A year ago the world’s leading medical-journal editors announced plans to require their authors to share with other scientists the data associated with their published articles about clinical trials.
Those editors have now backed off, and instead are predicting an even longer wait before such a mandate actually comes to pass.
“I realistically think this will take several years” for the right environment to be in place, said Darren B. Taichman, secretary of the International Committee of Medical Journal Editors, which proposed the now-abandoned data-sharing requirement.
The benefits of an open-data system are widely accepted by scientists. Sharing the data that underlie a journal article helps colleagues confirm the accuracy of the published finding, speed and expand their own research, and credit the originators, advocates have said.
But the coalition of journal editors, also known as the ICMJE, said last week that a rash of complaints from scientists about the proposed requirement had led it to conclude that the research community still was not ready for the mandate.
The obstacles to a blanket data-sharing requirement for authors reporting the outcomes of clinical trials, the ICMJE said, include a lack of resources and rules for handling, storing, and crediting the data.
We don’t right now feel that we can responsibly make a requirement that can reasonably be met by everybody.
“We don’t right now feel that we can responsibly make a requirement that can reasonably be met by everybody,” said Dr. Taichman, an adjunct associate professor of medicine at the University of Pennsylvania and executive deputy editor of the Annals of Internal Medicine.
As with its original proposal, the reversal also is drawing criticism to the ICMJE. Opponents of the ICMJE policy don’t necessarily dispute that conditions beyond the control of the journals need improving. But they contend that the journals could be pushing the change rather than waiting for it.
“You could make the case that journals do have the least leverage,” compared with universities and government funding agencies, in pushing for open-data policies, said Heather Joseph, executive director of the Scholarly Publishing and Academic Resources Coalition, which represents academic libraries committed to open-access policies. “But they certainly do have some, as long as universities require publication in them for tenure.”
And journals may have an even greater obligation to support open-data policies than do institutions or funders, Ms. Joseph said, because they market their journal names “as proxies for quality.”
A ‘Cry of Anger and Angst’
Rather than mandate any sharing of data from published studies, the new policy of the 15-member ICMJE will require authors merely to describe a plan to share data. Journal editors “may” consider those plans when deciding whether to accept a particular manuscript, the ICMJE said.
That revised policy for ICMJE members, including The New England Journal of Medicine, The Journal of the American Medical Association, the Annals of Internal Medicine, and The BMJ, formerly the British Medical Journal, will take effect in July 2018. The policy also will require a data-sharing plan in the pretrial registration statement for any clinical trials that begin enrolling participants in 2019 or later.
We’ve been doing this already, and haven’t found any problem with it.
Although represented by the ICMJE, The BMJ in 2013 began a policy requiring authors of reports of trials of drugs and medical devices to make relevant, anonymized patient data available at the request of outside scientists. It expanded that policy in 2015 to cover all clinical trials.
“We’ve been doing this already, and haven’t found any problem with it,” said Fiona Godlee, editor in chief of the BMJ Group. Dr. Godlee argued within the ICMJE to stick with the proposal announced last year.
Although the ICMJE heard “a real cry of anger and angst from the research community” over its plan, none of the cited obstacles in terms of handling, storing, and crediting data seemed insurmountable if data sharing were made a firm condition of publication in an ICMJE journal, Dr. Godlee said.
Groups such as the Center for Open Science, formed by scientists at the University of Virginia, already are building complex computer systems that ease the storage and sharing of large volumes of data, and that enable the crediting of data creators. The center’s director and co-founder, Brian A. Nosek, sees no reason for the journals to wait for the universities or funders to move first.
Of course, every group can point to the other groups, saying that they are the ones that need to change.
“Of course, every group can point to the other groups, saying that they are the ones that need to change,” Mr. Nosek said. “But that underappreciates the challenge of culture change in a decentralized system. Each of them is necessary; none of them is sufficient.”
Dr. Godlee cited the ICMJE’s requirement in 2005 that all medical trials be described in advance, in a public registry, as a condition for the results to be published in an ICMJE journal. The ICMJE also was accused then of arrogance, given that no such registry systems existed at the time, Dr. Godlee said. “Arguably, though, that forced the systems to become real,” she said.
Only When Forced
By contrast, Ms. Joseph said, when the National Institutes of Health introduced an essentially voluntary system to encourage free access to articles derived from NIH support, it saw a compliance rate of less than 5 percent.
Some universities are pressing ahead on their own. Harvard University this year announced the creation of a Data Science Initiative, an institutionwide effort to share data in fields that include arts and sciences, business, education, engineering, government, law, medicine, and public health.
The university’s interest is being driven by the vast volume of available data, the powerful new systems for processing such information, and the growing realization of its value, said David C. Parkes, a professor of computer science at Harvard. “It’s really all coming together at the same time,” he said in announcing the plan.
But over all, said Roger A. Pielke Jr., a professor of environmental studies at the University of Colorado at Boulder, universities and their researchers probably will act to share data only when forced. “It’s actually contrary to existing incentives, where publishing as much as possible is rewarded,” Mr. Pielke said.
Federal action, unfortunately, is probably necessary, he said. The best model for change may be the rules governing scientific misconduct. In that arena, Mr. Pielke said, “federal standards tied to federal grant funding compelled universities to respond.”
Similar to the new ICMJE policy, the NIH asks its funded researchers for data-sharing plans without mandating any particular action. The NIH also has been funding initiatives to expand data-processing capabilities, and it’s considering further policy changes to push data sharing, an NIH official said. “We are in the process of working through details,” the official said, “so no additional information is available at this time.”
Paul Basken covers university research and its intersection with government policy. He can be found on Twitter @pbasken, or reached by email at paul.basken@chronicle.com.