Guiding Principles for the Development of Policies on Sharing Clinical Trials Data (January 2014)

The ICMJE is considering policy related to the sharing of clinical trial data as we believe it has the potential to maximize the contributions of trial participants for the benefit of society. We welcome thoughtful efforts at evaluating the potential benefits and risks of different approaches to data sharing and their implementation. ICMJE has provided input to the Institute of Medicine’s Committee on Strategies for Responsible Sharing of Clinical Trial Data ( http://www.iom.edu/Reports/2014/Discussion-Framework-for-Clinical-Trial-Data-Sharing.aspx ) in which we outline the principles upon which we believe plans to enable clinical trial data sharing must be developed. Here are the comments ICMJE provided to the Institute of Medicine’s Committee:

Submitted to the IOM on January 29, 2014:

The International Committee of Medical Journal Editors (ICMJE) believes that the sharing of clinical trial data is an integral part of the scientific endeavor by enabling verification of published trials. The ICMJE is in the process of considering policy on this issue. The ICMJE believes that the following principles should guide the sharing of clinical trials data:

  • Authors, their institutions, and funders have an obligation to ensure that data supporting the submission of a clinical trial for publication is in a form that can be understood and reanalyzed by others.
  • Authors must maintain their data in a sharable state and they should commit to sharing data upon reasonable request. Further work is necessary to plan and establish the secure, reliable and sustainable mechanism(s) by which sharable data will be made available upon reasonable request. Further work is necessary to also define what constitutes a reasonable request according to factors such as the purpose of the requester, the timing (e.g., a defined period following article publication during which authors need not share data) and others. Who is to evaluate whether requests are reasonable also requires clarification.
  • Shareable data should include all data that underpin the published results, and data collected on all adverse events (serious and other, whether anticipated or not) up to the time of the request.
  • Sharable data must be in a format that is readable and sufficient to allow reproduction of the original analysis. It comprises: de-identified individual patient data, a data dictionary that specifies the definition of each variable and how and when it was measured, and the statistical plan and code used to analyze the data. Further work is necessary to begin to define how the quality of shared data is to be maintained (e.g., whether data should remain with the primary investigators, a third party, each requestor, and / or others).
  • If journals become aware that data sharing obligations are not being met, journals may choose to investigate, to publish an expression of concern, or in certain cases to retract the publication. Additional stakeholders (e.g., granting institutions) should be encouraged to consider policies aimed at ensuring that data sharing obligation are being met.
  • Authors and IRBs should ensure that the language of patient informed consent documents enables that data are de-identified and can be shared.
  • Those who receive shared data must commit to making the results of their analyses public, make clear if and how their analytic approach differs from the original, and give credit to the source of the original data.
  • Further work is necessary to define incentives for and protection of primary investigators who share their data.