Multivariate Failure Time Data: Representation and Analysis
While univariate failure time methods, including Kaplan-Meier survivor function estimators, censored data rank tests, and Cox regression procedures are well developed, corresponding flexible, standardized tools are not available for multivariate failure time analysis. This paper considers methods for the modeling and analysis of clustered failure times, with a focus on the estimation of marginal hazard functions and pairwise cross-ratio functions. First some representations of bivariate failure times are reviewed, along with corresponding nonparametric estimators and summary measures of pairwise dependence. Then procedures are outlined for simultaneous estimation of marginal hazard ratio, and pairwise cross-ratio parameters, for use in more general multivariate failure time regression problems. These estimation procedures are somewhat restrictive concerning the form of pairwise dependencies between failure times. Some approaches to relaxing these restrictions are briefly mentioned.
KeywordsFailure Time Frailty Model Failure Time Data Semi Parametric Model Frailty Variate
Unable to display preview. Download preview PDF.
- Cox, D.R. (1972). Regression models and life tables (with discussion). J.R. Statist. Soc. B, 187–220.Google Scholar
- Lee, E., Wei, L.J., and Amato, D.A. (1992). Cox-type regression analysis for large numbers of small groups of correlated failure-time observations. In Survival Analysis: State of the Art, Eds. J.P. Klein and P.K. Goel, pp. 237–247. Klewer Academic Publishers.Google Scholar
- Prentice, R.L. and Hsu, L. (1997). Regression on hazard ratios and crossratios in multivariate failure time analysis. Biometrika, in press.Google Scholar