Point of care and plasma glucose measurements in the newborn: agreement and precision.

Abstract

Point-of-care (POC) devices for glucose measurement provide rapid results, but the accuracy of the measurement compared with the reference standard (laboratory-based plasma testing) in newborns is unclear. We evaluated the agreement of paired POC (StatStrip glucometer) and plasma glucose measurement using the Bland-Altman method for repeated measures. Limits of agreement (LOA) falling within ±15 mg/dL were determined a priori to be acceptable. We analyzed 4071 paired POC and laboratory-based glucose measurements from 2065 unique patients; (mean ± SD) gestational age of 33.1 ± 5.6 weeks and birth weight of 2216.2 ± 1186.0 g. When accounting for repeated measures, the estimated mean bias was 0.86 mg/dL (95% CI: 0.40, 1.32) with LOA -61.77 to 63.49 mg/dL (95% CI: -62.91, 64.63), beyond the acceptable ±15 mg/dL. If the POC method was used alone, 7.7% of hypoglycemia and 8.2% of hyperglycemia would be undertreated. Conversely, 3.6% of euglycemia would be overtreated for hypoglycemia, and 1% overtreated for hyperglycemia. POC glucose measurements using the StatStrip glucometer showed poor precision with a greater tendency for undertreatment than overtreatment. Clinicians should be cognizant of this while interpreting bedside POC glucose measurements. Glucose point-of-care measurements in a newborn population showed poor precision and should not be considered interchangeable with the plasma measurements.