AI 中文总结
针对遗漏协变量导致非比例风险的比例风险模型偏差问题，提出基于模拟的方法恢复目标风险比，经模拟及III期乳腺癌临床试验数据验证，该方法在实际场景中表现良好。
AI 中文摘要
当从比例风险模型中排除一个纳入后可恢复比例风险的遗漏协变量时，估计治疗效应的偏差可能源于两个方面：对遗漏协变量分布的边缘化，以及因遗漏该协变量导致非比例风险而拟合比例风险模型造成的模型误设。该遗漏协变量可能代表不可观测的生物标志物状态、被忽略的分层因素或强连续预后因子。尽管已有理论框架在强假设下被提出以降低模型误设导致的偏差，但这些假设在实际数据应用中往往不切实际。我们提出一种简单的基于模拟的方法，用于恢复目标治疗效应风险比，该风险比定义为包含遗漏协变量的正确设定比例风险模型得出的风险比。在Weibull基线事件时间分布和遗漏协变量单位方差的最小假设下，我们的方法识别出使按治疗组分层的观测生存曲线估计值被最紧密的生存曲线带包围的目标风险比数值。模拟研究表明：（i）通过单变量脆弱项对指数化遗漏协变量建模可在一定程度上帮助恢复目标风险比，但效果不尽人意；（ii）所提方法在涉及一系列真实风险比、Weibull基线参数及随机均匀删失或其混合的实际场景中，无论遗漏协变量的真实分布如何，均能较好地恢复目标风险比。我们采用一项III期乳腺癌临床试验的数据对所提方法进行了示例说明。
英文摘要
When an omitted covariate whose inclusion would balance proportional hazards is excluded from a proportional hazards model, bias in the estimated treatment effect may arise. The omitted covariate may represent an unobservable biomarker status, an overlooked stratification factor, or a strong continuous prognostic factor. In this paper, we propose a simple simulation-based approach for recovering the target hazard ratio for treatment effect, defined as the hazard ratio from the correctly specified proportional hazards model that includes the omitted covariate. Our approach identifies the target hazard ratio value that generates a band of survival curves enclosing the observed survival curve estimates most, under the minimal assumptions of a Weibull baseline event-time distribution, (a mixture of) uniform censoring, and unit variance of the omitted covariate. Simulation studies indicate that the proposed method recovers the target hazard ratio reasonably well under practical scenarios involving a broad range of true hazard ratios, regardless of the true distribution of the omitted covariate. Consistency of the empirical estimates from the proposed procedure to the true values is proved. We illustrate the proposed method using data from a phase III breast cancer clinical trial.
Comments31 pages, 3 tables, 4 figures
