Performance evaluation of the Precision PCx point-of-care blood glucose analyzer using discriminant ratio methodology.

Point-of-care glucose testing needs to be integrated into a laboratory information system to provide continuous care. Selecting a particular glucose monitoring system is based on both analytical performance and on user's preference. We evaluated accuracy, performance and regulatory compliance of the Precision PCx glucose analyzer (Abbott), with automatic download into a central station, for remote quality control (QC) management and automatic upgrading. We used the discriminant ratio (DR) methodology, which determines the DR of a test (e.g. the ratio between the underlying SD and the within-subject SD), and compares it to that of another test allegedly measuring the same parameter. Accuracy was evaluated by Clarke's error grid method. Seven hundred and ninety four blood samples were taken from 22 diabetic patients, combined with two capillary blood samples: one for analysis by reference method and the second for PCx analysis. Linear regression analysis revealed, over a 2.1 to 26.9 mmol/l glucose concentration range, a correlation coefficient of 0.963, an intercept of 0.7 mmol/l and a slope of 0.851. Mean difference between meter-generated results and the reference method was -7.1+/-10.8%. Between-run imprecision for PCx using Abbott's controls at low and mid-low concentrations was 5.4 and 3.8%, respectively. Clarke's error grid did not show any clinical impact related to difference between methods. DRs were of similar magnitude using both methods. Nursing staff found PCx easy for everyday use. Our data show that PCx results agree with those obtained with the reference method, as shown by lack of significant difference in DRs, and by excellent correlation.