Abstract:
|
The negative binomial model is applied in the design and analysis of many clinical trials involving count data. Examples of endpoints for which this can be relevant include seizures in epilepsy, relapses in multiple sclerosis, bleeding episodes in coagulopathies, and exacerbations in pulmonary diseases such as chronic obstructive pulmonary disease and asthma. A closed-form sample size formula is available for the negative binomial model provided one actually knows the over-dispersion parameter. However, in some cases the over-dispersion is not known or there is no reliable historical information from which to estimate it. In this presentation we will consider an unblinded sample size re-estimation procedure that is restricted to only updating the unknown over-dispersion parameter at an interim analysis, and assessing any inflation of the type I error which may exist through the mean-variance relationship of the negative binomial model.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.