Background: Blood glucose meters (meters) are routinely used for self-monitoring of blood glucose (SMBG) and for titration of insulin therapy in diabetic pregnancies (DP). We have previously demonstrated that meter accuracy was suboptimal in an antenatal clinic (ANC)1. The Australasian Diabetes in Pregnancy Society (ADIPS) has proposed lower SMBG treatment targets in DP based on the Hyperglycaemia and Adverse Pregnancy Outcomes (HAPO) data2 and recent evidence of glycaemia in normal pregnancy3. This together with updated meter-strip enzyme technology prompted us to re-evaluate the accuracy of meters against a laboratory comparative method.
Method: Finger-prick blood glucose levels (BGLs) were performed on 109 women with DP attending our ANC using 5 different meters (in duplicate) and compared with plasma glucose collected simultaneously on a laboratory analyser. HbA1c and haematocrit (Hct) were also measured.
Results: The plasma glucose range was 2.4-13.6mmol/L, Hct 25-44% and mean HbA1c 5.3 ± 0.5 (SD). The absolute glucose difference [meter – plasma glucose] was 0.370 ± 0.60 to 0.871 ± 0.56mmol/L (mean ± SD). Bias ranged from 8.23-18.37% (see table). There was no significant difference in BGLs for all 5 meters adjusted for Hct. All meters failed to fulfil revised ADA performance goals of analytical error <5%4.
Conclusion: Positive bias in plasma glucose values continues to differentiate meters. Minimisation of these deviations and improved accuracy is essential to prevent over-treatment in DP leading to hypoglycaemia and small-for-gestational-age neonates. This is particularly critical in the setting of the proposed lower treatment targets in DP.