A new preprint server for the medical and health sciences – medRxiv – has launched, along with safeguards designed to mitigate the risk of non–peer-reviewed findings prematurely guiding clinical practice or misinforming the public.
Dr. Theodora Bloom
The new repository of preprints is intended for researcher-to-researcher communication – and mainly to facilitate faster sharing of research findings before publication in peer-reviewed journals. Papers will not be scrutinized for study design or the strength of the science, but they will be screened by an external clinical scientist and – at least for now – by an editor funded by BMJ, the London-based publisher and one of the three cofounding organizations of medRxiv (pronounced “med archive”).
The server’s six-person leadership team – comprising leaders from BMJ and cofounders, Cold Spring Harbor Laboratory in New York, and Yale University in New Haven, Conn. – will make final decisions about whether to post papers that generate concerns.
“We’ve put in place more stringent screening procedures than existed for bioRxiv, [a biological preprint server launched in 2013 by Cold Spring Harbor Laboratory],”said Theodora Bloom, PhD, executive editor of BMJ. “We’ll specifically ask the question, is there a risk to public health or health-related behaviors if this preprint is posted and [turns out to be] wrong?”
Concerns that poor information will be disseminated to the public or that the public will misinterpret information published, were heard by the medRxiv founders as they “work-shopped the idea and talked with the community,” said Joseph Ross, MD, an associate professor of medicine and public health at Yale and codirector of the Yale Open Data Access (YODA) Project
Dr. Joseph Ross
“We’re taking a cautious approach, particularly in the early days as we learn from the process,” he said. “How a paper [could potentially influence clinical practice] will be a guiding question.”
The cofounders had several conversations, Dr. Ross said, with Howard Bauchner, MD, editor in chief of JAMA, who took a strong stance against preprints and shortcutting the peer review process in a 2017 editorial titled “The rush to publication: An editorial and scientific mistake.” (Dr. Ross is an associate editor at JAMA Internal Medicine. Dr. Bauchner was unavailable for comment on the safeguards built into medRvix.)
Aaron D. Viny, MD, a hematologist-oncologist at Memorial Sloan Kettering in New York, said he has mixed feelings about preprints and believes the stakes are higher with medRvix, given that it will house clinical content – including, he anticipates, single-institution, nonprospective outcome studies of off-label drug uses. “These aren’t bona fide clinical trials and may not have the best data,” he said.
Still, there are advantages for investigators – and for the progress of research – with earlier dissemination of findings, Dr. Viny said. He recently had a paper posted on bioRvix for the first time. The paper was undergoing revision for a peer-reviewed journal and was being presented at a national meeting at the time it was posted.
“We timed it [as such], so that not only were we presenting it at a national meeting, but it also got more Twitter buzz,” he said. “I thought it was a good body of work, and I was excited to discuss it online with the scientific community.”
Dr. Aaron D. Viny
Dr. Viny’s decision is common among preprint authors and reflects the values of the preprint server, Dr. Ross said. “When people are reading or hearing about [new findings] at a meeting, they can go to the papers to get more complete information.” And, he said, the investigators themselves can get more feedback than they otherwise would.
In addition to papers that are well on their way to publication in peer-reviewed journals, Dr. Ross anticipates that medRvix will house papers on qualitative studies and observational research that face more arduous publication paths. He said he expects to see research on medical education and hopes to see papers on “quality improvement work, which typically involve small interventions at a single institution, and have important insights but are hard to publish because of generalizability and controls.”
And while there has been a “positive shift” in the past 10 years in the publication of negative results in peer-reviewed literature, medRvix may well capture studies that have negative results “because they have challenges with recruitment or other [elements of study design],” Dr. Ross said. “There is still a lot that can be learned by the scientific community from these negative studies, but they’re very difficult to publish in a peer-reviewed journal.”