Duke University
presents:
Keith Abrams
Department of Epidemiology and Public Health,
University of Leicester, U.K
"Bayesian Methods In Cancer Clinical Trials"
Abstract: Often in cancer clinical trials, death is the major event of interest, and in comparing different treatments the survival experiences of patients in the different groups are compared. A brief review of some the Bayesian approaches that have been advocated for the analysis of such data is given.
The use of some of the statistical models reviewed will be illustrated with data from a trial which was designed to assess the potential benefits of a new type of radiotherapy (Neutron therapy) for patients with cancer of the pelvic region. A particular feature of this trial was that considerable a priori information was available regarding the new treatment, in terms of both elicited clinical beliefs and previous trial-based evidence in the form of a meta-analysis.
A particular feature of this trial was that following an unplanned interim analysis the trial was stopped early by which point only half the planned number of patients had been recruited. The elicited prior beliefs and demands played a crucial role in the decision to stop the trial and the role of Bayesian approaches to trial monitoring with respect to this and other cancer trials is discussed.
Friday, December 8, 1995
11:45 - 12:45
116 Old Chemistry Building
***************************************************************************** Professor Abrams will also be speaking at the UNC Biostatistics Department on Wednesday, December 6, 1995 from 4:00 - 5:00 pm in #1301 MacGavran Greenberg Building -- "Model Estimation and Discrimination In Meta-Analysis - A Bayesian Prespective". There will be a reception in the Biostatistics Conference room on the 3rd floor of the MacGavran Greenberg Building. Please contact Professor Bahjat Qaqish with any questions concerning this seminar: