Measurement of glucose content in plasma from capillary blood in diagnosis of diabetes mellitus

Abstract

Overall, there is good correlation between glucose values obtained from ear capillary blood and those from peripheral venous plasma, but there are considerable individual differences. Results obtained with these two methods are generally not interchangeable and the converted values should not be used in the diagnosis of diabetes mellitus, because of the risk of misclassification. In Denmark this can affect 20–24 000 persons. The aim of our study was to investigate whether these differences might be less significant if measurements were taken at the plasma phase of capillary blood and expressed directly as capillary plasma results and if finger capillary blood were used instead of ear capillary blood. The Hitachi 717 instrument was used for measurements of glucose concentrations in venous plasma, the Cobas Mira S in capillary whole blood and the Accu‐Chek Inform from Roche in capillary plasma. The conclusions drawn were (1) capillary ear blood glucose concentration correlates well with capillary finger blood concentration and the two sites can be used interchangeably, yielding similar results in the individual patient; (2) sampling variation is almost the same (approx. 0.16 mmol/L) on capillary plasma and capillary whole blood from finger and ear. Sampling variation for venous plasma measured on the Hitachi instrument was 0.13 mmol/L; not significantly better; (3) the analytical imprecision of glucose measurements on capillary plasma (Accu‐Chek Inform) and capillary whole blood (haemolysate method) is almost the same (approx. 2.0%). The analytical imprecision of glucose measurements on venous plasma is 0.9% using a laboratory method and almost twice as high using Accu‐Chek Inform (2.1%); (4) determination of capillary plasma values in the finger did not improve the correlation with venous plasma values. Even though average values were in better concordance, individual differences did not change. For some persons, both ear‐ and finger capillary blood measurements deviate significantly from results on venous plasma, such that they cannot be used for diagnosis of diabetes mellitus; (5) the main factor for good correlation is the sampling site. Results obtained on plasma and whole blood from the same puncture correlate well; (6) neither capillary blood nor capillary plasma correlates with the venous plasma method recommended by the American Diabetes Association. It is concluded that physiologic differences in glucose content in capillary‐ and venous blood prohibit the random use of these two materials in the diagnosis of diabetes.