Data and Statistical Sciences, AbbVie Inc., North Chicago 60064, IL, USA.
Contemp Clin Trials. 2024 Jan;136:107401. doi: 10.1016/j.cct.2023.107401. Epub 2023 Nov 22.
Cox proportional hazards model has been an established model for survival analysis. The flexibility of incorporating time dependent covariates has made the analysis more suitable in many clinical trials when the time dependent covariates may be predictive factors for the events. Subjects are censored for various reasons, but they are usually nonnormatively censored in the analysis. Methods for informative censoring are not well studied for settings with time dependent covariates. In this paper, we propose a few methods for informative censoring in survival analysis by Cox model with time dependent covariates, including tipping point method and Reference Based Imputation (Jump to Reference and Copy Reference). The implementation of these methods by multiple imputation is described and illustrated with two data examples.
Cox 比例风险模型一直是生存分析的一种成熟模型。在许多临床试验中，当时间相关协变量可能是事件的预测因素时，纳入时间相关协变量的灵活性使得分析更加适用。由于各种原因，受试者被删失，但在分析中通常是非正态删失的。对于具有时间相关协变量的情况，针对信息性删失的方法尚未得到很好的研究。本文提出了几种在包含时间相关协变量的 Cox 模型中进行生存分析的信息性删失方法，包括拐点法和基于参考的插补法（Jump to Reference and Copy Reference）。通过多次插补描述了这些方法的实现，并通过两个数据示例进行了说明。
