实验室间比对结果统计方法的探讨  

Exploration of Statistical Methods of analysis of Inter-laboratory Comparison Results

作　　者：吴婷[1] 杨文凤[1] 张慧文[1] 

机构地区：[1]广州市药品检验所,广州510160

出　　处：《中国药学杂志》2016年第2期135-138,共4页Chinese Pharmaceutical Journal

摘　　要：目的对实验室间比对的结果统计方法进行探讨,提高药品检测能力验证项目组织者的统计能力。方法采用稳健统计法Z比分数方法和"迭代计算"方法,对关键统计技术进行讨论。结果 Z比分数方法和迭代计算方法的稳健平均值分别为77.71%、77.63%,稳健标准差分别为0.66、0.67,变异系数分别为0.85%、0.86%。结论选择正确的统计方法对能力验证结果的判断具有重要的作用。将迭代法得到的最终稳健平均值,和Z比分数得到的中位值进行参考比较,得到真值的最佳估计值范围,对于离群尤其是可疑数据的判断可以更为准确。OBJECTIVE To explore rational statistical methods for analysis of the resuhs of inter-laboratory comparisons, which will be helpful to improve the statistical analysis ability of organizers of proficiency testing project against pharmaceutical laboratories. METHODS Robust statistical Z-score method and iterative computation method were used to analyze the assay result of cefminox sodium powder for injection respectively, and the critical statistical parameters of the two methods were compared. RESULTS The critical statistical parameters of robust statistical Z-score method anf iterative computation method were as follows ： the robust average were 77.71% and 77.63% , the robust standard deviation were 0. 66 and 0. 67, and the coefficients of variation were 0. 85% and 0. 86% , respectively. CONCLUSION It is very important to select the correct statistical approach to evaluate the proficiency of pharmaceutical testing laboratories. By comparing the steady average and the median value of Z-score, the range of best estimates for the true value can be found, and the judgment for the outliers can be more accurate.

关 键 词：实验室间比对 结果统计方法 头孢米诺钠 

分 类 号：R927[医药卫生—药学]