Randomisation understood but not accepted

There is evidence that patients may neither understand nor accept randomisation in clinical trials. In a study in North Staffordshire (

–4) adults attending further education classes appeared to understand randomisation but not accept it.

The 130 participants in the study were aged between 18 and 70 (mean 32 years), 67% were women, and their occupations ranged from unskilled to managerial or professional. Most were not formally educated beyond GCSE or O levels (usually taken at 16 years). They were given two hypothetical scenarios (one medical and one nonmedical) in which people needed to be allocated to two groups. Five methods of allocation were proposed: computer allocation with no information about individuals, toss of a coin, drawing from a hat, personal preference, or allocation in turn. They were asked to decide for each of these methods whether it was random or not random. Most participants (75–90%) judged that computer allocation, toss of a coin, and drawing from a hat were random methods and asking people for their preference was not random. They were more or less equally divided about the status of allocation-in-turn. Similar answers were given whether considering the medical or the nonmedical scenario.

Participants were then asked to imagine they were to take part in a clinical trial comparing two drugs both known to be beneficial. They had to decide which of the same five methods would be acceptable for allocation to one or the other drug. Half were given a brief written justification for the use of randomisation and half were not. Among participants who correctly judged each method to be random or not random in the first part of the study most (60% if given the written justification for randomisation, 75% if not) considered allocation by patient preference acceptable. Most (62–72% of the group not given the written justification) thought that computer allocation, toss of a coin, drawing from a hat, or allocation-in-turn were not acceptable. The written justification did not change attitudes except towards computer randomisation which was acceptable to 38% of participants not given the justification and 58% of those given it.

Most people are able to distinguish between random and non-random methods of allocation but most would not find randomisation acceptable in a clinical trial. Computer allocation is seen as more appropriate than tossing a coin or drawing from a hat. More needs to be done to explain the reasons for randomisation in more detail and to address people’s questions.

Question—If some people choose drug A and some drug B, what’s wrong with giving them their choice and seeing how they get on? Answer—To obtain reliable results we must compare like with like. The purpose of randomisation is not just to allocate people to groups but to try to make sure that the groups are as alike as possible; one group is not older, sicker, or different from the other in any way that might affect their response to treatment. With large enough numbers of people any differences should be evened out by randomisation; chance should make it likely that more or less the same numbers of older or sicker people (or those different in ways nobody has thought of) are allocated to each group. Giving people their choice would not achieve that and the results of the trial would be unreliable; we would still not know for sure which drug was better. It would, of course, be completely unethical to allocate anybody to a treatment known beforehand to be inferior. Response—O.K. I’ll buy that.