Verifica dell’accuratezza di tre glucometri “point-of-care” per l’utilizzo in ambito ospedaliero = Verification of accuracy of 3 glucose point-of-care testing (POCT) devices for their use in a hospital setting

Inaccuracy in glucose POCT can lead to inappropriate therapeutic decisions. In this study, we evaluated the performance of 3 POCT glucometers by comparing results to those by an automated system traceable to higher-order references. Thirty-one heparinized venous blood samples were collected and assayed in duplicate for whole blood glucose concentrations with Roche Accu-Check Compact Plus, Nova Biomedical NovaPro and OK Biotech OKmeterDirect glucometers. All systems were calibrated to report plasma-equivalent results. Samples were then centrifuged and plasma glucose immediately determined by the hexokinase assay on Abbott Architect c16000 platform. The traceability of Abbott assay was checked by comparison with the hexokinase reference procedure performed on 3 samples. POCT performance was evaluated according to Cinical and Laboratory Standards Institute (CLSI) POCT12-A3 criteria [max. 5% of results >+/- 12 mg/dL (for reference results <100 mg/dL) or >+/- 12.5% (for reference results >= 100 mg/dL)] and consensus error grid (CEG) analysis. The Architect assay was perfectly standardized (mean bias, 0.2%). Sample glucose concentrations ranged from 62 to 326 mg/dL, with hematocrit spanning from 0.27 to 0.58 L/L. Average CV on duplicates was 3.5% for Roche, 3.6% for Nova and 5.9% for OK Biotech. All meters gave more than 5% of results (Roche 19.4%, Nova 16.1% and OK Biotech 22.6%) outside the CLSI criteria. However, all results, except two borderline values for OK Biotech, were within the low-risk zone according to CEG. In conclusion, by using CLSI acceptability criteria, the evaluated glucometers were not accurate enough for clinical use in a hospital setting. CEG analysis suggests, however, that this inaccuracy would not have any significant impact on patients' outcome.