In conducting, participating in and teaching about randomized clinical trials, I've found it useful to recognize that uncertainty exists at 3 levels. Because the levels each have unique properties, problems and solutions, they must be clearly distinguished.
The first level is community uncertainty, where sufficient numbers of clinicians, methodologists and ethics committees must become sufficiently uncertain whether an intervention is beneficial for a randomized clinical trial of the intervention to be judged both necessary and appropriate; the trial's data safety and monitoring board later resolves this community uncertainty in light of the emerging results.
The second level is the uncertainty of individual clinicians who are deciding whether to join a randomized clinical trial and then, if they join, whether or not to offer trial participation to any of their patients (for example, some clinicians were certain that endarterectomy was beneficial in symptomatic carotid stenosis and refused to join the North American and European trials in which only half their patients would undergo this operation).
The third level is uncertainty at the level of the individual partnership between patient and clinician, where unless both of them are uncertain which arm of the trial is better for the patient, the patient doesn't join the trial.
I think Francis Rolleston has both nicely described these 3 different levels of uncertainty and correctly pointed out that my original essay was concerned primarily with the third level of uncertainty, that within the individual patient–clinician partnership.1 Alas, he then proposes retaining the term “clinical equipoise” to denote the first level, community uncertainty. The disutility of his proposal is immediately revealed in the letter from Ian Shrier, who evokes clinical equipoise in addressing the third level of uncertainty.
I thank these 2 correspondents for making my point.
Reference
- 1.↵