Abstract
The allowable limits of error against 36 clinical biochemical analytes to establish the analytical quality specifications were investigated. These quality specifications were based on the analyses of physiological and biological variations with normal data in 9 studies published after 1990. These studies were performed against Japanese healthy volunteers and biochemical data were collected more than 10 weeks.
Desirable quality specifications for imprecision and bias were estimated using the following formulae, which was proposed by Fraser & Harris: Imprecision one-half of the average within-subject biological variation: 1/2CVI. Bias one-quarter of the group (within-plus between-subject) biological variation: 1/4 (CVG2+CVI) 1/2.
The results of quality specification were obtained very similar to those of Fraser's (Fraser CG, 1992) and Ricós' (Ricós C, 1999). These quality specifications are met with international requirement, and it will be possible to accept in laboratory medicine.
The quality specifications for imprecision can be used as the criteria of internal quality control, and those for bias can be also used as criteria of measurement values of control materials and the external quality assessment schemes.
