Premium
Withdrawal censoring and the sequential logrank procedure
Author(s) -
Richardson De Juran
Publication year - 1992
Publication title -
statistics in medicine
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.996
H-Index - 183
eISSN - 1097-0258
pISSN - 0277-6715
DOI - 10.1002/sim.4780111009
Subject(s) - censoring (clinical trials) , statistics , econometrics , log rank test , mathematics , survival analysis , conditional independence , type i and type ii errors
The effects of withdrawal censoring on the power of the group sequential logrank procedure are examined in the context of survival studies. Special attention is given to an assumption that the withdrawal censoring mechanism is only conditionally independent of survival time. Simulation studies of two‐sample, fixed duration trials, under random right censoring, staggered entry and exponential life times, reveal a general decrease in statistical power that is most pronounced when the withdrawals are unbalanced, the majority occurring in the group with the higher hazard rate. Generally, this decrease in power is not of practical concern under typical withdrawal rates except when conditional independence is present in which case large fluctuations in power and type I error rates can occur.