Performance of techniques for measurement of therapeutic drugs in serum. A comparison based on external quality assessment data

Ten assay techniques were compared using measurements of a range of 15 drugs spiked in freeze-dried samples of serum reported to the Heathcontrol External Quality Assessment Scheme between November 1988 and January 1991. Three measures of performance were studied: frequency of outliers greater than 3 standard deviations from the sample mean, the coefficient of variation (CV) of sample measurements, and the difference of the sample mean from the spike value. The most consistently precise technique was polarisation fluoroimmunoassay (PFIA). It was in the group of techniques producing significantly fewer outliers and lower CVs than other techniques for all its target analytes. However, a specific interaction with the animal serum used as sample matrix resulted in significant negative bias in PFIA measurements of carbamazepine. Other immunoassay techniques and high-performance liquid chromatography also performed well for a range of analytes, in most cases giving less than 6% of outliers with CVs of less than 13% and less than 5% bias. The least satisfactory techniques were nephelometry and gas-liquid chromatography with derivatisation, which for several analytes gave significantly more outliers and higher CV values than other techniques. In samples containing carbamazepine-10, 11-epoxide, immunoassay measurements of carbamazepine showed cross-reactivity with the epoxide metabolite of between 7 and 15%.