The aim of this prospective observational study was to compare a bedside test with the reference laboratory method in routine postnatal glucose monitoring. Term newborns with increased risk or clinical signs of hypoglycemia were screened with a bedside test. In case of a glucose value below 2.25 mmol/L, a second blood sample was taken and a duplicate glucose measurement done in the laboratory using a bedside test (Accutrend sensor) and the reference laboratory method (hexokinase method) at the same time and from the same sample. From 110 term newborns, 122 blood samples were obtained for duplicate measurements (median 1.69 mmol/L, SD 0.45 mmol/L). Of these 122, Accutrend correctly identified 97% as being <2.25 mmol/L by the laboratory method. A Bland-Altman plot revealed a mean underestimation of the Accutrend of only -0.09 mmol/L. However, due to high scattering, the maximal over- and underestimation was 0.89 and 1.39 mmol/L, respectively. Only 75% of the results from the Accutrend were within +/-20% of the result of the laboratory method. If the cut-off for low glucose concentrations was set 0.6 mmol/L higher for the bedside test as compared to the laboratory method, all patients except one would have been correctly identified as hypoglycemic.
Conclusion: When using the Accutrend sensor, single infants with even marked hypoglycemia might be missed. Some delay in receiving accurate measurements might be more helpful for clinical decisions and long-term outcome than immediate but potentially misleading results.