|
Activity Number:
|
246
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, July 31, 2007 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Biopharmaceutical Section
|
| Abstract - #308466 |
|
Title:
|
Nonparametric Confidence Intervals for Survival Probabilities
|
|
Author(s):
|
Kun Chen*+
|
|
Companies:
|
Bristol-Myers Squibb Company
|
|
Address:
|
5 Research Parkway, Wallingford, CT, 06492,
|
|
Keywords:
|
Kaplan Meier product-limit method ; confidence interval ; survival probability ; effective sample size ; Mantel-Haenszel
|
|
Abstract:
|
Kaplan Meier (KM) product-limit method has become the standard approach in estimating a survival distribution for censored data. Popular statistics are KM estimated probability of surviving beyond a specified time and median survival time. Simon and Lee's paper (1982) presents the method of obtaining CI for the probability of survival beyond a specified time. In this paper, I extend Simon and Lee's method to construct CIs for the difference of survival probabilities of two treatment groups with/without considering the strata effect. Effect sample size at that specified time is used instead of the regular sample size. The Mantel-Haenszel (MH) weighting procedure is used to adjust for strata effect. The method can be applied in oncology Phase II and III trials.
|
