[Structural relations in the comparison of methods in clinical chemistry. Application to the study of bias between a centrifugal analysis (Cobas) and the Kodak-Ektachem method].

In this paper, the authors present the aims of a comparison of 2 biochemical methods in terms of their accuracy. They discuss the inadequacies of classical statistical techniques (paired series, linear correlation coefficient, regression). Instead of them, the authors suggest other approaches based on the structural relationship and, more particularly, York's method which, at the present time, seems to be the best one adapted to the most frequently encountered experimental conditions. They illustrate their view by a study of bias between centrifuge analysis (Cobas) and the Kodak-Ektachem process in plasma calcium assay.