Abstract:
|
Sample sizes for clinical trials with comparison of survival data are calculated based on the expected number of events. In some trials, the number of events occurring until the end of planned follow-up period is lower than the expected number. One reason is that the actual survival probabilities of groups are higher than the expected survival probabilities. Other reason is that the experimental treatment is more effective than the expected effect. Trials should be continued if the former reason is true, however, trials should be finished if the latter reason is true. To determine whether to finish a trial, a sample size or an extension of follow-up period is re-calculated using blinded data in the case that an interim analysis is not planned. In this paper, we propose Bayesian method with historical data for blinded sample size re-calculation and compare with EM algorithm based method.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.