The evaluation of analytical performance of the Precision G point-of-care glucometer

In this study analytical and functional performance of the Precision G "point-of-care" glucometer (MediSense Inc.) was evaluated. Studies were carried-out using capillary blood collected for routine monitoring of glycemia in diabetic patients. Each glucose test measurement with the glucometer was paralleled by the laboratory measurement of glucose on the same blood sample, using the GOD/PAP method. Mean accuracy error in the glucose concentration range of 1.1-33.3 mmol/l calculated for the glucometer vs. the laboratory method amounted to only 0.2%. However, for glucose concentrations below 4.4 mmol/l the mean accuracy error was 3.9%, and for the concentrations above 10.0 mmol/l it was 4.6%. Within-run CV for three concentration levels was 2.76%, 2.89%, and 4.22%, respectively. Linearity of the meter response in samples with glucose concentration ranging from 1.7 mmol/l to 16.7 mmol/l, expressed as the correlation coefficient r, yielded r=0.996 and linear regression equation [y1 = 0.996 y2 - 0.005], where y1 is the measured glucose concentration and y2 is the target glucose concentration calculated in diluted samples. Correlation studies on a set of 114 blood samples collected from patients and assayed by glucometer and by the laboratory method yielded a relationship expressed by the equation: y = 0.84x + 1.13 where y is glucometer read-out and x is glucose concentration obtained by the laboratory method. Passing-Bablok test showed a significant agreement between the glucometer measurements and the reference laboratory results in the studied glucose concentration range. The error grid analysis of series of the paired patient's samples showed that 95% of results were in the clinically acceptable zone A and 1% of results in zone D.