Reliability of point-of-care capillary blood glucose measurements in the critical value range

Context.-Point-of-care glucose (POCG) testing on capillary blood specimens is central to maintaining glycemic control in patients with diabetes. Although there are known performance issues with POCG methods, especially for maintaining tight glucose control, there is little information about the accuracy of results in the critical ranges that may involve life-threatening conditions. Objectives.-To evaluate the reliability of POCG measurements in critical, high (.600 mg/dL) and low (,40 mg/dL) ranges. Design.-One-year retrospective analysis of POCG (ACCU-CHEK glucose meter, Roche Diagnostics Corporation, Indianapolis, Indiana) results for routine patient care were obtained. The frequency and accuracy of repeat testing after critical POCG results was analyzed. A convenience sample of noncritical capillary POCG measurements retested on venous blood specimens by another point-of-care device (RAPIDPoint 405 analyzer, Siemens Medical Solutions USA, Malvern, Pennsylvania) was also evaluated. Results.-Critical values were observed in 2.4 per 1000 POCG tests (256 of 105,928; 0.24%), with the highest rate (76 of 2289; 3.32%) from the emergency department. Twice as many critical high values as critical low values were seen. Nearly 80% of critical POCG tests (204 of 256) were repeated within 10 minutes. Of these 204 repeat measurements, 112 (54.9%) met accuracy criteria (615 mg/dL of low and 620% of high initial values). Accuracy was significantly higher when retesting was performed on the same meter or by the same operator (P < .05). Comparison of capillary and venous POCG testing of noncritical results showed no significant difference (P = .95), with 89.8% (125 of 139) meeting accuracy criteria. Conclusions.-POCG measurements in the critical range are frequently erroneous, which is likely caused by preanalytic factors associated with sampling capillary blood. POCG testing practices should include retesting to confirm critical results.