|
Abstract:
|
Because of the (by definition) large size of megatrials, there are necessarily smaller amounts of information collected on each patient. While this is usually an economic and practical necessity, we note that this need hold only on the average. Thus, features of smaller trials can be incorporated for subsets of patients (e.g., more safety data for the first "n" individuals, added testing on subjects with some specified baseline characteristics). In addition, adaptive designs for prespecified subsets may be incorporated in a partial sense, embedding prespecified smaller trials within the megatrial---albeit with some penalty to be paid.
|
