Technical Reports : Time-to-Event Endpoints

 
 

The use of weighted logrank statistics in group sequential testing and non-proportional hazards

  1. Summary: Daniel Gillen’s Ph.D. dissertation (University of Washington Department of Biostatisitics, 2003) that investigates of a number of issues related to analyses in the setting of non-proportional hazards. Results from this dissertation were also reported in the Sequential Analysis 2005, Biometrics 2005, Statistics and Probability Letters 2007, and BEPress 2007 manuscripts listed below


Information growth in a family of weighted logrank statistics under repeated analyses.

  1. Summary: An investigation of the sequential use of weighted logrank statistics in nonproportional hazards.

  2. Ref: Gillen DL, Emerson SS, Sequential Analysis 24: 1-22 (2005).



A note on P-values under group sequential testing and nonproportional hazards.

  1. Summary: Comparison of orderings of the outcome space and their impact on P values under nonproportional hazards.

  2. Ref: Gillen DL, Emerson SS, Biometrics 61: 546-551 (2005).



Non-transitivity in a class of weighted logrank statistics under non-proportional hazards.

  1. Summary: A demonstration of the non-transitivity of weighted logrank statististics under nonproportional hazards.

  2. Ref: Gillen DL, Emerson SS, Statistics and Probability Letters 77: 123-130 (2007).



A random walk approach for quantifying uncertainty in group sequential survival trials.

  1. Summary: An approach for quantifying uncertainty in future treatment effects when monitoring survival in an RCT.

  2. Ref: Gillen DL, Computational Statistics and Data Analysis 53: 609-620 (2009).



Evaluating a group sequential design in the setting of non-proportional hazards.

  1. Summary: An approach to the design and evaluation of a group sequential survival trial that relies upon the use of pre-existing pilot data to construct alternatives under non-proportional hazards treatment effects.

  2. Ref: Gillen DL, Emerson SS, (BEPress) (2007)



Exploring the benefits of adaptive sequential designs in time-to-event endpoint settings.

  1. Summary: When using a time to event as a primary endpoint there are some features that might suggest an advantage to a more adaptive approach.

  2. Ref: Emerson SC, Rudser KD, Emerson SS, Statistics in Medicine (in press).