The Hornet’s Nest of Data Sharing: Cardiologist-Led Proposals Do Not Necessarily Agree on Timelines, Reviews
The hornet’s nest of when and how clinical trial data could or should be shared is abuzz once again, with a series of conflicting proposals appearing today in the New England Journal of Medicine.
Some of the world’s leading clinical trialists in cardiovascular medicine have signed on to one proposal that would create a consortium of academic cardiologists that could “operationalize” the data-sharing mandates recently set out by the International Committee of Medical Journal Editors (ICMJE) and the Institute of Medicine (IOM). Two other proposals, however, including one endorsed by 282 investigators in 33 countries, offer differing views on timelines and on just who should be reviewing requests for data.
As previously reported by TCTMD, the ICMJE in January announced plans to require clinical trialists to share their data publically as a prerequisite for publishing trial results in their member journals. During the open public comment period that followed, a group calling itself the Academic Research Organization Consortium for Continuing Evaluation of Scientific Studies—Cardiovascular (ACCESS CV) signaled that it would be submitting a formal proposal.
“The ACCESS CV group was formed both to provide a consortium of likeminded people who want to [share data responsibly and] also respond to those efforts by the ICMJE and the IOM, by understanding the potential barriers then coming up with solutions,” Manesh Patel, MD (Duke Clinical Research Institute, Durham, NC), lead author on the ACCESS CV proposal, told TCTMD.
ACCESS CV Proposal
The international list of authors reads like a Who’s Who in academic cardiology, including principal investigators (PIs) for some of the biggest trials, past and present. A supplemental appendix lists several hundred additional names of investigators “who want to endorse and participate in” ACCESS CV.
Getting buy in from everyone wasn’t necessarily a cakewalk, Patel observed. However, “everyone agreed that we need to make a response to this ongoing movement about data sharing, and the response needed to be clear that we are all for data sharing, but that we as a group are going to use our abilities, our strength, our clout—anything we can do—to put together a way in which data can not only be shared better and safely without all of the risks.” There’s also the “aspirational goal of aggregating the data” he added, “so we may be able to ask questions that really matter but that may not be specific to one therapy versus another.”
The ACCESS CV proposal identifies key risks related to mishandling or miscomprehension of complicated data—leading to analyses that potentially alarm the public and result in publication bias; an unwieldy number of non-hypothesis-driven data analyses that could lead to type 1 errors; and the potential loss of patient privacy, which could ultimately result in fewer patients enrolling in studies.
While both the IOM and the ICMJE had proposed that individual patient data be shared within 6 months of publication, the ACCESS CV plan entails more time for trial investigators to conduct any secondary analyses they may have in mind. They propose a mechanism for accessing patient data to validate primary trial results within 12 months, but suggest a 24-month period, after which requests for secondary analyses would be considered. The consortium also proposes a web-based portal that would allow public access to data requests and review panel decisions in the hopes of reducing duplicate queries. Requests for access would have to include a publication plan, any conflicts of interest, ethics approval, links to any audits, and notification when a study is submitted for publication.
‘Fairness in Data Sharing’ Proposal
Of note, this timeline and some of the details differ strikingly from those set out by the other “consortium” proposal published in the NEJM, this one by the International Consortium of Investigators for Fairness in Trial Data Sharing. This was written by academic physicians from McMaster University in Hamilton, Canada, led by epidemiologist P.J. Devereaux, with cardiologist Salim Yusuf as senior author. Members of the McMaster group have authored some of the largest and longest running international trials in cardiology and beyond.
Yusuf and colleagues call the ICMJE’s 6 months “insufficient.” Instead, investigators for the original trial should be given exclusive use of “their” data for 2 years, they say, plus an additional 6 months for every year it took to complete the trial, up to a maximum of 5 years.
Because the public may want to validate initial trial data earlier, Yusuf and colleagues propose that the journals themselves undertake the confirmatory analyses prior to publication of the article, rather than have this be a principle of the open data plan.
One of the more controversial parts of both the ACCESS CV proposal and the McMaster group’s proposal is the role of the original investigators in reviewing requests for data use. ACCESS CV calls for the inclusion of the original trial’s principal investigator, a trial statistician, and a member of the data safety and monitoring board along with a panel of non-trial ACCESS CV members. The McMaster group calls for a review committee “with representation both of investigators involved in the trial and academic trialists who did not participate.” A third paper appearing in NEJM today, however—this one based on the Yale Open Data Access (YODA) project—specifically recommends excluding anyone involved in the trial from request reviews.
Asked whether having the PI on the review committee could bias subsequent research towards analyses that confirmed, rather than contradicted, the original trial findings or lead to academic rivals being barred access, Patel said the benefits outweigh the risks. “We believe that some of the people who conducted the original research have an important role in at least providing some stewardship for what’s feasible for these secondary questions,” he commented, adding, “I understand people will [worry about] bias, but we believe that there is actually strength here in having those interested primary investigators involved in saying yes, we want to share our data.”
C. Michael Gibson, MD, MD (Beth Israel Deaconess Medical Center, Boston, MA), senior author on the ACCESS CV proposal, agreed.
