Evaluation of automated glucose oxidase methods for serum glucose: comparison to hexokinase of a colorimetric and an electrometric method.

Two automated glucose oxidase methods have been evaluated with respect to accuracy, precision, recovery, linearity and various potential interferences. Trinder's method on an AutoAnalyzer II had a between-day coefficient of variation (C.V.) of 2.6% (mean 228 mg/dl), was linear to 500 mg/dl, and produced a mean recovery of 99.7%. Comparison of Trinder's method with a manual, blanked hexokinase method yielded the regression equation: TR=--1.95 + HEX (1.04); Spearman's rho correlation coefficient was: 0.974. The Beckman System I glucose method had a between-day C.V. of 1.6% (mean 198 mg/dl), was linear to 500 mg/dl, and recovered an average of 98.0% of added glucose. Comparison with the same hexokinase method yielded: SYI = 1.27 + HEX (1.02: Spearman's rho = 0.991. None of the possible interfering compounds tested caused significant deviation of results by either method within the range of concentrations encountered physiologically. Trinder's method on the AutoAnalyzer II and the System I method are accurate, precise methods and are highly recommended for routine use in the clinical laboratory.