Flexible Implementations of Group Sequential Stopping Rules Using Constrained Boundaries
Article first published online: 11 DEC 2003
Volume 59, Issue 4, pages 770–777, December 2003
How to Cite
Burington, B. E. and Emerson, S. S. (2003), Flexible Implementations of Group Sequential Stopping Rules Using Constrained Boundaries. Biometrics, 59: 770–777. doi: 10.1111/j.0006-341X.2003.00090.x
- Issue published online: 11 DEC 2003
- Article first published online: 11 DEC 2003
- Received August 2001. Revised May 2003. Accepted June 2003.
- Clinical trial;
- Error spending function;
- Group sequential;
- Interim analyses;
- Stopping rule
Summary. Group sequential stopping rules are often used during the conduct of clinical trials in order to attain more ethical treatment of patients and to better address efficiency concerns. Because the use of such stopping rules materially affects the frequentist operating characteristics of the hypothesis test, it is necessary to choose an appropriate stopping rule during the planning of the study. It is often the case, however, that the number and timing of interim analyses are not precisely known at the time of trial design, and thus the implementation of a particular stopping rule must allow for flexible determination of the schedule of interim analyses. In this article, we consider the use of constrained stopping boundaries in the implementation of stopping rules. We compare this approach when used on various scales for the test statistic. When implemented on the scale of boundary crossing probabilities, this approach is identical to the error spending function approach of Lan and DeMets (1983).