The need to secure informed consent is well established in individually randomized trials, as governed by principles dating back to the time of Hippocrates.
Norms regarding the need to obtain informed consent in trials randomizing intact social units have yet to receive full acceptance.
This is partly due to the great diversity that can be found in the size and nature of units that can be randomized (e.g., families, hospitals, cities) and partly because consent can be obtained, at least theoretically, at multiple levels.
These CIOMS guidelines suggest that the gate-keeper should sign a consent form clearly setting out the steps that will be taken for safeguarding the interests of the study participants. The implication here is that such a contract will adequately substitute for the need to obtain informed consent on an individual basis, which from a practical perspective will often be extremely difficult, if not impossible. This approach is also consistent with what has been recommended for “cluster-cluster” trials, in which the intervention is administered to the entire cluster, as in the case of a media message, rather than on a one-to-one basis to individual subjects (Edwards et al., 1999). In the latter case (“cluster-individual” trials) both the need and ability to obtain informed consent at the individual level are arguably much greater than when the intervention is “indivisible” at this level. However if the parent cluster has already been randomized in a cluster-individual trial, the limitation remains that one can only ask an individual subject at this stage to consent to an intervention that has been previously assigned.
Given the complexity of the ethical issues raised by this design, it would seem reasonable, at least as a first step, for researchers to more thoroughly report the steps taken to obtain informed consent in their own trial. However an ultimate goal would be to develop broadly acceptable norms that can be applied to a range of ethical issues that have yet to be adequately explored in the context of CRTs. Further discussion may be found in Klar and Donner, 2007a.