Contributed Commentary by Frank W. Rockhold

August 22, 2017 | Calls for greater transparency and open data access in clinical research—with its promise to benefit researchers, innovation, and patients—continue actively. Providing access to individual patient data from other investigators, researchers, and the public creates new value in multiple areas:

• Enables new discovery and answers research questions by using and combining existing data with increased statistical power
• Allows confirmation of existing research results by peer review and reanalysis
• Broadens research by enabling aggregation of data from disparate studies
• Helps in detailed evaluation of existing trials, avoiding unnecessary repetitive trials and additional participant risk
• Targets analysis of benefit/risk of therapies in subgroups for use in precision medicine. This relies on having access to individual patient-level data
• Strengthens trust in clinical research through enhanced openness and transparency
• Honors the commitment of individuals who volunteer for clinical trials.

In a recent development in June 2017, the International Committee of Medical Journal Editors (ICMJE) said that reports of clinical trial data submitted to ICMJE journals after July 1, 2018, must include a data sharing statement (DOI: 10.1056/NEJMe1705439). In addition, clinical trials that begin enrolling participants on or after January 1, 2019, must include a data-sharing plan in the trial’s registration. “These initial requirements do not yet mandate data sharing, but… editors may take into consideration data sharing statements when making editorial decisions,” notes the publication; there is an “ethical obligation to responsibly share data generated by interventional clinical trials because trial participants have put themselves at risk.”

A meeting titled, Aligning Incentives for Sharing Clinical Trial Data, was held in Boston and broadcast as a webinar by the New England Journal of Medicine on April 3-4, 2017. The meeting aimed to “examine the potential for data sharing to teach us things we did not know, to maximize the contribution of patients… and to make a positive impact on health outcomes.” Clinical trials require effort to accrue and organize data, and there is a need to ensure that all involved are adequately recognized and rewarded in a way that retains societal and academic incentives.

Reluctance To Share Patient-Level Data

In panel discussions at the NEJM meeting, some academic trialists appeared reluctant to share patient-level data, questioning its value and timing, and emphasizing the need to have exclusive access to data to incentivize research. Debate focused on the appropriate length for an “exclusivity period” to allow front line manuscripts to be completed, and on the value of granting access to society in increasing trust in clinical research, even if the data are not actually used.

Clinical trialists should be given incentives to share data, receiving appropriate acknowledgment and academic rewards when other researchers use “their” shared data to publish papers. Secondary investigators analyzing shared clinical trial data could provide a research question and data-analysis plan when requesting data access, submit their findings to a peer-reviewed journal, and share their own data analyses. Collaboration between clinical trialists and secondary users of data sets should be promoted and acknowledged. Data sharing is often perceived as a zero-sum game in which the original trialists lose if others perform secondary analyses. Trialists and secondary investigators should aim to be collaborators rather than antagonists.

Delegates heard that the expertise of bioinformaticists and data scientists is essential in overcoming complexity and enabling data sharing. NEJM editor-in-chief Jeffrey M. Drazen, MD, said that in the next year or two he would like to see “clinical trialists embracing data scientists” with a focus on helping patients improve their lives through the best science possible.

In future, clinical trialists, data scientists, and data analysts should strive to find a balance between maintaining an academic presence—including the publications that drive academic success—and appropriate collaboration to ensure that everyone involved is adequately recognized and rewarded, and future patients reap the benefits of the new information.

Increased access to individual patient-level data—essential for advances in precision medicine—will require several challenges to be overcome. These include the need to safeguard the privacy and confidentiality of research participants; ensure that the data are used for valid scientific purposes; providing data in ways that are understandable and navigable for external researchers; protecting intellectual property and maintaining commercial incentives; and creating incentives for researchers to share data.

Patients: Surprised At The Tension Between Trialists And Data Analysts

Patient panelists at the NEJM meeting expressed consensus that their trials data should be shared (with their privacy protected) and were surprised that there was debate over this issue. Patients seemed unaware of tension between trialists and data analysts, and of the existence of disincentives to data sharing. One speaker, Nancy Nagler, who has participated in multiple cancer trials, said she now feels even more strongly that data should be shared, the sooner the better, and in as much detail as possible. Nagler also advocated for a formal process to “close the loop” and share trial findings in an appropriate format with participants.

We owe it to patients who enroll in trials to make data available for further evaluation. The data access effort should be viewed as a journey that benefits patients in the first instance; other requirements from journals and legislators will then follow logically. Increased collaboration between trialists, data scientists, and secondary researchers will be an important step towards overcoming remaining challenges and some risks of data access, all of which are manageable and can be overcome with focus and effort.

Frank W. Rockhold is Professor of Biostatistics and Bioinformatics at Duke University and a member of the Duke Clinical Research Institute. He can be reached at frank.rockhold@duke.edu.