Cox模型下病例队列设计中两种伪似然推断方法及其应用

Inference and Application of Two Pseudo-Likelihood-Based Methods for Case-Cohort Design under Cox's Model

在许多大型队列研究中,采用节约成本并能提高效率的抽样设计至关重要.基于生存数据的病例队列设计正是这样一种有偏抽样机制.这种抽样设计最大的优点在于:昂贵的关键协变量的采集和测量仅对一个随机抽取的子队列以及子队列之外所有经历了感兴趣事件的病例个体进行.本文研究如何应用Cox模型拟合病例队列研究数据.探讨了两种基于伪似然思想的统计推断方法及其渐近理论.通过模拟研究和实际数据分析研究了病例队列设计相较于传统简单随机抽样设计的高效性,展示了其理论意义和应用价值.

In many large cohort studies,the major cost typically arises from the measurement of primary exposure variables.With a limited budget,the researchers need to find a cost-effective design to address this problem.The case-cohort design is such a biased sampling scheme.The principle idea of this design is to obtain the measurements of expensive exposure variables only on a subset of the entire cohort and all the subjects outside the subcohort who experience the event of interest.In this paper,we study how to fit the Cox model to case-cohort data.We introduce two inferential methodologies'which are based on pseudo-likelihood ideas and research their asymptotic properties.We conduct a series of simulations to assess the finite-sample performance of the proposed estimators and apply the proposed methods to real data analysis to demonstrate the,application value in practice.,