In earlier education sections, we have considered the formulation of the question for a study (1), the outcomes to measure (2), and the necessary sample size (3). In this issue, we discuss the choices that have to be made when deciding who should be eligible for the study.
Formulating the question for a trial should help to determine the interventions to be compared and the population to be recruited. The rules setting out who is eligible should be described in the inclusion and exclusion criteria for the trial, and these can be broad or narrow (4). The inclusion criteria might be kept very narrow if a trial is answering a focused, explanatory or efficacy question, but if it is seeking to determine the effects of interventions in circumstances that are closer to routine practice, the eligibility criteria need to be such that that the variety of types of patient that might be considered for the treatments in the future would be eligible. In these pragmatic or effectiveness trials, the eligibility criteria are likely to be broad.
In designing randomised trials so that they can be as large and wide ranging as possible, the “uncertainty principle” can be used to define the eligibility criteria (5). This means that people are eligible for the trial if there is sufficient uncertainty about whether the effect they would experience would differ between the interventions being compared. The people making the decision about someone's participation in the trial need to have this uncertainty. That includes the patient themself and those responsible for their care. Implementing the uncertainty principle in a trial requires that patients who are likely to be harmed by either interventions are excluded, as are those for whom one intervention is expected to be much better than the other. All other patients would be eligible, although it must be remembered that some of these might still not be suitable if, for example, they were not capable of providing the necessary outcome data or did not wish to be part of the research study.
A large and influential trial which used the uncertainty principle as the basis for its eligibility criteria is the International Subarachnoid Aneurysm Trial (ISAT), which took place in the 1990s and early 2000s (6,7). Patients with a ruptured intracranial aneurysm that had produced a subarachnoid haemorrhage were randomised to neurosurgical clipping or endovascular coiling of the aneurysm. If either treatment was clearly preferred for a patient, or if one of the two treatments needed to be avoided for them, the patient was not eligible. However, if there was uncertainty about the relative effects of the two interventions, the patient was eligible and would be offered participation in the trial. This allowed ISAT to become the largest ever randomised trial in the treatment of subarachnoid haemorrhage, randomising more than 2100 patients (8). This sample size gave it the power to identify moderates differences on the primary outcome of death or dependency at one year after randomisation, which it did. There was a significant difference in this outcome between the patients allocated neurosurgical clipping (326 bad outcomes, 30.9% of patients) and those allocated endovascular coiling (250, 23.5%). Based on this relative benefit of 24% for patients allocated coiling, 74 more patients would be alive and independent after one year among every 1000 patients treated. The use of the uncertainty principle by ISAT meant that more patients were eligible for the trial, more patients were recruited for it, and more future patients are able to benefit from its results.