|This month's CRASH Newsletter looks at the issue of sample size in head injury trials
|The histogram below shows the size of previous trials in head injury and is based on a search for all previous trials by the Cochrane Injuries Group. The curve shows the smallest absolute risk reduction detectable in a trial of a given size. The dotted line shows that only about 8 of the 203 previous head injury trials would have
|been large enough to detect a 10% reduction in death or disability from 20% to 10%. No previous trial would have been large enough to detect a 5% reduction in death or disability from 20% to 15%. These are large reductions and it is increasingly recognised that we cannot expect reductions this large in the treatment of head injury.
|Because even a 2% reduction in death or disability following head injury from an intervention as widely practicable as corticosteroids would represent a worthwhile benefit, CRASH has been planned to be able to detect a benefit of this size. As the graph shows, many
|thousands of head injury patients will need to be randomised. If we succeed, CRASH will be the largest trial in head injury ever conducted, but more importantly, will be the first to reliably detect or refute a realistically moderate but clinically important effect. Size matters!
|One reason why it’s been difficult to conduct large head injury trials may be the confusion about informed consent in unconscious patients. Dr Bernard Foëx, collaborator at Royal Bolton Hospital with a special interest in research ethics, writes on the issue of informed consent in CRASH:
|"CRASH is a 'no consent'
trial. In the UK, it has been agreed nationally, and at local level, that
consent does not need to be sought to enrol patients. Ethics committees
in several other participating countries have also agreed to this. Is this
Consent and informed consent form no part of the Hippocratic tradition. Informed consent was mandated for any research on human subjects in the Nuremberg Code of 1947. The 1964 Helsinki Declaration upheld this view. But neither of the codes addressed the problem of research in critical care settings, where, patients are unable to give informed consent. Does this mean we cannot conduct research to find better ways of treating them?
|Of course not. However, it does require a careful consideration of the principles of research ethics. This was done most effectively by the investigators in the ISIS-2 Trial. They invoked the principle of "uncertainty" in the treatment of acute MI. The treatments offered in the trial were all "acceptable" by the standards of the time, but, there was "substantial uncertainty" about which was the most effective. This is also the case in CRASH. There is substantial uncertainty about the benefits of corticosteroids. The rationale is that it would be ethically unacceptable to continue not giving steroids if they could be of benefit. The question of whether there is any benefit from giving steroids can only be answered by doing this trial."
There are now over 130 patients in the CRASH trial…Congratulations to all CRASH collaborators!
CRASH Co-ordinating Centre, FREEPOST LON 14211, LONDON WC1N 1BR
Tel: + 44 (0)20 7299 4684 Fax: + 44 (0)20 7299 4663 email: CRASH@lshtm.ac.uk