Seminars

General Departmental Seminar Series

Sequential Methods in Data Monitoring of Clinical Trials: Theory and Practice

David DeMets
Department of Biostatistics and Medical Informatics
University of Wisconsin-Madison

Friday, September 10, 1999, 12:00-1:00 pm

E5/584, Clinical Sciences Center

ABSTRACT

A brief review will be presented of group sequential methods typically used in the interim analysis of clinical trials. These methods are used to evaluate accumulating data for evidence of harm or early and convincing treatment benefit. The focus will largely be on group sequential methods using the alpha spending function concept as described by Lan and DeMets (1983). In addition, a few current topics of research in this area will be mentioned. The application of these methods will be illustrated by two recently completed heart failure clinical trials, VEST (1998, NEJM) and MERIT (1999, Lancet), with the Department serving as the independent statistical analysis center. The VEST trial was terminated with a harmful treatment effect while the MERIT trial was stopped at the midpoint with overwhelming benefit, both trials using mortality as the primary outcome. These trials raise several interesting clinical, statistical and ethical issues.

Back to General Departmental Seminar Series