Background: A point-of-care (POC) glucometer (G1) used for critical care at our institution is inaccurate in the presence of low hematocrit (HCT) values. The purpose of this study was to analyze error rates of three additional POC glucometer brands and determine mathematical correction formulas for each.
Methods: Blood samples (n = 196) from a cohort of surgical, trauma, medical, cardiothoracic, and burn intensive care unit patients were tested on three commonly used POC glucometer brands (G2-G4). Results were compared with reference laboratory values, and correction compared with the validated formula for G1. A mathematical formula specific to each glucometer type was derived from glucose measurements, associated HCT values, and the degree of difference relative to laboratory results.
Results: POC glucometer results were consistently elevated compared with reference laboratory values. Glucometer error rates for HCT </= 25% ranged from 15.4% to 22.3% for the three types. Error rates for 25% < HCT < 34% ranged from 16.4% to 18.4%. A correction formula for each glucometer based on the natural log transformation of the HCT predicted reference values with a mean error rate of -0.54% +/- 5.6% for G2, -0.6% +/- 5.5% for G3, and 0.2% +/- 8.0% for G4. Correction was similar to that previously established for G1 (-0.01% +/- 4.8).
Conclusions: Significant error rates because of HCT effect were found in all glucometer models tested with accurate prediction of reference values with a simple mathematical formula.