Recently, PRIM&R submitted comments to the International Committee of Medical Journal Editors (ICMJE) in response to their proposed requirements for sharing clinical trial data. PRIM&R stated that we fully support initiatives to promote data sharing on the grounds it enhances the value of clinical research and the participation of human research subjects.
However, as detailed below, we note that several practical and policy issues need to be addressed by ICMJE and its data sharing policy, including what it means for IRBs and the informed consent process.
ICMJE published their proposal in an editorial that ran in its member publications in January 2016. ICMJE’s members include Annals of Internal Medicine, Journal of the American Medical Association, New England Journal of Medicine, and PLOS Medicine. Although ICMJE only has eleven member journals, ICMJE’s polices are followed by many non-ICMJE journals. A change in ICMJE’s policy will therefore likely have a significant impact on the research community, as publication is a driving force for many investigators.
ICMJE asked the public to respond to four questions about their proposed requirements, as well as provide additional comments.
ICMJE first asked whether the public agrees with the proposal to require as a condition of consideration for publication of a clinical trial report in their member journals that authors share with others the deidentified individual patient data underlying the results presented in the article.
PRIM&R responded that we fully support initiatives to promote data sharing. As a culture, we ask individuals to assume risks associated with being research subjects for the benefit of science. Sharing patient-level data can optimize the use of existing data sets, enhance accountability, and reduce research redundancy. But the value of data sharing is not without risk. PRIM&R cautions that concerns about subject privacy, cost, and the impact on academic and commercial incentives will need to be addressed.
Second, ICMJE asked whether the public agreed with their proposal to require that deidentified individual patient data be shared within six months following publication.
PRIM&R responded that for some studies, the six month timeframe will not be problematic, but for others, a longer timeframe may be more appropriate (for example, for complex studies that involve multiple sub-analyses).
PRIM&R expressed our concern that requiring primary investigators to share data within six months could motivate them to delay publication of research out of concern that releasing their data before additional planned analyses are complete will be detrimental to their ability to publish their research. To address this possibility, PRIM&R suggests that ICMJE allow investigators to propose at the time of registration a reasonable alternative (but definitive) timeline within which they will share their patient-level data.
ICMJE’s third question asked whether the public agrees with a proposed requirement that authors include a plan for data sharing as a component of clinical trial registration.
PRIM&R supported this proposed requirement. Increased transparency fosters public trust in the research enterprise. However, we recommended that ICMJE provide further details on what constitutes an acceptable data sharing plan.
In addition, we note that ICMJE’s proposal does not make adequate provision for changes to data sharing plans. Researchers generating data may make changes to their protocols after registering their trial and providing a plan for data sharing. Disputes may arise between original data generators and those who later request data access about whether the new data sharing plan is adequate. PRIM&R recommends that ICMJE consider developing an independent committee to resolve disputes in this area, and that the committee include a human subjects research representative.
ICMJE’s final question asked for suggestions on how to provide credit to researchers who generate and then share their clinical trial data.
PRIM&R strongly agreed with ICMJE that those who generate and then share clinical trial data sets deserve credit for their efforts and work product. Without such credit, data sharing could be detrimental to researchers’ careers and, by extension, to the advancement of science. PRIM&R recommends that when collaboration between data generators and secondary researchers is not a viable option, journal editors ensure that the source of data is always clear in publications, thus ensuring that data-generators are given appropriate credit in any secondary analyses and that their institutions have a basis on which to give them appropriate academic or professional recognition, for instance, during merit-review or tenure-decision processes.
ICMJE then asked the public for any additional comments on their proposal.
PRIM&R notes that ICMJE’s proposal means guidance will need to be developed for IRB review of data sharing plans with respect to the adequacy of measures taken for protecting subject privacy and confidentiality. IRBs will be key partners in ensuring that informed consent processes include appropriate and understandable information about data sharing and conditions under which information will be shared. PRIM&R requested ICMJE encourage researchers to inform their IRBs at the start of research about their plan for data sharing, so that the IRBs can advise the researchers about what human subjects need to know about data sharing.
PRIM&R also urges ICMJE to call for and participate in the development of ethical standards for the conduct of secondary analyses on shared data. Finally, PRIM&R encouraged ICMJE’s members to engage the scientific community in a dialogue about how best to pay for new data sharing requirements.
We encourage and welcome you to share your thoughts on our comments and ICMJE’s proposal by leaving a comment below.
On a related note, in last month’s From the Director blog, we discussed data sharing and ClinicalTrials.gov.