数据计算表＃2：精密度评价实验 每日一批
分析物/浓度：葡萄糖/高值
仪器或实验室精密度标准差的计算： S2r（源自表1）：7.225 B2（源自表1）：8.88
N（＃每次运行的重复次数）：2
仪器：XYZ
ST?B2??3.53N?121Sr?8.882?(7.225) N2ST自由度的计算：
I＝20
ME＝S2r＝7.225
MD＝N?B2＝17.76
I???N?1??ME?MD???
T＝I?N?1??ME2??MD2I?1＝
2(7.225?17.76)2?7.225?202??17.76?192
＝
624.25
19.21＝32.49 T为32。
NCCLS consensus procedures include an appeals process that is described in detail in Section 8 of the Administrative Procedures. For further information contact the Executive Offices or visit our website at www.nccls.org. Summary of Consensus/Delegate Comments and Committee Responses
EP5-A2: Evaluation of Precision Performance of Quantitative Measurement Methods; Approved Guideline—Second Edition
General
1. A glossary would be useful either as a ―lead-in‖ section or with definitions clearly called out in the text.
? A section on “Definitions” has been added as recommended.
2. The protocols for evaluated precision in both EP5-A and EP5-A2 do not strictly apply to hematology testing, since it is not possible to have the same blood specimen tested for several times within a day for 20 days. We can have protocols to estimate repeatability (as defined in EP5-A2), but most of the time other variance estimates (day-to-day, laboratory-to-laboratory, etc) will be confounded.
? This concern is addressed in the third sentence in the Scope, which states: “These procedures may not be appropriate for some quantitative methods for which adequate test materials do not exist.”
3. I would have preferred to see procedures for the estimation of measurements of uncertainty (ISO).
? The committee agrees that MU (measurements of uncertainty) is an important concept and that EP5-A2 is closely related. This is addressed in the fifth paragraph of the Foreword, which states that the precision estimates from EP5-A2 are components of measurement uncertainty, but that GUM (ISO Guide to the Expression of Uncertainty in Measurement) estimates of MU that comply with ISO procedures are beyond the scope of the document. This is because laboratory-specific estimates of uncertainty may contain components of error other than precision, and may involve corrections for bias.
4. The recommended number of days (20) should be given in terms of ?repeats for a certain factor‘ (or degrees of freedom) to make it more general to any deviation from the base protocol. In addition, a factor that is expected to introduce more variability should be repeated more times than some other that is fairly constant.
? See response to comment number 6, below. See also Appendix C for discussion of modifications.
5. Formulas for calculating precision apply to experiments conducted exactly as described in the protocol. These formulas are totally or partially useless and sometimes misleading when even slight deviations from the protocol (e.g., having three replicates instead of two) are implemented. Currently, there are faster and more elegant ways for obtaining estimates of variance components. I think discussions should be based on the use of commercial statistical software (similar to EP6-A), while these formulas can be put together in an appendix for reference.
? The formulae for unbalanced designs are quite complex and beyond the scope of this document. To be fully general for all situations would reduce the usefulness for less statistically sophisticated users, as explained in the Foreword (third paragraph). While many statistical software packages correctly calculate variance components, not all of them do. Further, these software packages can be difficult to interpret and are best left to the attention of professional statisticians.
Section 7, Statistical Power of Precision Estimates
6. Section 7 does not provide any ?concrete‘ guideline for sample size. The discussion is purely academic.
? It is the intent of the committee to specify a minimum number and guidance for when a situation might require larger numbers. Goal-based experimental design is beyond the scope of the document, but might be considered in a future version.
Section 7.2, Statistical Comparison with the Manufacturer
7. Make degrees of freedom consistent. Use either 100 or 40 for both.
? Section 7.2 presents guidance that is consistent with the general protocol, which allows the user to choose the number of samples to test. In this document, 100 and 40 are presented as different ends of the range of recommended replicates, not as a choice of one or the other.
Section 10.9.1, Repeatability Comparison
8. In the description of equation (5) ―square of the‖ is not needed.
? While the committee agrees that “square of the” is not needed, the committee’s opinion is that the current text is less confusing, since Sis defined as the estimate of repeatability, which is squared in the equation.
r
质量体系方法
NCCLS在标准和指南的发展过程中签署了一种质量体系的方法，以利于项目的管理；它定义了文档的结构以及模版；并且提供了一种用于鉴定所需文档的过程。该方法是基于目前所提供的最新的NCCLS文档HS1－医疗卫生质量体系模型。该质量体系方法运用了一套用于任何组织的核心的“质量体系要点（QSEs）”，适于医疗卫生服务行业的任何分支的运作（换句话说，如何提供一项特定的产品或服务的操作方面）。QSE作为管理者的指南，提供了任何类型的产品或服务的形成的框架。QSEs包括：
文档记录 设备 信息管理 方法的改进 组织 购物清单 事件管理 服务及满意度 个人 方法控制 评估 设施及安全 ××××××××××××××××××××××××××× 服务及满意度 设施及安全 M29 文档记录 购物清单 方法控制 信息管理 事件管理 方法改进 组织 个人 设备 X C24 EP10 EP15 M29
评估
相关的NCCLS出版物
C24－A2 定量测量方法的统计学质量控制：原理和定义；批准指南－第二版（1999）。该文档提供了分析过程的定义；如何设计质量控制过程；以及质量控制应用指南 EP10－A2 临床试验室定量测量方法的初步评价；批准指南－第二版（2002）。该指南阐述了一种分析方法或仪器的性能初步评价的实验设计及数据分析。 用户精密度和准确度性能的验证；批准指南（2001）。 实验室工作人员的职业获得性传染病的防护；批准指南（2001）。基于美国的运作方式， EP15－A M29－A2