Practicality and accuracy of prehospital rapid venous blood glucose determination.

ABSTRACT

Blood glucose testing plays an important role in emergency medicine. Although the use of visual reagent test strips is widely established in this setting, the accuracy of reflectometric blood glucose determinations under emergency conditions has rarely been investigated. In a prospective study, 522 of a total of 3,217 patients undergoing emergency blood glucose testing had parallel blood glucose measurements performed using a specific enzymatic method. These 522 patients (aged 61.4 years, 54% men, 90 cases of severe hypoglycemia) had an intravenous access placed at the scene of the emergency. Venous whole blood from the introducer needle of the access was applied to the test strip and the glucose measured with a GlucoTouch reflectometer (LifeScan, Inc.). A blood sample from the intravenous access was then immediately collected in a monovette for subsequent glucose determination in a chemical laboratory (hexokinase method) within 20 to 40 minutes. The emergency glucose measurements (mean 7.3 mmol/L [95% confidence interval [CI] 6.9 to 7.7]; range 0.55 to 27.7) correlated well with the reference laboratory results (Pearson's r = .98; linear regression analysis: slope 1.0, axial intercept 1.74). Error grid analysis also showed good agreement between corresponding measurements zone A 96.7%, B 2.5%, C 0% and D 0.8%. The mean difference using the Bland-Altman method was 0.14 mmoVL; 2 SD 1.8 mmol/L; minimum -7.0 mmol/L; maximum 4.4 mmol/L. The accuracy of the rapid venous blood glucose determination by constantly changing emergency teams was high. Especially in 90 hypoglycemic patients, there were no deviations from the reference method that could have led to clinically relevant wrong decisions. The method of collecting whole blood directly from the venous access is simple and robust, and is independent of the hemodynamic status of the patient.