Clinical research involving human participants presents challenges for any journal with policies on data sharing due to the frequent presence of personally identifiable information in clinical datasets. These challenges are particularly acute for Scientific Data, given our mandate that data associated with our articles must be shared as a condition of submission.
Scientific Data established, in 2014, a working group with stakeholders from a diverse set of organisations to investigate how the transparency of clinical research data can be increased while still protecting the privacy of individuals (http://go.nature.com/duzM7O). A draft set of guidelines and recommendations for how this might be achieved was subsequently developed and has now being shared for public comment1.
After publication, journals have traditionally required or encouraged clinical datasets to be available from the original investigators upon reasonable request. However, these systems have been shown to be ineffective2–5 in securing access to data for legitimate reanalysis. Coupled with this, there is much evidence of bias in the medical literature favouring positive results6,7, which can ultimately lead to ill-informed treatment decisions by clinicians.
Recently, more robust systems for accessing clinical research data on request have emerged, such as the Yale Open Data Access (YODA, http://yoda.yale.edu/) Project and Clinical Study Data Request (CSDR, https://www.clinicalstudydatarequest.com/). These initiatives ensure patient privacy and scientific legitimacy is protected by checking anonymity of datasets and only providing access to qualified researchers in a controlled access environment. Protection of participants is codified in Data Use Agreements (DUAs) and independent committees review requests to access data.
What may be missing from these systems, however, is the permanence, discoverability, accessibility and quality assurance through peer review, provided by journals—as well as robust, persistent links between the peer-reviewed literature and data repositories.
Scientific Data is already glad to consider datasets derived from clinical studies, and has a flexible policy acknowledging that such datasets cannot always be openly shared on the web (http://go.nature.com/rlZ4ny). This policy, however, leaves open substantial questions regarding where authors should deposit clinical datasets, and how peer review of such datasets should be conducted. Today, Scientific Data published its first article with a restricted-access data component8. This highlights the potential value of Data Descriptor articles for such data even if there is not yet a broad consensus on how journals should link articles to restricted data.
Scientific Data sees a role for further collaboration between journals, repositories, and these new data-on-request services. The draft guidelines, entitled ‘Publishing descriptions of non-public clinical datasets: guidance for researchers, repositories, editors and funding organisations’, were developed with input and feedback from a working group including representatives from several pharmaceutical companies, research funders and the YODA and CSDR initiatives.
The guidelines are potentially applicable to all journals that consider clinical research studies, not just data journals. Aspects of these guidelines are fairly non-controversial even if they do not yet reflect widespread practice. For example, recommendations are made for additional article sections that encourage authors to transparently declare how data can be accessed, and for the creation of persistent public landing pages at data repositories even for non-public datasets9.
Elements of these recommendations, however, would represent much more significant changes in current clinical data sharing and publishing behaviour. The guidelines, for example, express a clear opposition to unnecessary restrictions embedded in DUAs that are not designed to protect participant privacy, particularly co-authorship mandates. It is also recommended that data repositories should routinely allow journal peer-reviewers access to clinical datasets, in accord with DUAs, prior to publication.
These guidelines do not directly address the issue of motivating clinical researchers to want to share their data—although a survey found they are interested in sharing data but may not be certain how to10,11. Scientific Data hopes that additional clarity about how to share, publish and peer-review clinical data will help lower barriers to sharing, and help researchers receive more credit for their work.
Scientific Data encourages comments from the community by September 7th, 2015. The authors will submit the guidelines for formal publication in due course. Importantly, we also invite researchers interested in sharing their clinical datasets to prepare and submit Data Descriptors for consideration by Scientific Data.
References
Hrynaszkiewicz, I., Khodiyar, V., Hufton, A. & Sansone, S. A. Publishing descriptions of non-public clinical datasets: guidance for researchers, repositories, editors and funding organisations. BioRxiv http://dx.doi.org/10.1101/021667 (2015).
Savage, C. J. & Vickers, A. J. Empirical study of data sharing by authors publishing in PLoS journals. PLoS One 4, e7078 (2009).
Vines, T. H. et al. Mandated data archiving greatly improves access to research data. FASEB J 27, fj.12–218164 (2013).
Jaspers, G. J. & Degraeuwe, P. L. J. A failed attempt to conduct an individual patient data meta-analysis. Syst. Rev 3, 97 (2014).
Wicherts, J. M., Borsboom, D., Kats, J. & Molenaar, D. The poor availability of psychological research data for reanalysis. Am. Psychol. 61, 726–728 (2006).
McGauran, N. et al. Reporting bias in medical research - a narrative review. Trials 11, 37 (2010).
Jones, C. W. et al. Non-publication of large randomized clinical trials: cross sectional analysis. BMJ 347, f6104–f6104 (2013).
Holmes, A. J. et al. Brain Genomics Superstruct Project initial data release with structural, functional, and behavioral measures. Sci. Data 2, 150031 (2015).
Starr, J. et al. Achieving human and machine accessibility of cited data in scholarly publications. PeerJ Comput. Sci 1, e1 (2015).
Rathi, V. et al. Data from: Sharing of clinical trial data among trialists: a cross sectional survey. Dryad Digital Repository http://dx.doi.org/10.5061/dryad.6544v (2012).
Rathi, V. et al. Sharing of clinical trial data among trialists: a cross sectional survey. BMJ 345, e7570 (2012).
Rights and permissions
This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0
About this article
Cite this article
Let’s be pragmatic about clinical data. Sci Data 2, 150034 (2015). https://doi.org/10.1038/sdata.2015.34
Published:
DOI: https://doi.org/10.1038/sdata.2015.34
This article is cited by
-
Five years of Scientific Data
Scientific Data (2019)
-
Finding a sensible approach to sensitive data
Scientific Data (2018)
-
Publishing descriptions of non-public clinical datasets: proposed guidance for researchers, repositories, editors and funding organisations
Research Integrity and Peer Review (2016)
-
Let referees see the data
Scientific Data (2016)
-
How should individual participant data (IPD) from publicly funded clinical trials be shared?
BMC Medicine (2015)