Korn E L, Dorey F J
Biometric Research Branch, National Cancer Institute, Bethesda, MD 20892.
Stat Med. 1992 Apr;11(6):813-29. doi: 10.1002/sim.4780110611.
Crude incidence curves display the cumulative number of failures of interest as a function of time. With competing causes of failure, they are distinct from cause-specific incidence curves that treat secondary types of failures as censored observations. After briefly reviewing their definition and estimation, we present five applications of crude incidence curves to show their utility in a broad range of studies. In some of these applications it is helpful to model survival-time distributions with use of two different time metameters, for example, time from diagnosis and age of the patient. We describe how one can incorporate published vital statistics into the models when secondary types of failure correspond to common causes of death.
粗发病率曲线显示了感兴趣的失败累积数量随时间的变化情况。对于存在竞争失败原因的情况,它们与将次要类型的失败视为删失观察值的特定病因发病率曲线不同。在简要回顾其定义和估计方法后,我们给出了粗发病率曲线的五个应用,以展示它们在广泛研究中的效用。在其中一些应用中,使用两种不同的时间度量(例如,从诊断开始的时间和患者年龄)来对生存时间分布进行建模会很有帮助。我们描述了在次要类型的失败对应于常见死亡原因时,如何将已发表的生命统计数据纳入模型。