Medical researchers are talking about starting an online archive where they could post unreviewed results from clinical trials, such as one that tested this light therapy for cancer patients.

NASA/MSFC/David Higginbotham

Plan for new medical preprint server receives a mixed response

CHICAGO, ILLINOIS—Medical scientists may get their own place to post unpublished studies. Researchers at Yale University and Yale School of Medicine are preparing to launch a preprint server, called MedArXiv, which would specialize in publishing the results of clinical research.

But the plan, presented here this morning at the Eighth International Congress on Peer Review and Scientific Publication by Yale cardiologist Harlan Krumholz, has received a mixed reception. Many in the medical community aren’t sure that posting preprints on the web is such a great idea; the general fear is that such papers might sway clinical practice, or prompt patients to try treatments on their own, before reviewers can vet the findings. “It would be helpful to be sure we don’t do harm,” Howard Bauchner, editor-in-chief of The Journal of the American Medical Association (JAMA), said after Krumholz had finished his talk.

Physicists have posted their preprints on arXiv for more than 2 decades, and the number of papers on bioRxiv, a repository for the life sciences launched in 2013, is growing exponentially. Krumholz argued that it’s time for the medical community to hop on the bandwagon; it would speed up research, he said. And data from clinical studies already get out before they’re published, for instance in press releases and at medical meetings, Krumholz said; why not make the full results public?

Although bioRxiv accepts certain clinical papers, Krumholz says MedArXiv would give medical researchers their own repository. “I’m not sure the clinical research community will feel bioRxiv is their home,” he says. “On the applied clinical side, many have never even heard about it.”

But MedArXiv will have a hard time attracting preprints if mainstream medical journal editors decide they won’t publish final versions of the papers. Currently, The BMJ and The Lancet are among the few medical journals that have explicitly said that posting a preprint doesn’t preclude publication; Nature and Science, which both occasionally publish medical studies, have the same policy. But at the JAMA Network, which publishes a dozen journals, the issue is hotly debated. “There are very strong opinions on both sides,” Bauchner says. “I suspect we’ll consider papers posted on preprint servers, but we’ll discourage it.”

Bauchner said he was “surprised” that Krumholz didn’t bring up the potential risks to patient health of publishing unreviewed papers in his talk.

Drummond Rennie, editor of JAMAevidence and former editor of JAMA, and the organizer of the meeting, says he's “not nearly as worried.”  “Whilst it’s true that people might give twice the dose as a result of your preprint, that’s a bit far-fetched,” he says. “People in general don’t go straight from a paper to the prescription pad.”  And papers that have passed peer review have all kinds of flaws as well, he adds.

For his part, Krumholz—who acknowledged the irony of promoting a repository for unreviewed research at a meeting about improving peer review—says there are ways to mitigate the risks. For instance, papers could be given a clear “watermark” indicating they haven’t been reviewed yet, and are primarily “science for scientists.” As to the debate occurring within the JAMA journals, Krumholz says, “They have a conservative editorial board that wants to learn more about this and see how it plays out. … I think that’s fine. It’s our job to show that [preprints are] not as scary as they think.”

Epidemiologist Steven Goodman, who is associate dean of clinical and translational research at Stanford University in Palo Alto, California, said that it’s probably “inevitable” that the preprint revolution will come to medical science, but that it should be done “in a very careful way … just like we’d do with a new therapeutic.” One way would be to start small, in just a single field of medicine, he said. But that doesn’t appeal to Krumholz. It would delay the introduction, and “I would not want to restrict who can use it,” he says.

Krumholz says the original plan was to open the server for business today, as part of the Yale University Open Data Access Project. But the launch was delayed; Krumholz says discussions with journal editors are ongoing, to avoid “surprises.” He and his colleagues are also talking to bioRxiv about possible collaboration, he says. “Our goal,” he says, “is not to be full owners and dictators of this process.”

John Inglis of Cold Spring Harbor Laboratory (CSHL) in New York, co-founder of bioRxiv, confirms that CSHL is in discussions with Krumholz’s team on developing MedArXiv. The server may draw on the same infrastructure as bioRxiv but “look different and have additional functionality,” Inglis wrote in an email. Inglis also expects that the collaborators will “adapt and extend” the process bioRxiv now uses to screen clinical trial and epidemiology submissions: “Screening is of particular importance for medically relevant information,” he says.

With reporting by Jocelyn Kaiser.