Stopping a clinical trial very early based on unplanned interim analyses: a group sequential approach.

In the conduct of a clinical trial, unexpectedly high rates of toxicity may cause a researcher to want to terminate the trial early even though no formal stopping rule had been specified. The experience of one such clinical trial is used as an example of the ways in which group sequential methodology can be applied in deciding to stop the study, as well as in reporting the results of the clinical trial. This approach is then compared to a Bayesian analysis.