Table Drug

and determined the drug content of the two- and fourfold diluted samples. The results were in proportion to the dilu-tions. However, when the same pa-tients ’ samples were assayed by the automated dilution protocol (page 9.27 of the TDx System Operation Manual) we found that only for a twofold dilu-tion was the result correct, the error occasionally being significant for dilu-tions greater than twofold. Comparison with TDM-AACC target values: The accuracy of the TDx sys-tem was also evaluated by assaying four TDM-AACC quality-control speci-mens for all five drugs during May-September, 1984. The TDx results were all within the acceptable ranges. The linear regression analysis showed good correlations: y(TDx) = 1.lOx(TDM)- 3.92 mol/L; r = 0.9994; S. = 19.8; n = 20. Mean for all five drugs: TDx = 188 mol/L; TDM = 174 LmoI/L. Correlation with HPLC method: We compared results for 331 patients ’ sam-ples as measured by both methods [val-proic acid was assayed by gas-liquid chromatography (4)]. The correlation coefficients (r) were: 0.9818 (n = 81) for phenobarbital, 0.9893 (n = 118) for phenytoin, 0.9875 (n = 38) for primi-done, 0.9838 (n = 64) for carbamaze-pine, and 0.9339 (n = 30) for valproic acid. The good correlation between re-sults by the two methods agrees well with data quoted in the TDx manual. We find the Abbott TDx system easy to operate, and the routine start-up from assay to assay is quick and sim-ple. The good stability of the calibra-tion curve is definitely an advantage over other methods. Precision and ac-curacy of the TDx system meet the supplier’s claims. Assaying the sam-ples by manual dilution, we found the linearity of the assay to be acceptable, but we urge that the automated dilu-tion protocol not be used with more than twofold dilutions. The expert assistance of Dr. H. Blu-menthal on HPLC and a generous supply o