Borenstein M
Hillside Hospital, Division of Long Island Jewish Medical Center, Glen Oaks, New York 11004, USA.
J Clin Epidemiol. 1994 Nov;47(11):1277-85. doi: 10.1016/0895-4356(94)90133-3.
If the purpose of a clinical study is not only to test the null hypothesis but also to estimate the magnitude of the treatment effect, the study design should ensure not only that the study will have adequate power but also that it will enable the researcher to report the relevant parameters with an appropriate level of precision. This paper discusses the factors that control precision in survival studies and shows how a computer program may be used to address these issues. The program allows the user to systematically modify assumptions about the population (e.g. the magnitude of the hazard ratio or the attrition rate) and elements of the study design (e.g. sample size and trial duration), quickly identify the impact of these factors on the study's precision, and modify the study design accordingly. The program may also be used to compute power for a planned study, and confidence intervals for a completed study.
如果一项临床研究的目的不仅是检验零假设，还在于估计治疗效果的大小，那么研究设计不仅应确保研究具有足够的检验效能，还应使研究者能够以适当的精度水平报告相关参数。本文讨论了生存研究中控制精度的因素，并展示了如何使用计算机程序来解决这些问题。该程序允许用户系统地修改关于总体的假设（例如风险比的大小或损耗率）以及研究设计的要素（例如样本量和试验持续时间），快速识别这些因素对研究精度的影响，并相应地修改研究设计。该程序还可用于计算计划研究的检验效能以及已完成研究的置信区间。
