In a recent editorial from the BMJ, arguments for sharing data were once more repeated (ensures reproducibility, allows testing of secondary hypotheses, facilitates development of new statistical methods, provides a resource for teaching, aids design of new studies, simplifies data acquisition for meta analysis, and helps prevent fraud and selective reporting). Although substantial support to the principle of open access to data has come from many sides, including the IEA, “the system is broken and needs fixing”, BMJ claims, and we agree.
The BMJ suggests that time has come for scientific journals to refuse publication unless data are openly available and for funding agencies to request that the data they have paid for are available to external researchers. Many journals and funding agencies have this policy now, but they do not always follow up on the requests or accept access by committees that are not independent of the research groups that generated the data. Such a committee may be more interested in preventing open access than providing data to others.
The IEA supports the idea of making data freely available, and time may have come to use brute force to implement the policy, as BMJ suggests. We do not, however, believe that problems will be solved just by refusing publication or closing funding if raw data are not freely available. Time is still available for using both the carrot and the stick methods.
First of all, making data useful for other researchers is not a simple job. For complicated large-scale studies it requires several man years of work. Funding agencies must add this cost to their grants. Furthermore, we need to find some way of recognizing people who spend years of their lives collecting data and then see this investment disappearing and others benefit while they are busy cleaning and documenting the data source for others to use. The solution is not to violate the International Committee of Medical Journals co-authorship rules (known as the Vancouver rules, i.e. a co-author should be involved in the conception and design of the scientific work or in the analysis and interpretation of the data), as clearly happens many of the large GWAS consortia and in other areas of collaborative research. Part of the solution is to recognize that building an infrastructure for research is an important part of epidemiology, and those who do this should be properly rewarded for their efforts. The IJE initiative of accepting papers that describe an open data source is a step in the right direction, but more is needed. At present, this effort does not count much in any bibliometric measures. This is not what you should do to improve your h-index.
Jørn Olsen, Cesar Victora, Neil Pearce, Shah Ebrahim