Screening for gestational diabetes mellitus with a reflectance photometer: accuracy and precision of the single-operator model.

Abstract

To investigate the accuracy and precision of a recent-generation reflectance photometer, Accu-Chek III, for gestational diabetes mellitus screening with a single operator. Four hundred ten paired capillary and serum glucose samples collected prospectively were evaluated by a reflectance photometer and a glucose oxidase method (Beckman), respectively, during routine gestational diabetes mellitus screening with a single operator. The mean Accu-Chek III result was 12% greater than serum glucose oxidase (115.5 +/- 1.4 and 103.4 +/- 1.3 mg/dL, respectively). A serum glucose level of at least 135 mg/dL by glucose oxidase produced sensitivity and specificity of 0.81 and 0.68, respectively. At 155 mg/dL, the Accu-Chek III sensitivity was 0.81 and specificity 0.74, and the method produced fewer false-positive results (37 of 410 [9.0%] versus 41 of 410 [10%]) than a 135-mg/dL serum cutoff. Adjusting the Accu-Chek III threshold by two times the coefficient of variation resulted in the same number of false positives as the Beckman technique at 135 mg/dL. The receiver operating characteristic curve cutoff for the venous value of 135 mg/dL was an Accu-Chek III value of 141 mg/dL. Linear regression analysis predicted y = 20.6 + 0.919x, r = 0.838. The mean coefficient of variation was 3.9%. The Accu-Chek III produced accurate and precise gestational diabetes mellitus screening results with a single operator. Because capillary blood is a combination of arterial and venous blood, an Accu-Chek III cutoff of 155 mg/dL may be more appropriate for screening, considering the 10-15% higher capillary glucose level in the fed state. An Accu-Chek III cutoff of 155 mg/dL produced a higher specificity than venous serum at 135 mg/dL, with no decrease in sensitivity.