Search

As part of a tender AP-HP Paris Hospitals, an assessment of the reliability record of five blood glucose monitoring systems (BGMSs) (Optium Xceed (Abbott), Contour TS (Bayer), One Touch Ultra (Lifescan), Stat Strip Xpress (Nova) and Accu Check (Roche) and an evaluation of their sensitivity to changes in hematocrit were conducted in 4 hospitals of Paris. In terms of inaccuracy, all BGMSs have submitted CV repetability under the limits of acceptability. One BGMS (Lifescan) presented a CV of reproducibility outside limit of acceptability (13.1%). The inaccuracy was measured by a comparison method on multiparameter analyser relative to the hexokinase method for two sites, the glucose oxidase for the two others. The coefficients of correlation varied from 0.8405 to 0.9303. However, according to both defined acceptability criteria (absolute value difference between the result acquired on analyzer and those determined with the BGMS), the percentage of results outside acceptability was above 20% for two BGMSs (Abbott and Lifescan). Similarly, a net effect of changes in hematocrit was observed on the results of those two BGMSs. BGMS Nova was the most reliable, because of the correction device for hematocrit and blank substractions owed to interferences. In terms of expertise, BGMSs Nova and Roche have been selected with the best analytical performance and practicability satisfactory. In the future, accreditation with standard NF/EN 22870 requested for point of care testing, will require a close collaboration between biologists and clinicians to establish a system of strict quality control to detect deviations of these BGMSs.

In order to build a continuous quality improvement approach for control of glucose meters in clinical divisions at Necker-Enfants Malades hospital, the analytical performances (precision and accuracy) of 2 glucose meters have been evaluated in our laboratory according to SFBC recommendations. Fifty-six heparinized whole blood specimens from patients and thirty-nine from healthy volunteers were analyzed on each of the two meters and compared to plasma glucose measurement on the Roche Hitachi 917 system. The correlation coefficient was 0.938 for Optium Xceed and 0.911 for One Touch Ultra. However, 14.7% and 18.9% of the results (n = 95) for respectively Optium Xceed and One Touch Ultra were discordant, i.e. higher than a 20% difference compared to reference blood glucose concentrations. Inaccuracy was more important for low glucose concentrations (< 5 mmol/L; 12/14 discrepant samples for Optium Xceed and 16/19 for One Touch Ultra). This data suggests a lack of accuracy, particularly for low glucose concentrations. Capillary blood glucose concentrations must therefore be interpreted with caution concerning the diagnosis of hypoglycemia and treatment of unstable patients. Moreover, quality control of glucose meters (blood glucose determinations concurrently at bedside and in the laboratory) is difficult to perform. It also raises questions about the responsibility of "point-of-care testing", an area still subject to discussion.