Comparison by external quality assessment of performance of analytical systems for measurement of therapeutic drugs in serum.

The precision and accuracy of analytical methods currently in use for therapeutic drug monitoring were evaluated from proficiency test data provided by laboratories participating in the international Healthcontrol external quality assessment scheme for a range of eight antiepileptic drugs, theophylline, caffeine, and digoxin. Different analytical systems were assessed after grouping according to the reagent source and the analyzer used. The majority of analytical methods produced comparable levels of performance with coefficient of variation of < 10% and accuracy within +/-7% of the spike value. Emit reagents were implemented successfully on diverse analyzers but data from the Cobas Mira were generally in the technique group with significantly lower precision. Bias problems were evident for a number of FPIA assays for specific drugs. For example, caffeine interference was present in theophylline measurements by Sigma FPIA reagents whereas use of nonhuman matrix caused a negative bias in Abbott FPIA measurements of carbamazepine. Measurements in the group with highest positive bias were produced by Roche FPIA reagents for phenytoin, phenobarbitone, and carbamazepine. Chromatographic and turbidimetric techniques performed satisfactorily. The variable performance of the different reagent/analyzer combinations demonstrates the value of the narrower technique classification in the assessment of assay performance.