Fang Weiyu, Zhou Jie, Xie Mengqi
School of Mathematics, Capital Normal University, Beijing, 100048, China.
Lifetime Data Anal. 2025 Jan;31(1):187-204. doi: 10.1007/s10985-024-09637-8. Epub 2024 Oct 12.
Recurrent event data with a terminal event arise in follow-up studies. The current literature has primarily focused on the effect of covariates on the recurrent event process using marginal estimating equation approaches or joint modeling approaches via frailties. In this article, we propose a conditional model for recurrent event data with a terminal event, which provides an intuitive interpretation of the effect of the terminal event: at an early time, the rate of recurrent events is nearly independent of the terminal event, but the dependence gets stronger as time goes close to the terminal event time. A two-stage likelihood-based approach is proposed to estimate parameters of interest. Asymptotic properties of the estimators are established. The finite-sample behavior of the proposed method is examined through simulation studies. A real data of colorectal cancer is analyzed by the proposed method for illustration.
在随访研究中会出现带有终末事件的复发事件数据。当前文献主要聚焦于使用边际估计方程方法或通过脆弱性的联合建模方法来研究协变量对复发事件过程的影响。在本文中，我们提出了一种针对带有终末事件的复发事件数据的条件模型，该模型对终末事件的影响提供了直观的解释：在早期，复发事件的发生率几乎与终末事件无关，但随着时间接近终末事件时间，这种相关性会变得更强。我们提出了一种基于两阶段似然的方法来估计感兴趣的参数。建立了估计量的渐近性质。通过模拟研究考察了所提方法的有限样本行为。为作说明，使用所提方法分析了一个结直肠癌的真实数据。
