Accuracy of bedside glucometry in critically ill patients: influence of clinical characteristics and perfusion index

Abstract

Objectives: To determine the accuracy of bedside glucose strip assay on capillary blood and on whole blood and to identify factors predictive of discrepancies with the laboratory method.

Patients and methods: We conducted a prospective 3-month (July 1-September 30, 2003) study in 85 consecutive patients who required blood glucose monitoring. Values obtained with a glucose test strip on capillary blood and on whole blood were compared with those obtained in the laboratory during serial blood sampling (up to 4 samples per patient). The test strip values were considered to disagree significantly with the laboratory values when the difference exceeded 20%. Clinical and biological parameters and the perfusion index, based on percutaneous oxygen saturation monitoring, were recorded when each sample was obtained.

Results: Capillary glucose values conflicted with laboratory reference values in 15% of samples. A low perfusion index was predictive of conflicting values (P=.04). Seven percent of values obtained with glucose strip on whole-blood samples conflicted with laboratory reference values; factors associated with these discrepancies were mean arterial hypotension (P=.007) and generalized mottling (P=.04).

Conclusion: Bedside blood glucose values must be interpreted with care in critically ill patients. A low perfusion index, reflecting peripheral hypoperfusion, is associated with poor glucose strip performance. Bedside measurements in whole blood seem to be most reliable, except in patients with arterial hypotension and generalized mottling.