More organizations with interests in clinical research – most recently leaders in evidence-based medicine the Cochrane Collaboration – are calling for better access to research data. So what’s getting in the way?
In human health research the benefits of increased transparency – more reliable, efficient research to better inform clinical practice – are arguably the most tangible, yet it’s been reported that researchers in this discipline are amongst the least likely to share their data. The case for sharing data and results from all clinical trials is made by Peter C. Gøtzsche (Nordic Cochrane Centre) in a commentary published today in Trials.
Gøtzsche reviews a number of past high-profile cases where lack of access to clinical trial data has been detrimental to human health (such as celecoxib, rosiglitazone and reboxetine) or lead to misplaced tax-payer resources (oseltamivir). Gøtzsche argues that commercial, academic and regulatory interests can lead to a lack of transparency, and that the benefits of data sharing greatly outweigh the potential harms – such as selective interpretation by competing scientists.
Going further than a code of conduct for data sharing previously proposed in Trials, Gøtzsche calls for data to be made available to other researchers for “any relevant purpose”, without needing to pre-specify analyses and obtain permission from the original researchers. Maintaining competitiveness in drug development has been put forward as a reason not to share all data immediately, but Gøtzsche argues that transparency regarding failures in drug research and development could be beneficial, if costly research and development is not unnecessarily duplicated.
BioMed Central encourages and supports the sharing and publication of underlying research data, but recognizes the challenges and caveats associated with different data types and research domains. Requirements for sharing have been successfully established in some fields, for example in genomics where mechanisms and policies and a clear need for collaboration on large datasets exist. In this field data must be released, often immediately, and researchers allowed relatively short periods in which to exclusively analyze the data. Gøtzsche proposes a similar, but legally enforced, supranational model for clinical trials: “It should be a legal requirement to provide all results and raw data within an appropriate period of time, which, in accordance with most calls for data sharing, should be no later than 12 months after the randomized phase of the trial ended”.
The article aims to convince those with doubts that data sharing is needed and does not focus on the associated practical issues to be overcome, but recognizes efforts aimed at developing guidance and best practice to overcome these practical issues, such as those initiated by Trials.
“It is a moral imperative and we should act now, as it [full access to trial data] will empower citizens and convey tremendous scientific, economic, and social benefits,” concludes Gøtzsche.
Central to, or rather the starting point of, transparency in clinical trials is prospective study registration in databases such as the ISRCTN register. Trial registration is required by law in the US and under international ethical guidelines in the Declaration of Helsinki, parts of which are considered as ‘customary international law norm‘. Incentivizing and providing platforms for sharing of all trial-related publications is a major aspect for BioMed Central’s threaded publications initiative.
This substantial, non-commissioned commentary forms part of Trials’ growing thematic series on Sharing clinical research data, edited by Dr Andrew Vickers.