Comparison of point-of-care and laboratory glucose analysis in critically ill patients.

BACKGROUND Blood for point-of-care analysis of glucose levels is often obtained from different sources (fingerstick, arterial or central venous catheter). OBJECTIVES To examine agreement between point-of-care and laboratory glucose values and to determine effects of hematocrit, serum carbon dioxide, and mean arterial pressure on the accuracy of point-of-care values. METHODS Point-of-care values were compared with laboratory values. In 49 critically ill patients, blood was obtained first from a catheter for laboratory testing and then from the catheter and via fingerstick for point-of-care testing. Bias, precision, and root-mean-square differences were calculated to quantify differences in values between the 2 methods. A t test was used to determine differences in values between each point-of-care blood source and the laboratory value. Multiple regression analysis was used to determine if serum level of carbon dioxide, hematocrit, and/or mean arterial pressure significantly contributed to the difference in bias and precision for the point-of-care blood sources. RESULTS Mean laboratory glucose level was 135 (SEM 5.3, range 58-265) mg/dL. In point-of-care testing, bias +/- precision and root-mean-square differences were 2.1 +/- 12.3 and 12.35, respectively, for fingerstick blood and 0.6 +/- 10.6 and 10.46 for catheter blood. Values for point-of-care and laboratory tests did not differ significantly. For catheter samples, hematocrit and serum carbon dioxide contributed significantly to difference scores between point-of-care and laboratory values (P < .001). CONCLUSIONS Glucose values for point-of-care samples did not differ significantly from laboratory values. For catheter samples, hematocrit and serum carbon dioxide levels accounted for the difference between point-of-care and laboratory glucose values.