06 Oct 2011 We need access to all data from all clinical trials
In an editorial published this week in The Cochrane Library, Peter Gøtzsche (Director, The Nordic Cochrane Centre) calls for access to all data from all clinical trials. The editorial exposes the pernicious practice of withholding clinical trials data by sponsors of studies and illustrates how the consequences of selective reporting can be disastrous (e.g. cox-2 inhibitors, antidepressants).
- All randomised clinical trials to be registered at their inception, before recruitment of the first participant (see the Cochrane statement);
- All data from all randomised clinical trials, including raw anonymised individual participant data that do not allow identification of individual participants, and the corresponding trial protocols to become publicly available free of charge and in easily accessible electronic formats;
- Governments to consider introducing legislation that makes it a requirement to provide these data from all trials to the public within 12 months from the end of the randomised phase of the trial, in accordance with most international calls for data sharing;
- Governments also to consider punitive measures for non-compliance; a requirement to continue to hold and make available core data indefinitely, or to pass such data to a central and accessible repository; and a recognition that ownership of trial data should be shared between sponsors, investigators and trial participants.
The Cochrane Collaboration recognises that the sharing of data from all clinical trials would lead to tremendous benefits for our societies for the following reasons: the current situation is harmful to patients and has been shown to lead to the avoidable death of tens of thousands of patients; we would become much better informed about the true benefits and harms of our interventions, which would lead to better treatment with fewer harms throughout health care; transparency would be increased, making it possible for independent scrutiny of the methods and the calculations reported in the trial publication compared with the trial protocol and the raw data, which would increase the likelihood that any malpractice was detected; the efficiency of healthcare research would be much improved, as many important research questions can be answered by using existing data, sparing researchers and patients from unnecessary, potentially dangerous and wasteful duplication of effort; and it would help identify healthcare strategies and uncertainties that require research, and to set priorities for research.