同源性生存率方差估计值和置信限在癌症生存率研究中的应用

在癌症临床研究中经常用生存率以评价疗效或分析病人预后。然而有人用Greenwood和simon分式,其结果分别严重低估和高估生存率方差,使置信限偏窄和偏宽。相比之下,同源性生存,方差估计分式偏性最小,适于各种生存时间样本,包括无重合,有重合,和分组的,且计算简便。由该式构造生存率置信限可获最佳区间估计,手算即可完成,易于推广应用。附有工作实例描述计算过程。

THE ESTTMATION FOR SURVIVAL RATE OF CANCER PATIENTS THE HOMOGENETIC ESTIMATE FOR THE VARIANCE OF SURVIVAL PATE AND THE CONFIDENCE INTERVAL

The estimation for survival rate of cancer patients is often used for assessing the effect of treatment or analyzing the prognostic factors in clinical work. Greenwood and simon formulae have been commonly employed for this purpose. However, the former grossly underestimates the variance of survival rate and shortens the conitdence intervals and the latter strictly overestimates and lengthens the infervals. By contrast, the formula of the homogenetic estimate for the variance of survival rate has the least bias and is suitable for various survival time samples: with and without ties as well as grouped, with a simple operation. The confidence interval constructed on the basis of this formula can provide the best interval estimation with written calcultion and is easily applied in clinical practice. A worked example fllustratrates the methodology.