Interlaboratory variability in drug assay: a comparison of quality control data with reanalysis of routine patient samples. II: Digoxin. Clinical Pharmacology and Toxicology Study Group, Italian Society for Clinical Biochemistry.

Fifty-four laboratories providing a serum digoxin monitoring service participated in a 16-month prospective quality control (QC) study that involved (a) determination of spiked QC samples based on a human serum matrix and (b) reanalysis by two to four reference laboratories of patient samples randomly selected among those assayed routinely. Interlaboratory variability on spiked samples was limited, coefficients of variation being usually in the 11-21% range. Correlations between values reported by individual reference laboratories on routine samples were reasonably good irrespective of the techniques used, which were the enzyme-multiplied immunoassay technique (EMIT) or the fluorescence polarization immunoassay (FPIA). When all data were considered, the correlation between the original routine values and reference results (r = 0.93) was comparable to that observed for the other analytes (phenytoin and theophylline) included in the survey. When subsets of results were evaluated according to the technique used, the agreement between routine and reference results was good for samples originally measured by FPIA (r = 0.96), moderate for samples originally measured by non-EMIT/non-FPIA techniques (r = 0.92), and poor for the few samples originally measured by EMIT (r = 0.55). The poor correlation with EMIT results could be ascribed largely to erratic performance of two individual laboratories. It is concluded that interlaboratory variability in routine drug measurements is greater for digoxin than for other analytes such as phenytoin and theophylline, and that the analytical performance of some laboratories is grossly inaccurate.