
Note: We use the term "medical writer" to represent the spectrum of professionals who prepare documents either for submission to regulatory authorities or for publication in peer-reviewed journals 5 .

Preprints
Preprints are preliminary scientific reports that are made publicly available online for anyone to read, comment on, and discuss before they have been peer reviewed. Some preprint servers scrutinize submissions for scope and for basic quality standards before making them publicly available [6][7][8] . Once the preprint is posted, most reputable preprint servers assign a unique digital object identifier (DOI) to aid traceability. Authors can revise preprints according to readers' comments and post iterative versions. Preprints are often not indexed on mainstream bibliographic services, although Europe PMC now indexes preprints 9 , and there are standalone tools for searching named preprint servers to improve discoverability 10 .
Preprints have been rapidly adopted by physicians and scientists, their obvious benefits being the immediate availability to their peers and the public, avoiding lengthy peerreview processes prior to release, and the option of readers to leave comments. However, there are issues associated with preprints that ideally should be addressed by standards jointly developed by a convened body of all stakeholders.

Issues with preprints
While preprints enable rapid release and discussion of data, many are never revised, and only about a third to a half are ever fully published 11,12 . "Once the toothpaste is out of the tube, it cannot (easily) be stuffed back in." 13 Provocative or poor quality research results could be reported by the media, or posted and discussed on social media, with little regard to the preliminary nature of the findings 14,15 . No amount of retrospective "tagging" will have much effect. Misinformation or deliberately misleading or sloppy science can be freely circulated, cited, and believed ad infinitum, regardless of whether it is ultimately debunked and retracted.

Our suggested solutions
Preprints should not be used as references in any medical publication unless these are cited in the manner of a personal communication, that is, as an in-text reference (using the preprint link, DOI, or both) rather than as bibliographic references. It should be clearly disclosed that the source is a preprint. Clearly distinguishing preprints from peer-reviewed articles might help to reduce the tendency of readers to view the work as fully vetted 14,15 . This should be done by Watermarking the article, as is done, for example, by medRxiv and bioRxiv, with the information that it has not been peer reviewed. Placing a clearly-worded disclosure in the body of the article highlighting that the findings have not been formally peer reviewed. Pre-publication vetting: Pre-publication checks by server hosts. MedRxiv performs a basic screening process for plagiarism, nonscientific content, and material that might pose a health risk, including material that might compromise existing public health measures 7 . However, these checks should be more extensive and consistent across server hosts, and a comprehensive checklist should be used (Appendix).
Encouraging authors to ensure that preprints that have been subsequently fully published be marked as such on the preprint server and linked via DOI to the fully published article.

Post-publication peer review
In post-publication peer review, an article is published in its original form, then subjected to informal (as with preprints) as well as invited peer review. For instance, with the model used by the F1000 publishing platform 16 , articles are posted online after passing pre-publication checks and after an article processing charge (APC) is paid. When posted, articles are assigned a DOI and opened to comment from registered users. Expert peer reviewers are invited to review in the usual way. All comments, peer review reports, and article revisions are available with the article, and once the article receives two favorable peer review reports, the final, peerreviewed version is indexed in external bibliographic databases and becomes fully discoverable. The benefits of this model are similar to those of preprintsrapid access to the readers and the option for readers to comment.

Issues with post-publication peer review
The issues with post-publication peer review are basically identical to those of preprints, but it should be noted that the requirement for an APC would potentially discourage casual or low-quality submissions. Articles are clearly marked as "under peer review," and the progress of that review is accessible to readers. As with preprints, articles undergoing post-publication peer review should not be used as references in any medical publication until the peer review process is completed and the article is approved for publication. If the article is cited, we suggest the citation be made in the same manner suggested for preprints. Issues associated with traditional peer review also apply and are addressed in Section 2.3.

Our suggested solutions
Our suggested solutions include those proposed for preprints; however, we suggest that the publication be indexed by mainstream bibliographic databases (if applicable) once it has been fully peer reviewed, as is done on the F1000 platform.

