Accuracy of the hemocue portable glucose analyzer in a large nonhomogeneous population.

Several studies have reported inconsistent results between HemoCue (HC) whole blood glucose measurements compared to plasma glucose. We selected a large patient population with diverse pathologies and healthy volunteers to evaluate HC. For this comparison, whole blood glucose concentration was measured using HC and referenced to laboratory plasma glucose. The population (n = 512) included healthy volunteers, diabetics, and patients with heart failure, liver failure, renal failure, renal and liver transplant, and other chronic diseases. Patients were on a wide variety of medications, vitamins, and food supplements. Venous blood samples were collected in tubes containing potassium oxalate and sodium fluoride. Comparison of the results was made using the method of Bland and Altman and ANOVA at three selected glucose ranges. The glucose measurement ([HC + laboratory]/2) ranges were 24-75, 76-129, and 130-404 mg/dL. A positive bias for all three glucose ranges was observed: 38 +/- 17 mg/dL for the high glucose group compared to 24 +/- 9 mg/dL and 22 +/- 10 mg/dL for the middle and low groups, respectively. In the high glucose group 90% of the values were within 10% (R = 0.97) of the laboratory reference values compared to 81% and 55% in the normal and low glucose groups, respectively. HC glucose measurements were generally within two SD from the laboratory plasma reference. HC consistently yielded lower whole blood glucose measurements than plasma with the largest differences seen in the low glucose range (29%). HC measured more consistently at the higher glucose concentrations and was 16% lower than plasma, although the mean absolute error was highest for that range. No significant effects in the bias could be attributed to disease while possible effects from instrument modifications by the manufacturer remain uncertain.