Matthews D R, Burton S F, Bown E, Chusney G, Dornan T, Gale E A, McKinnon G, Steemson J
Diabetes Research Laboratories, Radcliffe Infirmary, Oxford, UK.
Diabet Med. 1991 Nov;8(9):875-80. doi: 10.1111/j.1464-5491.1991.tb02128.x.
The aim of the study was to evaluate the precision and accuracy of the ExacTech home blood glucose meter when used with either capillary or venous blood and to compare this with a reference whole blood glucose assay. Non-fasting glucose measurements were used since a validation study showed no capillary-venous differences between fasting and post-prandial states. In a cross-sectional study, blood was taken from 182 patients and measured in duplicate on three batches of strips. Altogether we analysed 1089 readings. The regression of the data from capillary blood samples (meter vs reference method) had a correlation coefficient, of 0.93, and a mean bias of 0.2 mmol l-1. The corrected 90% confidence interval was +/- 1.5 mmol l-1 overall, and +/- 0.9 mmol l-1 for readings under 7.0 mmol l-1. Regression of the data from venous blood samples (meter vs reference method) had a correlation coefficient of 0.93 and a slope of x 1.1. The corrected 90% confidence interval was +/- 1.7 mmol l-1. Thus venous blood may be used even though the meter is calibrated for capillary samples but the value must be corrected by dividing by 1.1. Error-grid analysis showed that day-to-day clinical decisions could be made on the basis of ExacTech readings, although a diagnosis of borderline diabetes may not be possible.
