Immunochemical methods for measurement of transferrin in serum: effects of analytical errors and inappropriate reference intervals on diagnostic utility.

We estimated analytical errors of the Calbiochem, Kallestad, Hyland, Meloy, Helena, and Beckman immunochemical methods for serum transferrin. Intermethod biases were determined by analysis of the "Reference Preparation for Serum Proteins" of the College of American Pathologists and by analysis of 106 patients' serum samples. We judged the acceptability of errors by comparing confidence limits for total errors with 1/4 of the normal reference intervals. The transferrin status of each patient's sample was interpreted by comparing the result of each method with the normal reference interval claimed by the corresponding manufacturer. We found that the combined effects of medically unacceptable analytical errors and inappropriate normal intervals caused results of the tested methods not to be interchangeable. The Calbiochem method identified 61 serum samples (57%) as having abnormally high transferrin concentrations. In contrast, for the same specimen, with the Meloy method we found abnormally high transferrin concentrations for only two samples (1.8%).