Author:
Ichihara Kiyoshi,Kawai Tadashi
Abstract
AbstractIn proficiency testing (PT), it is difficult to set rational criteria for determining target values and allowable limits. In Japan, the largest national PT program has for the past 8years used an evaluation scheme based on a “common coefficient of variation” (cCV) to overcome this difficulty. The cCV, the average within-method variation, is derived by one-way analysis of variance from data of peer groups after excluding those with large variations. A unit interval used for grading participants is obtained by multiplying the cCV by the peer-group mean which is computed using an iterative method featuring repeated removal of a large portion of the tails of the distribution and reinflation of the standard deviation (SD) to compensate for the trimming. The evaluation scheme gives a relatively better grade to those participants in peer groups with small CVs, compared to use of individual peer-group CVs. The scheme has expedited the transition to more reproducible methods, resulting in a notable overall improvement in laboratory performance. The cCV evaluation scheme is uniformly applicable to any analyte measured quantitatively. The iterative method, an integral part of the scheme, effectively precludes bias in the evaluation.
Subject
Biochemistry, medical,Clinical Biochemistry,General Medicine
Reference9 articles.
1. A theoretical basis for clinically relevant proficiency testing evaluation limits;Ross;Arch Pathol Lab Med,1988
2. Standardization of immunoassay for CRM - related proteins in Japan : from evaluating CRM to setting reference intervals;Itoh;Clin Chem Lab Med,2001
3. Proficiency testing as regulatory device : a CAP perspective;Hamilin;Clin Chem,1992
4. Reference values and participant means as targets in proficiency testing;Tholen;Arch Pathol Lab Med,1993
5. Serum protein standardization project in Japan : evaluation of an IFCC reference material and establishment of reference intervals;Itoh;J Clin Lab Anal,1997
Cited by
3 articles.
订阅此论文施引文献
订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献