Comparison of two automated nephelometers.

The intercalibration precision and linearity were determined for two representative analytes, apolipoprotein A1 (apo A1) and immunoglobulin G (IgG), on the Beckman Array and the Behring Nephelometer 100 (BN-100). For two of nine samples analyzed, poor precision was observed for IgG with the Array. The poor precision for these two samples is attributed to large systematic shifts. A statistical non-linearity was observed for apo A1 with the BN-100, but this non-linearity does not exclude use of this assay for clinical diagnosis. Method comparisons were done for 12 analytes: apo A1, apo B, IgG, IgA, IgM, IgE, C3, C4, albumin, C-reactive protein, alpha-1-antitrypsin, and ceruloplasmin. These comparisons showed proportional biases of > 10% for seven of 12 analytes. Furthermore, correlation coefficients were < 0.96 for seven of 12 analytes. We conclude that comparing results obtained from the two different nephelometers is of only limited value for the individual patient.