TIME ON TRIAL ESTIMATES WITH BIVARIATE RISKS
成果类型：
Article
署名作者：
TOLLEY, HD; NORMAN, JE
署名单位：
National Academies of Sciences, Engineering & Medicine
刊物名称：
BIOMETRIKA
ISSN/ISSBN：
0006-3444
发表日期：
1979
页码：
285291
关键词：
摘要：
A common estimator for comparing mortality rates in epidemiological studies, the time on trial or person years at risk estimator, is based on failure times with an underlying exponential distribution. Since in most follow-up studies individuals are also at risk of dying of unrelated causes, the time on trial estimator is generalized to the competing risk situation. Behavior of the resulting estimator is studied when the failure time of the 2 causes are distributed as dependent exponentials or Weibulls with joint distribution function expressed in the Morgenstern-Farlie form. Certain results of this investigation suggest a more cautious use of this estimator than is presently exhibited.