Restrictions on the number of patients who can be treated in one arm of the trial

This situation can arise when one of the treatments is in short supply, perhaps a drug is only being made available for use in the trial, and the company supplying the drug impose a restriction on the number of patients they will provide free treatment for. This was the situation at an early stage of the ICON3 trial in advanced ovarian cancer, which compared platinum-based therapy with paclitaxel plus carboplatin [13]. At the launch of the trial, paclitaxel was not widely available in the UK and Europe outside a trial setting. Initially, the company funding the trial would only provide free paclitaxel for 200 patients. Allocating equal numbers of patients to the two arms would give a total of only 400 patients which was insufficient to detect even quite large differences in survival. The trial was therefore designed with a 1: 2 allocation to paclitaxel and control, enabling 600 patients to be entered into the trial, increasing the power for the comparison. In this

50 60 70 80 90 100 Percentage of patients allocated the test treatment cent, test group survival = 60 per cent

Fig. 5.9 Power according to allocation ratio; control group survival = 50 per case, it would theoretically be possible to use an even more extreme ratio, because adding extra patients would always increase the power relative to the smaller size trial. However, in practice there were concerns about the willingness of patients to accept randomization into such a trial, knowing that their chances of being allocated paclitaxel were not 1: 1 or even 1 : 2 but in fact very much less.

0 0

Post a comment