Choice of design

How would you choose between designs in a given situation? There is no clear-cut answer. Deciding factors include the importance of stopping early if a treatment is clearly insufficiently active - this might be high if there is an alternative treatment, in which case one may wish to avoid a single-stage design such as Fleming's. On the other hand, if you have good reason to expect a response rate above 50 per cent, the Gehan design is perhaps unappealing; the first stage would require only between two and four patients at conventional error rates. Both the Gehan and Simon designs allow early stopping if a treatment is inactive; the Simon design provides a numerical rule for deciding, at the end of the second stage, whether the treatment is worth pursuing. Although this may be useful, the final decision will also need to incorporate the information gained on toxicity and general feasibility and the structure of the Gehan design may sit more comfortably with the less prescriptive aims of many phase II trials. Finally, the benefit of early stopping is lost if the outcome measure is not a short-term outcome such as response. In cancers for which radiological response is particularly difficult to assess, it may be appropriate to use a later endpoint such as progression-free survival. In such situations, the Fleming design maybe most appropriate; an example of such a trial in stage I testicular teratoma is given in Box 5.1.

0 0

Post a comment