最小化随机分组方法介绍及其SAS实现

摘要

目的 介绍最小化随机分组方法的基本原理及运算过程,编制专用的SAS宏程序. 方法 通过文献查阅,综述最小化法的基本原理及其运算过程.依照运算过程,结合模拟实例,编制专用的SAS宏程序,并给出模拟实例的分组结果. 结果 最小化法作为一种动态随机分组方法,依据已入组病例重要预后因素组间分布情况,动态确定新入组病例的分配概率,最大限度保障重要预后因素组间分布均衡.它根据因素不平衡函数、总体不平衡函数和最优分配概率三个参数确定病例的分组,实施过程相对繁复是限制其应用的一个重要方面.通过编制SAS宏程序,可提高最小化法的分组效率. 结论 最小化法适用于小样本、基线特征复杂的临床试验,SAS宏程序能简化实施过程,有利于最小化法的应用.%Objective To introduce the basic principle and operation process of minimization,and compile a customized SAS macro program.Methods Through literature review,the basic principle and operation process of minimization were summarized.By using a simulation example,we introduced a customized SAS macro program of minimization,and the results of allocation for the simulation example were provided.Results As a dynamic randomization method,minimization assigned the next patient into treatment groups with dynamic probability which depended on the balance of some prognostic factors,in order to achieve well-balanced groups.The assignment depended on three functions:the factor level imbalance function,the total imbalance function,and the optimal allocation probabilities.Although the method of minimization has advantages compared with other methods of randomization,its application has been limited probably because of the relatively complicated implementation procedure.As demonstrated by the simulated example,the utilization of minimization could be simplified by using the customized SAS macro.Conclusion Minimization was a method suitable for clinical trials with small sample and complex baseline characteristics.The SAS macro program could simplify the implementation process and facilitate the application of minimization.