Identification of Prognostic Factors Related to Survival Time: Nonproportional Hazards Models

  1. Elisa T. Lee and
  2. John Wenyu Wang

Published Online: 30 JUN 2003

DOI: 10.1002/0471458546.ch13

Statistical Methods for Survival Data Analysis, Third Edition

Statistical Methods for Survival Data Analysis, Third Edition

How to Cite

Lee, E. T. and Wang, J. W. (2003) Identification of Prognostic Factors Related to Survival Time: Nonproportional Hazards Models, in Statistical Methods for Survival Data Analysis, Third Edition, John Wiley & Sons, Inc., Hoboken, NJ, USA. doi: 10.1002/0471458546.ch13

Author Information

  1. Department of Biostatistics and Epidemiology and Center for American Indian Health Research, College of Public Health, University of Oklahoma Health Sciences Center, Oklahoma City, Oklahoma, USA

Publication History

  1. Published Online: 30 JUN 2003
  2. Published Print: 4 APR 2003

ISBN Information

Print ISBN: 9780471369974

Online ISBN: 9780471458548

SEARCH

Keywords:

  • prognostic factors;
  • identification;
  • survival time;
  • nonproportional hazards models;
  • time-dependent covariate models;
  • stratified proportional hazards models;
  • competing risks model;
  • multiple events models;
  • related observations models

Summary

There is only one cause of failure in the proportional hazard model for identification of important prognostic factors, in which the covariates are assumed to be independent of time, that is, the event or failure is allowed to occur only once for each person, and there is no correlation among failure times of different persons. However, in practice, failure may be due to more than one event or cause, the same event or failure may recur during a follow-up study, and the event or failure time observed may be from related persons in a family or from the same person at different times. In this chapter we discuss several models for these situations. The first two models are extensions of the proportional hazards model to handle time-dependent covariates and to perform stratified analysis. Other models introduced in this chapter are for multiple causes of failure, recurrent events, and related observations. The chapter concludes with a problem solving section.