Electroimmunoassay of albumin in human serum: accuracy and long-term precision.

An improved procedure for the Laurell "rocket" technique is described. Samples were electrophoresed in an agarose gel containing anti-human albumin. The gel plates were processed, the peaks stained, and peak heights used to calculate albumin concentrations. Factors affecting precision were (a) adequate heating of agarose gel before antibody is added, (b) accurate leveling of the gel surface during plate formation, (c) applied voltage during sample application, and (d) avoidance of the "edge" effect on sample placement in the gel. Multi-plate long-term precision (CV) for the method was 6.2% at a mean albumin concentration of 13 g/L and 3.0% at 37 g/L. Analytical recovery of 8 and 11 g of albumin per liter was 99 to 100%. There was negligible interference from hemoglobin and dextran as well as several common substances that bind to albumin--bilirubin and salicylate. Because of its high accuracy and good long-term precision, the method is a possible candidate reference method for serum albumin.