A particular random number table, read from the top left-hand corner, generates the digits : 75792 78245 83270 59987 75253.

Suppose we have three treatments, and we aim to have equal numbers on each, and thus use the digit allocation described in section (b) above. This would give the treatment allocation sheet shown in Fig. 4.7.

Note that here, while wishing to have equal treatment allocation, we actually have four patients each on treatments 1 and 2, but 10 on treatment 3. In practice with trials of 10s or 100s of patients, simple randomization may by chance lead to such imbalances, or to fairly long runs of the same treatment. In very small trials, the impact on the overall power of the study can be substantial, but in general you may wish to have slightly more control over the overall treatment allocation than is possible with simple randomization. A number of methods are available to permit this, while maintaining the essential elements of randomization; these are described in Sections 4.7.3—4.7.5.

0 0

Post a comment