如何正确利用参加者结果对药品检测领域能力验证项目进行合理评价

Determination of assigned values and standard deviations from participant results for performance statistics in proficiency testing schemes

关键词:  能力验证，指定值，能力评定标准差，统计方法选择，能力评定统计量

Key words:  proficiency testing  assigned value  standard deviation for proficiency assessment  statistical method selection  performance statistics

摘要:

目的：帮助能力验证提供者正确利用参加者结果确定能力评定统计量及其中的指定值和能力评定标准差,从而确保能力评定结论的有效性。方法：对国际标准化组织发布的准则、中国合格评定国家认可委员会出台的认可准则及指南性文件、相关部门颁布的中华人民共和国国家标准以及计量技术规范中的技术要求进行逐一梳理,并对其中涉及的统计学相关知识进行拓展介绍和解读。结果与结论：汇总出利用参加者结果确定能力评定统计量这种方法的优缺点。为降低该方法局限性可能对能力评定结果产生的不利影响,本文分别从数据审查、可供选择的常用统计方法及其应用条件、各统计算法的性能指标等关键环节展开了较为全面的分析和解读,并通过具体示例诠释了“没有任何统计方法可以完美适合所有情境”的基本理念,从而提示能力验证提供者在运作计划过程中应高度关注统计学方法的正确选择和使用。

Abstract:

Objective To offer methods for proficiency testing providers to correctly apply the performance statistics that includes assigned value and standard deviations from participant results for the validity of performance evaluation. Method The technical requirements defined in the guidelines issued by the International Organization for Standardization, the accreditation guidelines and guidance documents published by the China National Accreditation Service for Conformity Assessment, the national standards of the People"s Republic of China promulgated by the relevant departments and the technical specifications for metrology are sorted out, and the expanded technique information referring to statistics involved in them is introduced and interpreted. Results and Conclusion The advantages and disadvantages of the determination of criteria for evaluation of performance from data obtained in the same round of a proficiency testing scheme are summarized. For minimizing the limitations of such method, this paper also presents a more comprehensive analysis and interpretation from the essential aspects of data review, selection of available statistical methods with application conditions and the key performance indicators referring to each statistical method. Examples of using different statistic methods are illustrated and these results demonstrate that there is no statistical method that is perfect for all situations. To sum up, this paper intend for proficiency testing providers to highly focus on the selection and use of statistical methods during the implementation of proficiency testing.