Abstract:
|
In longitudinal clinical trials, we may define the onset of treatment response or treatment failure based on assessing relative change on clinical instruments that are commonly used in clinical practices. In some cases, the defined event needs to be confirmed at pre-defined later visit/visits after the event onset. However, when patients drop out of the study early or get an alternative medical intervention after the initial onset of the event, the subsequent clinical assessments are missing or considered as missing for the latter visits. Therefore, the event (either response or failure) cannot be confirmed and the information becomes censored. In this research, we compare several methods that are commonly used to deal with the sustained or confirmed response or failure when the missing data exist and the censoring may be informative. Simulation results are given to illustrate the methods to be applied and the bias introduced is also assessed for the corresponding method.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.