Practicalities of using placebos or blinded treatments

As with the term 'randomization,' a trial which is said to be double-blind may by anything but. Despite best efforts, treatments may not be perfectly matched [30]. Whether this is a problem or not may depend on the curiosity and motivation of the subjects. For example, a group of patients within a trial could compare their capsules and look for differences in internal or external appearance. While not being able to tell which is active and which placebo, pooling and dividing the capsules so that each patient has some of each could satisfy the desire of a desperate patient to have at least some active treatment. If this happened on a large scale, it would make the trial impossible to interpret.

This hypothetical example illustrates the need for great care in matching treatments and gives an indication of the additional level of complication and potentially considerable cost which treatment blinding may bring to a trial. In addition, as discussed below, if the side effects of the treatments being compared are known to be very different, it will soon become clear which patients are on which treatment. For the purposes of blinding the clinician therefore, it would not be sufficient simply to label the treatments A or B, since it takes only one patient with a 'classic' reaction to drug B to unblind the trial to the trial physicians at least. Rather, the randomization list should be prepared such that each patient in the trial is allocated a code, and a separate list (often held by an independent member of the pharmacy staff) identifies which codes correspond to which treatment. There must be a facility to break the code - not just for the final trial analysis, but also in case of emergencies. One example is the unanticipated need to give a patient treatment, perhaps for an unrelated condition, which may interact badly with one of the trial therapies even if they are stopped (naturally any patient for whom such a need is a known possibility should not have entered the trial in the first place since they were not 'fit to receive either treatment,' an eligibility requirement for any trial-see Section 4.2.3). This again emphasizes the need for a complex coding system for which knowledge of one patient's true allocation does not reveal any other patient's allocation.

0 0

Post a comment