ACCESS CV on Trial Data-Sharing: Beware Unintended Consequences

Good idea, but beware unintended consequences. That’s the bottom line of a new proposal from a consortium of academic cardiovascular researchers in response to calls from major medical organizations for greater transparency of clinical trial data.

The proposal, published in the Aug. 4 issue of The New England Journal of Medicine (NEJM), dubs itself “a strategy to thoughtfully operationalize the recommendations of the International Committee of Medical Journal Editors (ICMJE) and the Institute of Medicine (IOM) for sharing clinical trial data.” The ICMJE put out its recommendations in January 2016, while the IOM recommendations came a year earlier.

The consortium of researchers — known as ACCESS CV (Academic Research Organization Consortium for Continuing Evaluation of Scientific Studies – Cardiovascular) — lauds the spirit of the ICMJE and IOM recommendations but warns that the devil is in the details.

“Full disclosure of clinical trial results has long been a cornerstone of the academic medical community’s participation in clinical trials,” says ACCESS CV member A. Michael Lincoff, MD, a co-author of the proposal in NEJM.

“The proposals by the IOM and ICMJE to provide access to primary data from clinical trials may have value for some academic research efforts, but they also carry a number of potential dangers,” adds Dr. Lincoff, Director of the Cleveland Clinic Coordinating Center for Clinical Research (C5Research) and a practicing cardiologist.

The primary risks

Part of the ACCESS CV paper in NEJM outlines those dangers, which include:

• Generation of analyses by insufficiently informed researchers. The proposals by the IOM and ICMJE include the availability of patient-level data from trials via a publicly accessible website. “This would allow people who have had nothing to do with the planning or enrollment of a trial to perform analyses of the trial data,” says Dr. Lincoff. “They might not fully understand the data sets, which can be extensive and complex, and this could lead to findings of discrepancies where none exist.”
• Agenda-driven biases and spurious findings. Similarly, unfettered access to a trial’s primary database would leave the data subject to analyses by people with agendas and secondary motives — as well as to spurious subanalyses. “If you do enough subgroup analyses of a dataset, you’ll end up with some sort of positive finding purely by statistical chance,” Dr. Lincoff explains. “When investigators do a trial, we must prespecify a limited number of subgroups we’ll analyze. The lack of such requirements with a publicly accessible database would lead to selectively published subanalyses on various groups’ pet subjects and/or to spurious findings that actually mean nothing.” Such findings, he adds, could cause needless alarm and compromise public health.
• Violation of patient privacy. Although publicly available databases would not contain patient names, research suggests that the identities of up to 85 percent of patients can be reconstructed based on date of birth and other demographic details typically contained in trial databases, Dr. Lincoff notes.

“For reasons like these, hundreds of investigators throughout the world have come together to form ACCESS CV to propose a framework that would allow sharing of data from clinical trials, but in a responsible, accountable and scientifically valid manner,” Dr. Lincoff says.

An alternate approach

Much of the NEJM paper lays out the principles of an alternate framework proposed by ACCESS CV for sharing data from cardiovascular trials. Among the key recommendations:

• After publication of a trial’s primary results, researchers not involved with the trial could send requests to conduct secondary analyses to the study’s publications committee. Two years after initial publication, requests would go to a review group made up of selected ACCESS CV members along with the trial’s principal investigator, a trial statistician and others. Requests would be evaluated based on feasibility, the researchers’ technical capabilities, nonduplicativeness and other factors.
• When necessary, ACCESS CV would offer a mechanism for validating the primary results using the underlying individual patient data within 12 months after publication.
• ACCESS CV would develop a web-based portal for public dissemination of all data requests and the subsequent decisions. The portal would also serve to systematically ensure disclosure of conflicts of interest and publication plans and confirm adherence to data-use agreements.

Next steps

The proposal’s authors identify “how to provide meaningful academic credit to the team that designed and conducted the trial” as an unresolved issue. Noting that funding will be needed to realize their proposal, the authors state that they plan to work with trial sponsors, the IOM, the ICMJE, and governmental and regulatory authorities to develop the needed data-sharing infrastructure.

The proposal notes that it may well apply to trials in clinical areas beyond cardiovascular disease, and Cleveland Clinic oncologist Mikkael Sekeres, MD, MS, was among a handful of noncardiovascular specialists on the ACCESS CV writing team.

“Availability and openness of trial results is fundamental to what we do across all disciplines of academic medicine,” notes Dr. Lincoff. “This additional step of complete transparency of all primary data is new. It would be beneficial in allowing unprecedented aggregation of trial results for greater statistical power to answer research questions. But it needs to be done in a way that minimizes potential risks to patients and scientific validity.”