Traditional peer review
Traditional peer review occurs after a submitted article is accepted for consideration by a journal, then passed to expert peer reviewers. The reviewers' comments are sent to the authors to use in revising their article, or else the article is rejected after review. For rejected articles, authors can start the process again with another journal. If an article is revised to the peer reviewers' satisfaction, the article is published and assigned a DOI, after which the article is indexed in mainstream bibliographic databases. Peer review reports and revisions may or may not be available with the final article, depending on the peer review model the journal uses. The benefit of traditional peer review is that information is released to the readers only after there has been quality control applied by subject matter experts.

Issues with traditional peer review
Lengthy review process, which may impede the timely release of valuable informationparticularly in a pandemic or public health crisis. Inadequate time for high-quality peer review. Inconsistency among reviewers. Difficulty in "recruiting" qualified reviewers, given time commitment, particularly in times of health crises when the most appropriate reviewers are likely to have a high clinical workload.

Our suggested solutions
Authors: Submit rejection comments to second-choice journals, with itemized rebuttals and updates to the manuscript (portable peer-review) 17,18 . Be more accepting of editor referrals to cascade journals 19 . Journal Editors: Accept, request, or require portable peer review as described above, thereby reducing the need for additional review cycles. Consider commercial back-end services that expedite peer review (e.g. ResearchSquare [https://www. researchsquare.com/], as used by the BMC journals and others). Form a rapid response team of reviewers, with appropriate expertise, who can provide peer review with a quick turnaround time. Publishers: Standardize formatting requirements to expedite resubmission 20 . Offer fast-track options for potentially practice-changing work. Consider incentives for reviewers 21 .

Quality control
Make use of existing publication guidelines [22][23][24] and available checklists 25 to ensure high-quality publication development.
Include Clinical Trial Protocols and Statistical Analysis Plans (SAPs) as Appendix.

EDITORIAL
Ask all authors to sign an author form confirming that they had full access to the relevant data reported in their article, along with the acceptance of responsibility for submitting the article for publication. Furthermore, the contributor statement should name the authors (at least 2) who have accessed and verified the underlying data, as suggested in the revised Lancet publication guidelines 26 . Journals should clearly explain the initial quality review that editors perform on newly submitted manuscripts.

Training in peer review
Authors, peer reviewers, and editors should be adequately trained in the nature and technical aspects of peer review. Guidelines should be used, such as those created by the Committee on Publication Ethics (COPE) 25 , along with the reviewers' checklist in Appendix.
Medical journalists and the public should be educated on how preprints and pre-publications differ from peerreviewed literature.

The role of professional medical writers and scientific communicators in expediting the publication process
Evidence suggests that the use of professional medical writers enhances publication quality and speed [27][28][29][30][31][32][33] , and such assistance has been associated with a reduced risk for retractions due to misconduct 34 . If a qualified medical writer is part of the team, they should be involved in the process as early as possible 5 . The medical writer should have access to the clinical study report (if available), source data, and related documents, including statistical outputs and patient narratives, to the extent that dataprotection regulations allow. Professional medical writers should have an active role in ensuring the high quality of publications, including their development, editing, and referencing 22,24,35 , and the use of appropriate publication checklists 36 . Medical writers and statisticians should be actively involved in peer review, during which the medical writer will critically assess the quality of the manuscript according to common appraisal criteria, thereby augmenting the traditional subject-matter-expert review (Appendix). Medical writers could also be involved in pre-publication vetting, act as trainers, or both (see Section 3.2). As professional medical writers and communicators, we have identified areas that could benefit from increased quality assurance. We have suggested some processes that we believe would better ensure effective oversight of scientific and medical publications, whether in the context of a health emergency or not. To maintain confidence in published science, each involved party (including the reader) must take responsibility for exercising their best judgment and selecting information from sources with good publishing practices that are rigorous and transparent.