Accuracy of fingerstick glucose determination in patients receiving CPR

Recent demonstration of the dangers of hyperglycemia in the setting of cerebral hypoperfusion has resulted in recommendations to abandon the long-standing practice of empiric administration of dextrose to patients with altered mental status. Patients receiving basic cardiopulmonary resuscitation (CPR) have been identified as being at particular risk from inappropriate administration of dextrose. We conducted a prospective trial to determine whether the blood glucose of 50 patients receiving CPR could be accurately assessed using a portable rapid-reagent device in the emergency department (ED). We analyzed both capillary and venous blood samples in the rapid-reagent system, using laboratory analysis of venous blood to define the patient's true blood glucose level. Capillary blood analysis identified 8 patients as hypoglycemic; only 3 of these patients were truly hypoglycemic, and 2 patients identified as hypoglycemic by fingerstick were actually hyperglycemic on laboratory analysis. The only instance in which a patient was classified differently by venous rapid-reagent and laboratory analysis occurred when a patient assessed as normoglycemic (169 mg/dL) by the laboratory had a venous rapid-reagent glucose reading of 182 mg/dL (hyperglycemic). We recommend that venous (rather than capillary) blood be used to analyze blood glucose in all patients receiving CPR before dextrose administration is considered.