[Inaccuracy of glucose meters. Automatic correction for hematocrit variations and the presence of exogenous interfering components].

Seven hospital-based glucose monitoring systems (meters) were evaluated with particular attention to those analytical interferences encountered in intensive care patients. Imprecision differed little between meters and remained altogether within acceptable limits. Inaccuracy, as measured by comparison with a hexokinase method presented with significant differences, yet without exceeding acceptable limits either. All meters but one showed an important bias when hematocrit departed from the reference interval. Two meters would not distinguish maltose from glucose. Three showed an important positive bias in the presence of acetaminophen and four a comparable bias in the presence of ascorbate. Only one meter was unaffected by both such exogenous interferences and hematocrit variations, owing to built-in hematocrit and electrochemical blank measuring devices. This meter also showed narrowest correlation with hexokinase methods. At a time when intensive care patients are being submitted to ever tighter glycemic control, it is desirable and our results show that it is now possible to tighten accordingly the acceptability criteria of glucose meters used to this end.