Background: When a new measurement method is introduced into a clinical laboratory, a comparison study is often needed to ensure that the measurement by the existing method is reproducible by the new method with clinical acceptability. The comparison study of ARCHITECT i2000SR analyzer and i-CHROMATM reader analyzer in measuring plasma human chorionic gonadotropin beta subunit (β-HCG) has not been reported.
Methods: Blood samples for β-HCG were collected from pregnant women seen at the outpatient clinic, and they were divided into two groups, those below 20 mU/mL or above, due to its wide concentration range in pregnant women. A comparison study was performed according to EP09-A3 guidelines of the National Clinical and Laboratory Standards Institute (NCCLS). β-HCG’s levels measured from the analyzers being compared were inspected on Bland-Altman plot and outliers were identified by Extreme Studentized Deviate (ESD). Correlation analysis was performed using Passing-Bablok model.
Results: Passing-Bablok regression analysis showed that slope B 95% CIs (confidence intervals) of the two groups fall outside of 1, indicating there was a proportional difference between the two methods. Both groups had a ratio of less than 95% percent of the values in the ± 1.96 RSD (residual standard deviation) interval, indicating that there might be inconsistencies between the two methods with respect to random differences. According to Bland-Altman analysis, 95% Limit of Agreement (LOA) between the two methods exceeded the clinically acceptable limits. The deviation between these two detecting platforms was beyond clinically acceptable ranges when samples fell within the concentrations of 15,000 - 30,000 and 1.2 - 20 mU/mL.
Conclusions: Measurements of β-HCG by ARCHITECT i2000SR and i-CHROMATM Reader are consistent and reproducible only at a certain concentration range. Further research is needed to reduce the biases between these two analyzers.