UW Biiostatistics & Medical Informatics UW Biostatistics & Medical Informatics UW Madison UW Biostatistics & Medical Informatics Site Map
UW School of Medicine and Public Health UW Madison



Other Seminar Series


General Departmental Seminar Series

Switching Endpoints in Clinical Trials with Simple and Composite Outcomes

Yonghua Chen, Graduate Student, Departments of Statistics and
and Biostatistics, University of Wisconsin

October 8, 1999 at 12:00-1:00 pm

G6/164, Clinical Sciences Center, 600 Highland Avenue


Mortality is an important outcome in many clinical studies. Since it may take a long time to obtain enough information or events for a death outcome (simple outcome), some composite outcome such as death plus disease occurrence (or recurrence) or death plus hospitalization is often used as the primary outcome. The simple outcome can be viewed as both an efficacy and a safety outcome and is almost always included in the interim analyses for ethical purposes. The trial would be stopped if this simple outcome shows significant adverse effect or benefit in interim analyses while the primary outcome or the composite outcome will be tested at the last analysis, if the trial continues to the end.
Since we test the simple outcome in the interim analyses and ``switch'' to the composite outcome at the last analysis, appropriate adjustment is needed to protect the type I error rate. Two ``switching'' methods are proposed in our paper. The first is a Bonferroni method in which we monitor the simple outcome by using an alpha-spending function and spend all the rest alpha on the composite outcome at the last analysis. The second method is to adjust the alpha spent on the composite outcome based on the joint distribution of the test statistics which hopefully could be more powerful. Since the joint distribution of the test statistics is derived under the assumption that the joint distribution of the simple and the composite outcomes is identical for both the treatment and the control groups, the Bonferroni method which assumes the identical marginal distributions seems more natural. Further, since typically a small alpha is spent on the simple outcome in the interim analyses, Bonferroni method may not be as conservative as we usually expect. The Bonferroni method could be a simple, natural and efficient solution to the switching problem.
Simulated trials are used to compare the two methods with the fixed composite-only design. The two switching procedures enable us to stop the trial early and are essentially as powerful as the fixed composite-only design. Bonferroni method is slightly less powerful than the switch-procedure using full correlation structure. At the end, we apply both methods to a real clinical trial, Praise-1. The trial could have been stopped early if either of the two methods was applied.

Back to General Departmental Seminar Series


Internal Use | Site Map | Search |
Overview | People | Training | Research | Seminars | Employment | Links |
Biostatistics Program | Clinical Trials Program | Medical Informatics Program | Biomedical Computing |

Copyright © 2006 The Board of Regents of the
University of Wisconsin System


UW Madison UW School of Medicine and Public Health