Assessment of immunochemical methods for determining low concentrations of albumin in urine

Four immunochemical methods (radioimmunoassay, RIA; radial immunodiffusion, RID; immunoturbidimetry, IT; enzyme-linked immunosorbent assay, ELISA) for measuring urinary albumin at low concentrations were assessed for their assay characteristics and practicability. Precision and accuracy were comparable between the methods when studied individually. We made a method comparison, with RIA as reference, using urine samples from diabetic patients with albumin concentrations ranging from 1 to 120 mg/L. There was no significant systematic difference between RID and RIA, but IT and ELISA gave consistently lower values than RIA, the mean differences being 1.8 (p less than 0.01) and 9.7 mg/L (p less than 0.001), respectively. Random error, compared with that for RIA, was in increasing order: RID (residual SD = 3.8 mg/L); IT (4.3 mg/L); ELISA (7.3 mg/L). The difference between the methods increased with the albumin concentration. Operational cost was highest with IT, lowest with RIA. Capital cost was highest with RIA and lowest with RID, which required most technical skill. ELISA had intermediate overall costs.