Search

We assayed 100 capillary whole blood specimens from 50 adults on the Precision QID blood glucose testing system (MediSense, Waltham, Mass) and in duplicate on a whole blood reference method. We centrifuged aliquots of each specimen and analyzed the resulting plasma in the clinical laboratory. The Precision QID achieved maximum total error of 15.5% at a glucose concentration of 42 mg/dL, 12.3% at 83 mg/dL, and 6.4% at 280 mg/dL. The amount of error demonstrated satisfies the performance requirement established by the 1986 Conference on Self-Monitoring of Blood Gluco

In this article we examine the statistics that summarize random error (imprecision) and systematic error (inaccuracy) in the evaluation of blood glucose—monitoring devices. Random error is estimated from replicate testing of patient samples or manufacturers’ controls; systematic error is estimated from the comparison of methods study. Total error combines the measurements of random error and systematic error into a single value and allows comparison of glucose meters’ performance to established standards. Recommendations from the American Diabetes Association Consensus Conference in 1986 specify a 15% maximum performance error. These recommendations are standards currently accepted for evaluations of blood glucose meters.

Well-designed evaluations of blood glucose meters are fundamental to determining the acceptability of meters. Unfortunately, no standard guide exists to evaluate these meters. In this article we outline the meter evaluation process from establishing objectives to generating final recommendations with results that are statistically significant. Proper planning and execution of an evaluation can mitigate or even eliminate problems associated with glycolysis, hematocrit, and blood collection, and thus permit valid inter study comparison.