A Stochastic Approximation Method for Assigning Values to Calibrators (Laboratory Management)
2,99 €
Kopen bij Amazon Bekijk in Apple BooksLiever op papier? Alle uitvoeringen bij Amazon →
Prijs in Apple Books. Bij Amazon kan die afwijken.
- Verschenen
Beschrijving
The problem of transferring the analyte concentration value of a reference material to a new production calibrator is a common manufacturing activity in the in vitro medical diagnostic industry. Many popular approaches to this problem, whether they use consensus values from many laboratories, reference laboratories, or master lots, involve fitting calibration curves to assay responses of reference standards to determine the concentration values of the new production calibrators, which are assayed as unknowns (1, 2). There are at least two potential problems with this approach: Any consistent systematic error in fitting the calibration curve is propagated to the value assignment of the production calibrator regardless of how many assay setups (or runs) are used, and the required sample sizes, even with a single instrument, are direct functions of the magnitudes of the random variabilities of fitted calibration curves and/or assay setups together with the within-run variabilities. With consensus value methods, the required sample sizes are further increased by the variabilities among laboratories, analyzers, and assay methods. The object of this paper is to provide a simpler, more economical method of value assignment that is robust to calibration curve-fitting errors and has its required sample size driven only by the smallest component of variability, which is the intraassay signal response SD, rather than by the interassay or interlaboratory sources of variabilities of recovered values, which drive up sample sizes. For the case in which both the production calibrators and the reference materials have the same signal-response curve, which is linear in analyte concentration with a zero intercept, this problem has already been solved by the following well known equation (1):