Comparison of automated Abbott Architect ci 4100 and Abbott Axsym plus analyzers for determination of Phenytoin, Valproic acid, Paracetamol, Vancomycin, Theophylline, Carbamazepine, Phenabarbital, Tricyclic antidepressant and Benzodiazepine in patient’s

Abstract

The purpose of this study was to identify the relationship of the quantitative determination of drugs in serum such as phenytoin, valproic acid, paracetamol, vancomycin, theophylline, carbamazepine, phenobarbital, tricyclic antidepressants and benzodiazepines between automated Abbott Axsym plus analyzers (fluorescence polarization immunoassay) and newly developed automated Abbott Architect ci4100 analyzers (chemiluminescent microparticle immunoassay and enzyme immunoassay) before introduced into a routine determination. This study included 40 samples for each drug which collected from patients for examination in Clinical Toxicology Laboratory Siriraj Hospital. The data were assessed using the linear regression analysis, the percentage of bias, precision and calculated total error. The precision for internal quality control using automated Abbott Architect ci4100 analyzers were 1.22– 7.12 % CV.  The results of phenytoin, valproic acid, vancomycin, theophylline, carbamazepine, phenobarbital, paracetamol and benzodiazepines revealed an excellent correlation (r > 0.993, r² ≥ 0.986). The results of benzodiazepines revealed a good correlation (r > 0.988, r² ≥ 0.976). There were the percentages of bias between -3.52 – 2.51%. The calculated total errors were in acceptable range. However, the result of tricyclic antidepressants revealed low correlation (r > 0.922, r² ≥ 0.85). There was nearly a -40 % bias. The calculated total errors were out of acceptable range. We could conclude that automated Abbott Architect ci 4100 analyzers showed good analytical, it could be introduced into a routine determination of phenytoin, valproic acid, paracetamol, vancomycin, theophylline, carbamazepine, phenobarbital, and benzodiazepines. In tricyclic antidepressants determination, it should be used to predicted equation for patients with long term therapeutic drug monitoring.