August 13, 1997

In Reply.—

Author Affiliations

Israel Institute of Technology Haifa
Centers for Disease Control and Prevention Atlanta, Ga

JAMA. 1997;278(6):478. doi:10.1001/jama.1997.03550060054034

Much progress has been made in understanding the effect of within-subject biological variation on total variation of a reported laboratory result since the publication on variations and practical utility of blood lipid measurements in 1992.1 Before 1994, it was accepted that, since CVa is usually determined by replicate measurements on a single quality control or reference serum pool, the number of specimens should be considered as a single specimen for calculating CVa in the equation for total CVt. While we were reviewing the effect of multiple serial specimens on CVa, it became evident that the number of multiple specimens does affect CVa, since the average mean would vary with the number of serial specimens on the same subject. The number of serial specimens (NS) modifies the effects of analytical variation CVa on total variation CV, by 1/NS.

The modified equation for total variation