Bedside capillary glucose monitoring in the general hospital

We assessed the accuracy and reliability of capillary glucose monitoring (CGM) for 20 hospitalized patients as an alternative to repeated venipunctures for laboratory blood glucose (LBG) determination. A total of 330 pairs of observations was obtained. Pearson correlation between patient estimates using Chemstrip bG and the laboratory glucose-oxidase method was 0.87. The mean LBG determination was 197 +/- 5 mg/dl compared with 176.4 +/- 4.3 mg/dl for patient CGM estimates. The mean deviation was 17.6 +/- 2.9 mg/dl or 8.9%, which is well within the 20% range that is generally accepted as sufficiently accurate. Certain individual patients may require relatively more instruction and supervision to reliably carry out these measurements. These patients may represent 10-25% of all diabetic patients. Neither age nor years of schooling is a useful index to identify these patients. Therefore, we suggest as a routine that 10 or more CGM be checked against simultaneous LBG to confirm accuracy before relying on bedside CGM estimates alone in managing the hospitalized patient. Patients who can carry out CGM accurately may be managed with CGM alone.