In the third part of our blog series from the Open Pharma roundtable ‘Pharma and Publishers Forum’, we discuss the role of data sharing as a critical element of open science. Here, we highlight key barriers to data sharing from both the pharma and publisher perspectives, and how they might be overcome. The final report in this series, focusing on plain language summaries, will be released in the coming weeks.
Key talking points
The roundtable participants believe that data sharing increases both the public trust in, and the perceived value of research. Authors, publishers and pharmaceutical companies have increasingly recognised these benefits and endeavoured to increase their data sharing practices. However, many researchers and companies find data sharing confusing and risky. Best practice for data sharing is often unclear, and data privacy requirements vary between nations and pharma companies.
These concerns hamper the efforts to reach the data sharing ideal of “as open as possible, as closed as necessary” that is in line with the FAIR Data Management Guidelines.
Challenges and solutions
Authors are not always aware of what best practice looks like
One of the participant publishers at the roundtable pointed out that, when asked, the majority of researchers and authors publishing in their journal often believed they were doing well in complying with the journal’s data sharing protocol. However, many authors are simply submitting the lowest level of detail required by the journal, sometimes as minimal as just stating the location of the data.
These observations demonstrate a disparity between data sharing best practice and authors’ understanding of best practice. The roundtable participants suggested that authors would benefit from education and training on data sharing early in the research process. This support for authors would include guidance on what to share, which repositories to use and how to manage data sharing in an effective manner that complies with data protection laws. The participants believed that these educational measures could improve authors’ understanding of data sharing policies and their confidence in choosing more open data sharing options.
Data privacy requirements can be difficult to meet, and the risks of harm can outweigh benefits
Data protection laws vary from country to country, and even from region to region, as in Switzerland and the USA. Failure to comply with these laws can have serious consequences, from risks to patient safety and trust in the research, to a shattered career reputation or prosecution. The risks of data becoming de-anonymized, particularly sensitive varieties such as clinical trial patient data, quite quickly seem to outweigh the benefits of data sharing for involved parties. In addition, if guidelines change during an ongoing trial, pharma companies might not be able to re-obtain patient consent in a reasonable time frame to meet the new data-sharing requirements.
Some data repositories are addressing concerns over sensitive data handling. Dryad uses salaried workers to flag up and investigate sensitive data, and dbGaP has been used successfully by some pharma companies for the same purpose. Furthermore, dedicated teams within some pharma companies have been successfully set up to ensure transparent access to data alongside protected patient anonymity, and this could prove a valuable model to follow.
The roundtable participants suggested that creating and using universal data-sharing guidelines could further help to build researcher and institutional confidence in complying with data sharing best practice. These guidelines should be used across journals, companies and data sharing systems. Looking at other fields that have had high success rates with data sharing, such as psychology, could also help the pharma industry learn about the best ways of sharing sensitive data.
Differences in data sharing policies between pharma companies and publishers can cause problems
The participants highlighted that authors and pharma companies are sometimes reluctant to change their usual data repository, even when the new repository might have some advantages from a publisher’s point of view. This is perhaps due to a gap in education and guidelines, with some pharma companies choosing their own preferred repository based on convenience, control over their data and their data sharing ethos. However, several companies are already using common repositories, such as ClinicalStudyDataRequest.com, where the data requests are reviewed by an independent committee. Still, the roundtable participants noted that while multiple separate policies and initiatives focusing on data sharing, do exist, the lack of a single unified policy proved a hindrance when pharma companies decided to work together to share data.
The group favoured a collaborative approach to developing data sharing policies. Pharma companies and publishers should seek harmonization in the systems and processes they recommend for data storage. This harmonization would make it easier to handle data and share them appropriately.