A clinical comparison of the performance of four blood glucose reagent strips

Abstract

The widespread use of visually read blood glucose reagent strips for initiation of emergent treatment of hypoglycemia and hyperglycemia has produced concern over the accuracy of this method. This study evaluated the accuracy of Chemstrip bG (Bio-Dynamics, Boehringer Mannheim, Indianapolis, IN), Dextrostix (Ames, Miles Laboratories, Elkhart, IN), Glucostix (Ames), and Visidex II (Ames) as compared with hospital laboratory values in an emergency department (ED) setting. Blood samples from 96 ED patients were tested for glucose concentration by each of the four strips and by the hospital laboratory. Each strip was evaluated for sensitivity, specificity, correlation coefficient ( r), 95% confidence intervals, and kappa statistic (κ 1 a measure of agreement between nonparametric data) using laboratory values as reference. In addition, six observers scored each strip for ease of interpretation using an ordinal scale of 1 (poor) to 4 (excellent). From the samples, no patients were hypoglycemic (≤60 mg/dL), 83 were euglycemic (>60 and <160 mg/dL), and 13 were hyperglycemic (≥160 mg/dL). Results suggest that over the range of glucose concentration sampled, there is good to excellent correlation with laboratory values for all strips except Dextrostix. The lower r value for Dextrostix is in part artifact due to limitation of its range of measurement to ≤250 mg/dL. Decreased accuracy for all strips in the hyperglycemic range may have been attributable to small sample size. Chemstrip bG and Visidex II were found to be subjectively easier to interpret. The authors conclude that Chemstrip bG, Visidex II, and Glucostix are suitable for use in an ED setting. Accuracy in the hypoglycemic range warrants clinical study.