Most biomedical and health researchers who declare their willingness to share data behind journal articles do not respond to access requests or fail to turn over data when asked, study finds1.
Livia Puljak, who studies evidence-based medicine at the Catholic University of Croatia in Zagreb, and her colleagues analyzed 3,556 biomedical and health science articles published in one month by 282 BMC journals. (BMC is part of Springer Nature, the publisher of Nature; Nature’s news team is editorially independent of its publisher.)
The team identified 381 papers with links to data stored in online repositories and another 1,792 papers for which the authors indicated in statements that their datasets would be available upon reasonable request. The remaining studies said their data were in the published manuscript and its supplements, or had not generated any data, so sharing did not apply.
But of the 1,792 manuscripts for which authors said they were willing to share their data, more than 90% of corresponding authors declined or did not respond to requests for raw data (see “Data Sharing Behaviour”). Only 14%, or 254, of the authors contacted responded to email requests for data, and just 6.7%, or 120 authors, actually delivered the data in a usable format. The study was published in the Journal of Clinical Epidemiology May 29.
Puljak was “flabbergasted” that so few researchers shared their data. “There’s a gap between what people say and what people do,” she says. “It’s only when we ask for the data that we can see their attitude towards data sharing.”
“It’s quite appalling [researchers] don’t present the data,” says Rebecca Li, who is executive director of the nonprofit global data-sharing platform Vivli and is based in Cambridge, Massachusetts.
Claims of data availability are of little value because many datasets are never actually made accessible, says Valentin Danchev, a sociologist at the University of Essex in Colchester.
Puljak’s findings match those of a study led by Danchev, which found low rates of data sharing by authors of articles in leading medical journals that state that all clinical trials must share data.2.
Researchers who declined to provide data in Puljak’s study gave varying reasons. Some had not received informed consent or ethical approval to share data; others had left the project, misplaced data, or cited language barriers when it came to translating qualitative interview data.
Aidan Tan, a pediatrician and evidence-based medicine researcher at the University of Sydney in Australia, says the study demonstrates that persistent barriers prevent researchers from sharing their data. His own research looking at clinical trial leaders revealed concerns about data privacy, participant privacy, and the misuse of data in misleading secondary analyses.3. Investigators may also want to publish more original research first or fear being rounded up.
Previous research suggests that some fields, such as ecology, embrace data sharing more than others. But several analyzes of COVID-19 clinical trials – including some from Li4,5 and tan6 – reported that between about half and 80% of investigators do not want or plan to freely share data.
“It is concerning,” says Tan, “that data sharing practices do not appear to have improved at all during the COVID-19 pandemic, despite high-profile calls for data sharing” from funding organizations. such as Wellcome in London, USA. National Institutes of Health and World Health Organization.
Tackling the problem
Li assumes that many researchers don’t fully understand what data sharing really entails: that the data underlying manuscripts “should be ready, formatted, and available to anyone who asks for it,” she says.
To encourage researchers to prepare their data, Li says, journals could make data sharing statements more prescriptive. They could require authors to specify where they will share the raw data, who can access it, when and how.
Funders could also raise the bar on data sharing. The US National Institutes of Health, in an effort to reduce unnecessary and non-replicable research, will soon require grant applicants to include a data management and sharing plan in their applications. Eventually, they will be required to share the data publicly.
“The power to drive data sharing forward lies with those who engage with researchers to change the culture,” Li says.