Accuracy of Devices Used for Self-Monitoring of Blood Glucose

We measured the inaccuracy of 17 home blood glucose monitors (two visually read, eight colorimetric and seven amperometric). Using strips from a single batch, blood glucose measurements were performed by three medical laboratory technologists on at least 50 capillary blood specimens from patients attending two diabetes clinics. Additional capillary blood was deproteinized and assayed with hexokinase to give a whole blood glucose result. A dedicated glucose analyser was also studied to cross-validate the methodology. At a mean glucose concentration of about 9 mmol/L, monitor readings differed from the reference results by −5.1 to + 19.5% with three systems failing to meet the American Diabetes Association guideline for total error of less than 15%. This problem would be alleviated by manufacturers adopting a common policy on calibration and on reporting as a plasma or whole blood value.