Sharing all types of clinical data and harmonizing journal standards
Current diversity in journal policies does not optimally serve the cause of data sharing as it allows varying academic standards. Through this system, study authors are able to choose their standard of preference when submitting a study report – a decision which might be guided, among other considerations, by their willingness to share.
As a first step towards better harmonization, biomedical journals should require a data sharing statement for all types of clinical study reports, and not simply for randomized studies. If audit and accountability are the ‘bread and butter’ of good medicine and science [6], accepting various policies for different clinical study types would imply that studies with an observational design are not good science.
Even if publishing a data sharing statement does not mean making all data available, such a policy would swiftly implement a cultural change in the definition of scientific outputs. Currently, a scientific output only corresponds to a study report published in a medical journal, while in the near future it might consist of all materials described in the manuscript, including all relevant raw data.
With such a policy uniformly implemented, researches, who are currently interested in designing and conducting studies with the aim of meeting the highest methodological standards and requirements to ensure publication in major medical journals, would consider the issue of data sharing from the inception of their research projects. This would imply, for example, (1) the inclusion of a data sharing plan as part of a study protocol and its registration in international repositories of study protocols [7]; (2) agreement with local ethics committees on a procedure to preserve patient confidentiality and privacy when de-identified individual patient data are shared [8, 9]; (3) the inclusion of financial support for data sharing in grant applications; (4) drafting of a detailed publication plan in order to allow the best use of the database [10]; and, even more importantly, (5) the development of a high-quality database in a way suitable for secondary uses, written and coded in English, for example, but also meeting other requirements that expert methodologists would need to further develop and define [11]. There should also be careful development of web-based infrastructures for open data, as it would be rather disappointing if the promising development of open sharing of data led to no more than researchers piling their data in fairly unsearchable data repositories [12, 13]. Additionally, reporting guidelines, such as the CONSORT for clinical trials, PRISMA for systematic reviews of clinical trials, STROBE for observational studies, and MOOSE for systematic reviews of observational studies, would need to be updated by adding items for proper data sharing plans (what to share, when, and how) [7].
This policy would make data sharing the norm, with some reasonable exceptions that authors may publicly declare in their data sharing statement [14]. As the majority of published studies are not clinical trials, but rather studies with an observational design, it may be expected that most researchers would easily adhere to the spirit and practicalities of data sharing. Paradoxically, therefore, observational rather than randomized data may pave the way towards full implementation of a data sharing culture.