Sharing Research Data to Improve Public Health

Commentary As a charitable research foundation, the Wellcome Trust is strongly committed to ensuring that the outputs of the research we support—including research publications and data—can be accessed and used in a manner that delivers the greatest possible benefit to global health. In common with other research funders around the world, we believe that many of the increasingly rich and complex datasets that are generated by the research we support have to date been largely untapped resources of considerable value. Through making these data more widely accessible and useable to other researchers, they can be used to advance avenues of research that were never envisaged by the original data generator , as well as to verify published research findings. In this light, many funders have introduced explicit policies on data management and sharing that require the researchers we support to maximize access to datasets in a timely and appropriate manner and with as few restrictions as possible (Wykstra, 2013). In implementing their policies, funders are aware that there are significant challenges to overcome. Different research disciplines are at very different stages in terms of developing the infrastructures, resources, and skillsets needed to manage and share data. Some sections of the research community have been much quicker to embed data sharing practices than others, and across the research enterprise, incentive structures do not yet recognize adequately the cost and effort required to generate and make available high quality data resources. Furthermore, for many areas of biomedical research that involve the collection of data from patients and populations, limits on data sharing are required to safeguard the privacy of research participants. Public health and epidemiology research has been highlighted as an area in which tremendous benefits could be gained from data sharing, but where the community has been relatively slow to embrace it and where key enabling infrastructures are not yet in place (Pisani & AbouZahr, 2010). This is in part the result of some justifiable barriers and concerns (van Panhuis et al., 2014). In particular, researchers who invest significant time and effort in compiling large-scale long-term datasets from cohorts and populations are understandably reluctant to share these data before they have had sufficient time to conduct and publish their own analyses. This concern is amplified for researchers in resource-poor settings, where fears are particularly acute that the benefits will be lost to better resourced groups with enhanced capacity for data analysis. …