Login to Your Account

Global Standards Needed for Sharing of Trial Data


By Mari Serebrov
Washington Editor

The question is no longer should clinical trial data be shared but, rather, how can it best be shared.

Hoping to answer that question, an Institute of Medicine (IOM) committee heard from a range of experts Wednesday on strategies to responsibly share raw trial data.

Now that the European Medicines Agency (EMA), various consortia and even industry are setting up data sharing systems, there’s a growing demand for global standards for the data, patient privacy protection and research quality. The response to those demands could impact how broadly raw data is shared in the future.

One of the big fears is the spate of “rogue science” open access could release, especially if trial data are available to everyone. Experts should be the only ones analyzing the data, Robert Califf, of the Duke Translational Medicine Institute, told the committee. Otherwise, bad science will become prevalent.

Debunking false media reports that result from that information can sap up a lot of effort and resources from both industry and regulators. “The idea that there’s no value in expertise is a real mistake,” Califf said.

To ward off suspect science, some data sharing sites require requesters to present their credentials and an analysis protocol before they’re given access. That’s the tack the EMA started on when it began developing its draft consultation on trial data sharing. But then it had second thoughts stemming from an incident with a drugmaker that requested data about another company’s trial, said Hans-Georg Eichler, a senior medical officer with the agency.

After the second company refused the data, EMA got an identical data request, this time from a medical student. The agency realized that anyone who wants access to data could create false credentials or go through an acquaintance who’s a medical student or researcher. Thus, the agency decided it wouldn’t judge a requester’s qualifications or protocol.

“Are we worried about blatantly stupid analysis? No,” Eichler said. The stupid stuff will be caught by peer review, he explained. The bigger worry is more sophisticated analyses that look OK on the surface but have underlying faults that make the results questionable. If the EMA were to approve protocols in advance, it would appear to be endorsing such results.

Initially, the EMA also was going to require summary data only. “But when you eat, the appetite grows,” Eichler said, adding that the demand for data has snowballed. When the agency’s consultation takes effect in January, all trial data will have to be shared – unless a few drugmakers are successful in a legal challenge. (See BioWorld Today, March 13, 2013, and July 25, 2013.)

Europe is further ahead than the U.S. on requiring data sharing, largely because of political pressure. The EMA was told it wouldn’t get a budget if it didn’t require sharing of all trial data, Eichler said. The pressure stemmed from media reports of specific incidents in which drugmakers were accused of killing patients through misdeeds and secrecy.

Protecting Patient Privacy

Another concern is patient privacy. Patricia Teden, of Teden Consulting LLC, pointed out that a person could get the entire history of a single patient from a clinical summary report, unless the patient listings are redacted. The threat to confidentiality is more acute in trials involving very rare diseases.

While patient privacy is a valid concern, it “cannot be an excuse that we hide behind to avoid sharing data,” Deborah Collyar, of Patient Advocates in Research, said, adding that privacy protections are not about protecting patient data from research but protecting them from misuse.

Many trial participants are willing to have their data shared because they’re hoping for a cure. They recognize that “the more information researchers have, the faster they’ll figure [a cure] out,” she said.

Since there is a range in participant willingness to share data, committee members asked if it would be possible to give trial subjects a choice in the informed consent process about how their data can be used.

That’s not a good idea scientifically, Eichler said, as it would lead to a self-selection bias in the data.

Patient privacy also raises issues about what information can be shared from completed trials, as sponsors can’t assume that participants would want their data shared for unrestricted use, said Frank Rockhold, senior vice president of drug development sciences at Glaxosmithkline plc (GSK).

This year, GSK opened the raw patient data from all of its trials of approved or failed drugs to independent researchers, who must sign a confidentiality statement. The research requests are vetted for scientific merit by an independent panel of experts, and then access is granted via a secure website. (See BioWorld Today, Oct. 12, 2012.)

The panel takes a triage approach to the release of patient information, Rockhold said. For instance, if the research request involves a similar indication as the one in the trial, it’s more likely to be granted than one dealing with a different use. This won’t be an issue with new trials, as GSK has changed its informed consent to reflect use of the data for all medical research.

Industry Signing on

The GSK model is catching on with other drugmakers. Roche is now participating and two other drug companies are on the verge of announcing their participation, Rockhold said. Two more companies are considering joining.

“It’s a first step,” Rockhold said of GSK’s data sharing program. The goal is to move the data to an independent custodian.

The GSK program is one of several now in use. Some, like Project Data Sphere, focus on trials for a specific disease. Others are consortia involving academia, industry and nonprofit groups.

Because of the different models in use, members of the IOM committee expressed the need for common data standards so the data can be shared across the different systems in the future. A central repository or directory also would help in sharing the data that are available, they noted.

“Data is not knowledge,”Collyar reminded the committee, “so we have to turn them into knowledge.” That will happen by sharing the information. “We have to get this right,” she added.

But Eichler cautioned against overthinking how data are to be shared. As the U.S. confronts these issues, stakeholders should focus on what’s practical, rather than on what is “ideologically desirable,” he advised the IOM.