Answers are grounded in this paper and cite their sources.
Conducting randomized controlled trials (RCTs) of potentially effective treatments is time consuming and expensive, and for industry sponsored studies, the information is commercially sensitive. The notion that information from RCTs can be shared to ensure openness and transparency has been proposed by a number of groups, but at the present time there is no consensus on how data sharing should be carried out or the balance of benefit and risk inherent in this process.1–3 Naci et al.4 argue that providing original datasets from clinical trials in an open access anonymized format will avoid publication bias where ‘negative’ trials (which do not show a difference between treatment and control) have less chance of being published, resulting in important information not being available. Other reasons for making clinical trial data available are to improve reproducibility and validity, pool data across different studies to provide better estimates of treatment effects (especially using individual patient data), and to provide a more efficient method to improve care by using the available evidence in a timely manner. As Naci et al. point out, many organizations including the US National Institutes of Health and the UK Medical Research Council already allow access to their data under a data-sharing agreement. This process could be termed a ‘limited’ data-sharing policy rather than open access, which implies a very simple procedure to access and download data. The reasons for data sharing are coherent and sound so why is there not a stampede among researchers to provide open access clinical trial data in the public domain? There appears to be three main reasons obstructing the cause for open access: It appears that there is very little incentive or reward for researchers to post their data in the public domain and perhaps, this is an area to explore further. Incentives could include:3 An example of where data sharing may be helpful is the recent controversy about recommendations from the Cholesterol Treatment Trialists' (CTT) Collaboration about potential benefits of statins for primary prevention.5 Articles questioning these findings were felt to be erroneous by the CTT investigators who requested retraction.6,7 An independent panel set-up by the British Medical Journal (BMJ) found some errors, but did not recommend formal retraction. Fiona Godlee, Editor of BMJ, suggested that independent analysis of the CTT data under data-sharing schemes would help to resolve important aspects of this controversy.8 Clinical research is a competitive environment. Posting hard earned datasets in the public domain allows competitors easy access to publish contrary or misleading views. This is not popular among research groups who have spent years conducting trials. Concerns that many individuals or groups who can access public domain clinical trial data may not have the expertise to manage and analyse the data appropriately leading to spurious or incorrect analyses. Data may also be accessed by people with an intent to discredit the original findings. Early access to clinical trial data that is commercially sensitive could lead to loss of patents and intellectual property and be financially damaging. Invitation to co-author any publications from groups which use open access datasets. Agreement among leading journals that posting data in an open access environment increases transparency and validity. This could become a pre-publication requirement in the same way as registration on a recognized clinical trials' website. For clinical trials of new drugs or devices, regulators could mandate that trial data are posted on an official website that has restricted access for a period until results are presented and published, but subsequently data are more easily accessible. Current models of data access by public funding organizations, such as NIH or MRC, are setting the trend for increasing transparency and validity and reducing the risk of publication bias.3 The NIH BioLINCC model requires a registration process, validation of reason for data access, and an agreement of how the data can be used.9 This process seems to work well to ensure that data are accessed by genuine researchers with good intentions, and helps to address some of the barriers to open access. However, this is still far from the genuine open access model and does not address the issues of commercial sensitivities which may take far longer to resolve. In the meantime, Naci et al. are to be commended for providing a well-argued academic case for open access of clinical trial data.
No takes yet. Share an insight, caveat, or question.
Marcus Flather (2015) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: