Why medical journals must make researchers share data from clinical trials

This month a new BMJ policy on sharing data from clinical trials takes effect. From July 1 2015, the authors of all clinical trials published by the journal must agree to make individual patient data from the trial available to other researchers upon reasonable request. Among major medical journals, only PLOS also requires data sharing as a condition of publication.

There are many advocates for data sharing, including the National Institutes of Health, The Cochrane Collaboration and the World Health Organisation. A 2011 statement from the Cochrane Collaboration, for example, called for sharing:

All data from all randomised clinical trials, including raw anonymised individual participant data that do not allow identification of individual participants, and the corresponding trial protocols to become publicly available free of charge and in easily accessible electronic formats.

A landmark report released this April by the Institute of Medicine concluded that data sharing is in the public interest and should become “the norm”.

But sharing is as yet not standard in the medical research community. Instead, the current research enterprise is probably better characterised as a “walled garden” in which most industry and academic investigators carefully control access to the information collected in clinical trials.

This hoarding of information has many bad consequences. It prevents other researchers from checking the accuracy of any results the original researchers might have published. It also makes it impossible to tell if those researchers have manipulated data or selectively published information about the study, which are problems that continue to occur. Most importantly, data hoarding prevents other scientists or patients from using clinical trial information to inform or accelerate their own research.

Hoarding or selectively publishing data.Data by Shutterstock

The the problems caused by making research data inaccessible are not mere abstractions: they affect real people. One widely mentioned example is the case of the Italian scientist Alessandro Liberati, who was unable to obtain information from clinical trials that were relevant to decisions he had to make about how to treat his own cancer. Among the evidence pertinent to his decision were “four randomised controlled trials, whose results haven’t yet been fully published …”

In an opinion piece describing his plight he asked “why was I forced to make my decision knowing that information was somewhere but not available?” Liberati concluded by saying that “research results should be easily accessible to people who need to make decisions about their own health”.

Clinical trial data should be thought of as a public good. This has been defined as:

a good that does not diminish in value with use, and has (almost) no marginal cost for users after the first use. A scientific public good is special because its value increases with use and it is important to maximise us of the information obtained from the sacrifices of human participants.

Sharing data will accelerate medical progress because discovery is a cumulative process that builds on previous work. If that previous work is not available then scientific progress suffers. Harlan Krumholz, Yale researcher and open data advocate, has written that “many trials yield only a fraction of the knowledge that could be produced with more resources and creativity.”

There is already a degree of collaboration among scientists and research groups around the world, but imagine how much more productive it might be if rivalries were dissolved by freeing up information. For one thing, sharing of clinical trial data would make possible creative research outside the current set of stakeholders and scientific insiders.

An additional benefit of wider sharing of data from clinical trials is that it might help restore trust in the clinical research enterprise. At present, according to the Center for Information and Study on Clinical Research Participation: “Public sentiment toward the clinical research enterprise is at an all time low … more than 70% of Americans believe that drug companies put profits ahead of patient needs.”

It is important to counter cynicism about medical research. If patients and doctors lack confidence in medical evidence, what hope is there for evidence-based medicine? If the evidence on which guidelines and recommendations are based is not trusted, how likely is it that doctors will apply therapies or act in other ways that are consistent with the best evidence?

Despite all of the benefits we can expect as a result of increased sharing of research data, there will of course be problems. It is inevitable that serious unanticipated consequences will occur. But even knowing that, it is still the right thing to do. That is because there will also be many unanticipated serendipitous and important benefits as well. The balance of benefits and harms is almost certain to be favourable. We should have confidence that human and scientific ingenuity will be able to solve the problems that emerge.

We are engaged in one of the great struggles of human knowledge – the struggle to liberate clinical trial data and ensure it is put to its highest and best use, now and in the future. Medical journals are in a position to help make data sharing the norm by requiring it as a condition of publication. The BMJ calls upon other journals to help make this happen.

Elizabeth Loder is employed as a research editor by The BMJ, which is an advocate of data sharing and was a founding member of the AllTrials Initiative. Dr. Loder has an academic interest in the topic of data sharing and research integrity. She has lectured on the topic and has participated in a number of conferences and workshops aimed at expanding data sharing requirements for medical research.