“It’s important to note that many large trials involve many hundreds of sites from around the world, and [they’re] often co-led by instigators from competing institutions such as Harvard, Duke, Stanford, Cleveland Clinic, and Europe,” he said. “Engaging ‘frenemies’ from so many institutions at the outset offers important checks and balances and minimizes any one group monopolizing the data or conclusions from the data.”
Moreover, Gibson stressed, “the PI is only one member of the panel and is kept in check by multiple ACCESS CV members who did not conduct the trial and the DSMB member who was not involved in the publication. The goal of the review process is to assure that the data is adequate to provide a valid answer to the question being posed.”
Roxana Mehran, MD (Mount Sinai Hospital, New York, NY), one of the signatories to ACCESS CV, pointed out that additional problems will arise if investigators are excluded. “If the PI is not involved, on the other hand, you could have a very important loss of what was the original question and hypothesis,” she said. “I do agree there is the potential for some bias if the PI is involved, but at the same time having the PI in there as one of the votes, but not a veto, is important for hearing how the trial was conducted. Watch a 15-minute presentation or read a paper that is 2,500 words, and people lose sight of what it takes to get there.”
These views are at odds with the system of data sharing used by the YODA project. Cardiologist and clinical trialist, Harlan Krumholz, MD, who with Joanne Waldstreicher, MD (both of Yale School of Medicine, New Haven, CT) details the YODA approach in the NEJM today, confirmed to TCTMD that their proposal involves detailed reviews of access requests, but not principal investigators.
“We are interested in giving credit to those who produced the data, see wisdom in people reaching out to the investigators, but also recognize that independent evaluations often need to be truly independent—of the funders and the scientists who may have strong feelings about the data,” Krumholz told TCTMD in an email. “The goal is to serve society’s needs, accelerate progress toward breakthroughs and practical new knowledge, and credit those who produced data, but move away from a system in which data were considered an asset of the investigators and data owners, to be used at their discretion (including whether it would be published at all) and more toward a public good (yet protecting participant privacy and proprietary interests).”
YODA currently manages access requests for trials sponsored by Johnson & Johnson—the only company currently participating in the Yale data-sharing program.
A final—and pressing—question is who should fund the work involved in reviewing requests for access and preparing the data for being shared more widely.
In an accompanying comment, US Senator Elizabeth Warren (D-WA) writes that trial sponsors should be on the hook for funding subsequent studies. “Expenses associated with administering data-sharing protocols,” she writes, “must be treated as a standard, necessary aspect of the costs of carrying out a clinical trial. And over the long run, data sharing may help reduce costs by allowing researchers to avoid duplicating trials or to answer questions without undertaking a separate data-collection effort.
The McMaster group takes a different view, making the bold suggestion that third parties who want to conduct secondary analyses should have to compensate the original investigators themselves. In YODA, the data holder—Johnson & Johnson or a subsidiary—pays for accepted requests, although Krumholz and Waldstreicher acknowledge that affordability is an issue that remains unresolved. The ACCESS CV authors note that the “significant matter” of funding and resources “has not been addressed” adding, “We plan to work with sponsors of clinical trials, the IOM, the ICMJE, governmental bodies, and regulatory authorities to develop the infrastructure for the data-sharing proposal we describe.”
Gregg Stone, MD (Columbia University Medical Center, New York, NY), one of the authors of ACCESS CV, told TCTMD that “funding is obviously a major issue and nobody has come up with a solution yet.” Three possible solutions include having future data requests be funded by the original clinical trial funding (industry or government), the requesting physicians, or governmental organizations. “If this is something that the government believes in strongly enough, there could be a new funding allocation along, for example, the lines of ClinicalTrials.gov,” he said. “But where all of these databases are housed, is it one or a hundred different possible sources? All of this needs to be worked out.”
Indeed, as the various NEJM proposals make clear, there is no consensus on how large-scale data sharing should move forward, and many questions remain unanswered. “We’ve seen multiple ideals published in this series of NEJM articles,” Stone said. “I believe the groups should get together, put their best ideas forwards, and together we come up with one integrated solution that takes the best of what everyone has thought of.”
“Data sharing holds incredible promise for strengthening the practice of medical research and the integrity of our clinical trial system,” Warren concludes. “I look forward to following these proposals as they continue to develop and urging their implementation.”
- Warren E. Strengthening research through data sharing. N Engl J Med. 2016; 375:401-403.
- Krumholz HM, Waldstreicher J. The Yale Open Data Access (YODA) project — A mechanism for data sharing. N Engl J Med. 2016; 375:403-405
- The International Consortium of Investigators for Fairness in Trial Data Sharing. Toward fairness in data sharing. N Engl J Med. 2016; 375:405-407.
- The Academic Research Organization Consortium for Continuing Evaluation of Scientific Studies. Sharing data from cardiovascular clinical trials—a proposal. N Engl J Med. 2016; 375:407-409.
- Disclosure forms provided by the authors are available with the full text of this article at NEJM.org.
- ICMJE Gets an Earful on Its Data-Sharing Proposal for Clinical Trials
- The Gift of Data: Medical Journals Back Public Data Disclosure as Prerequisite for Publishing
- On Record: Open Science and Data-Sharing