Zhang Xu, Loberiza Fausto R, Klein John P, Zhang Mei-Jie
Department of Mathematics and Statistics, Georgia State University, Atlanta, GA 30303, USA.
Comput Methods Programs Biomed. 2007 Nov;88(2):95-101. doi: 10.1016/j.cmpb.2007.07.010. Epub 2007 Sep 11.
Often in biomedical research the aim of a study is to compare the outcomes of several treatment arms while adjusting for multiple clinical prognostic factors. In this paper we focus on computation of the direct adjusted survival curves for different treatment groups based on an unstratified or a stratified Cox model. The estimators are constructed by taking the average of the individual predicted survival curves. The method of direct adjustment controls for possible confounders due to an imbalance of patient characteristics between treatment groups. This adjustment is especially useful for non-randomized studies. We have written a SAS macro to estimate and compare the direct adjusted survival curves. We illustrate the SAS macro through the examples analyzing stem cell transplant data and Ewing's sarcoma data.
在生物医学研究中，一项研究的目的通常是在对多个临床预后因素进行调整的同时，比较几个治疗组的结果。在本文中，我们专注于基于未分层或分层的Cox模型计算不同治疗组的直接调整生存曲线。估计量是通过对个体预测生存曲线取平均值来构建的。直接调整方法可控制由于治疗组之间患者特征不平衡而可能产生的混杂因素。这种调整对于非随机研究特别有用。我们编写了一个SAS宏来估计和比较直接调整生存曲线。我们通过分析干细胞移植数据和尤因肉瘤数据的例子来说明这个SAS宏。
