be a leading cause of inappropriate treatment or misdiagnosis.While precision is periodically checked using internal controls and accuracy is verified by comparison to reference materials or external quality assessment, erroneous results due to interference from endogenous or exogenous substances are subjectively evaluated or overlooked by clinical laboratories. It has long been recognized that hemolysis, bilirubin, and lipids (HIL) are the most common and most significant sources of error in laboratory medicine [2]. Due to their spectral characteristics, these substances can cause optical interference. Moreover, inherent chemicals such as potassium in the cytoplasm of erythrocytes can disrupt the results of measured components.Previously, inspection of individual specimens by the naked eye was routinely applied as the system for detecting and reporting HIL interference. However, visual interpretation of these inter-
Background:The amount of interference due to hemolysis, bilirubin, and lipemia can be measured on the AU5800 autoanalyzer (Beckman Coulter, USA) by spectrophotometry. This is reported as semi-quantitative indices, specifically H-index, I-index, and L-index, respectively. In this study, we evaluated the impact of interference using chemistry assays and established the concentration of interfering substances and HIL-index above which analytically significant interference exists, according to CLSI guidelines C56-A and EP7-A2. Methods: Pooled sera including different concentrations of analytes were prepared and mixed with hemoglobin, bilirubin, or Intralipid. These samples were then tested for 35 clinical chemistry analytes by AU5800 and the bias based on interferent concentrations was computed. The interferent concentration above which significant interference exists was calculated from the 50% within-subject biological variation (desirable analytic goal), and the corresponding index was assigned. Results: Among 35 items evaluated, interference was detected for 12 analytes by hemoglobin, 7 analytes by bilirubin, and 12 analytes by Intralipid. We proposed HIL-index1 and HIL-index2 for each analyte according to 2 different medical decision levels. HIL-index1 and HIL-index2 were considered more reasonable criteria than the HIL-index from the manufacturer's technical document (HIL-indexTD). This is because HIL-indexTD was empirically set to 5% or 10%, and had a wide tolerance range, which was not sufficient to reflect the presence of interference, compared to HILindex1 and HIL-index2. Conclusions: We have demonstrated hemoglobin, bilirubin, and Intralipid interferences according to CLSI guidelines using the desirable analytic goal. Our results provide applicable information for Beckman Coulter automated chemistry analyzers.