Comparison of Blood Glucose Measurement Methods at Low Concentrations in Neonates at Risk for Hypoglycemia.

Background

Accurate assessment of low-level blood glucose is essential for managing neonatal hypoglycemia. This study compared low-level neonatal glucose measurements across commonly used platforms and evaluated the potential clinical implications of method-specific differences.

Methods

Paired neonatal glucose were measured using the Accu-Chek Inform II blood glucose monitoring system (BGMS) device and the Radiometer ABL90 Flex Plus blood gas analyzer. A subset of paired samples was also compared with the Beckman Coulter AU5800 hexokinase method. Method agreement was assessed using Deming regression and Bland-Altman analysis. Exploratory analyses evaluated impact of clinical characteristics, including gestational age, birth weight, growth status, and delivery mode.

Results

Agreement between Accu-Chek and ABL90 was modest (R = 0.558) with wide limits of agreement (-13.21 to 12.95 mg/dL; -0.73 to 0.72 mmol/L) and a bidirectional bias pattern. Positive-bias samples showed a mean bias of 6.13 mg/dL (0.34 mmol/L), whereas negative-bias samples showed -5.29 mg/dL (-0.29 mmol/L). By contrast, Accu-Chek demonstrated strong agreement with the AU5800 hexokinase method (R = 0.976) and narrower limits of agreement (-2.53 to 6.64 mg/dL; -0.14 to 0.37 mmol/L). No significant associations were identified between absolute bias and neonatal clinical characteristics.

Conclusions

At low glucose concentrations, agreement between the Accu-Chek and ABL90 was variable and bidirectional, while agreement between the BGMS meter and the hexokinase method was strong. The lack of association between analytical bias and neonatal characteristics suggests that platform-specific analytical performance, rather than patient factors, accounts for variability at clinically relevant glucose levels.