A commitment to act on data sharing

The argument is simple enough. Ensuring that research is registered in advance, the findings reported quickly, and the data and code shared readily is in the best interests of science and the public good. Ultimately, the public pays for the research enterprise, and the scientific community must be held accountable. But nothing that involves commercial interests, academic rivalry, political agendas, and the perverse incentives of scientific publishing is ever simple. What is simple and obvious, such as the importance of sharing data and code from research, rapidly becomes complex and insoluble (doi:10.1136/bmj.o102).1

The movement for more openness and transparency in science, which The BMJ supports, is intimately connected to the movement for evidence based (or evidence informed) medicine, another cause that The BMJ is closely identified with (doi:10.1136/bmj.312.7023.71).2 Both serve the interests of science and the public. Yet both are often presented as idealistic and unpragmatic. In Gulliver’s Travels, our hero encounters a scientific society that is perfecting how to extract sunbeams from cucumbers, an exercise in the absurd.3 Open science and evidence based medicine are not akin to extracting sunbeams from cucumbers. They are, when properly implemented, pragmatic solutions to understanding, disseminating, scrutinising, and implementing research findings.

The choice here, as ever, is whether to tolerate the status quo or agitate for change. Is it sufficient to avoid doing harm or should you also act to promote what is good? On whom does the moral responsibility to act fall? For example, do you stand by as your rivers and waterways are once more polluted with sewage (doi:10.1136/bmj.p1481)?4 Are you content, in another throwback to the 19th century, that abortion remains a criminal offence (doi:10.1136/bmj.p1508)?5 What if the action, when taken, seems good in theory but isn’t well enough thought out, such as England’s new workforce plan (doi:10.1136/bmj.p157)6 or the rapid introduction of physician assistants into clinical practice (doi:10.1136/bmj.p1596)?7

What should you do, then, when you think that a doctor is spreading disinformation about harms of covid-19 vaccines (doi:10.1136/bmj.p1568)?8 Do you do nothing or do you take your medical regulator to court for inaction, as one doctor is doing (doi:10.1136/bmj.p1581)?9 The regulator in this instance, the UK’s General Medical Council, continues to lose the profession’s confidence on several fronts (doi:10.1136/bmj.p1549).10 The GMC has vowed to change, promised a new approach, yet it remains mired in minor complaints and is struggling to demonstrate the impact of the recommendations from the Arora review (doi:10.1136/bmj.o3015).11

Some people act by moving continents, not only for adventure like Gulliver, but for better working conditions, although they might discover that the new world isn’t necessarily better than the old one (doi:10.1136/bmj.p1559, doi:10.1136/bmj.p1536).1213 Equally, the medical profession must take responsibility for making working lives better. A survey finds that many doctors feel discriminated against and burnt out as a result of the behaviour of colleagues (doi:10.1136/bmj.p1601).14 Systemic and institutional factors matter too, for instance when eroding compassion in provision of mental health services (doi:10.1136/bmj-2022-073055).15

Demands and agreements for action don’t readily translate into progress. The UK might now be a “world leader on clinical trials transparency,” but what should be done about scientists who fail to meet their obligations on clinical trial registration and reporting (doi:10.1136/bmj.p1547)?16 One answer, argue Nicholas DeVito and Ben Goldacre, is for regulators and funders to take action by rejecting new applications for trial funding from non-compliers.

The distance between promoting what is good and doing what is good often seems great. While the struggles over trial registration and reporting of results continue, the next obvious step is sharing study data. New research finds that only 8% of medical and health articles declared that their data were publicly available, and a mere 2% shared data publicly. The same estimates for public code sharing were less than 0.5% (doi:10.1136/bmj-2023-075767).17 Sharing data and code is best practice for public accountability and patient outcomes (doi:10.1136/bmj.l6365).18

A group of editorialists say that “journals and regulators should adopt mandatory policies” for data and code sharing (doi:10.1136/bmj.p1434).19 This demand is morally right, and one that we are willing to accede to. The BMJ currently requires data sharing on request (doi:10.1136/bmj.h2373).2021 Authors agree to this on publication, but enforcement proves difficult. Our intention is to strengthen this policy by demanding data sharing, such as by deposition in a suitable repository, for research papers that we offer to publish. We will make code sharing mandatory too. We understand that this next leap in transparency is complex. Researchers might not be able to share observational data from databases. Code might be unwieldy and even inaccurate. But our principle is clear, and we will consider the complexities before the final policy is announced later this year. We might begin by limiting mandatory data sharing to clinical trials, for example, and we welcome readers’ views on how best to make this policy work.

In this case, the moral responsibility to act is on us. We will do so, and we hope others will too. We are not extracting sunbeams from cucumbers; we are shining a light for the public good.