Hauer-Jensen M
Department of Surgery, University of Arkansas for Medical Sciences, Little Rock 72205.
Comput Appl Biosci. 1993 Feb;9(1):45-7. doi: 10.1093/bioinformatics/9.1.45.
Single estimates of sample size for a study may be easily obtained by use of a hand calculator or from published tables. In contrast, performing multiple calculations is a tedious and time-consuming task, which is greatly simplified by a computer program. The computer program presented here assists the investigator in calculating sample size estimates, determining statistical power and creating randomization tables for a study. The program is designed primarily for clinical trials and thus includes some features not found in other software packages performing similar tasks. Sample size calculation and power analysis are performed for dichotomous, continuous (parametric and non-parametric tests) and time-to-failure (exponential distribution and log-rank test) response variables, and for correlation coefficients. Sample size estimates and significance levels may be adjusted for multiple participating centers, non-compliance, interim analyses and.multiple testing. The randomization subroutine generates tables for studies with up to nine treatment arms and with any valid block size. As a Windows application, the program runs in a multitasking environment, allowing switching between programs and easy pasting of results into word-processing documents and other applications. It is very simple to use, with a completely menudriven interface and sufficient built-in help to obviate the use of a manual.
通过使用手持计算器或查阅已发表的表格，可轻松获得研究样本量的单一估计值。相比之下，进行多次计算是一项繁琐且耗时的任务，而计算机程序可极大地简化这项工作。本文介绍的计算机程序可协助研究人员计算样本量估计值、确定统计功效并为研究创建随机化表。该程序主要为临床试验设计，因此具备一些其他执行类似任务的软件包所没有的功能。针对二分变量、连续变量（参数检验和非参数检验）、失效时间（指数分布和对数秩检验）响应变量以及相关系数进行样本量计算和功效分析。样本量估计值和显著性水平可针对多个参与中心、不依从性、期中分析和多重检验进行调整。随机化子程序可为多达九个治疗组且具有任何有效区组大小的研究生成表格。作为一个Windows应用程序，该程序在多任务环境中运行，允许在程序之间切换，并可轻松将结果粘贴到文字处理文档和其他应用程序中。它使用非常简单，具有完全由菜单驱动的界面，并提供了足够的内置帮助，无需使用手册。
