Background
In New Zealand, it is recommended that all pregnant women have a haemoglobin A1c (HbA1c) test performed with their booking antenatal bloods to identify previously unrecognised diabetes. However, screening rates in some groups are low. Use of a point‐of‐care device may improve compliance with screening.

Aim
To assess the accuracy of the COBAS b101 point‐of‐care system referenced against a laboratory method, for measurement of HbA1c levels in pregnant women.

Materials and methods
Convenience sample of 40 obese pregnant women enrolled in a clinical trial. HbA1c was assayed in paired capillary and venous whole blood samples using the COBAS b101 point‐of‐care system and Primus Ultra2 high performance liquid chromatography laboratory analyser, respectively. The accuracy of the point‐of‐care system was assessed by Bland–Altman analysis.

Results
The mean (SD) laboratory HbA1c was 35.9 (2.0) mmol/mol. The COBAS b101 point‐of‐care system, compared with the laboratory reference method, had a small negative bias for HbA1c (−1.0 mmol/mol, 95% CI −2.0 to −0.03, P = 0.03) and relatively wide 95% limits of agreement (−7.2 to 5.1 mmol/mol).

Conclusion
In conclusion, we found that in pregnancy, the COBAS b101 point‐of‐care system has a small negative bias and modest point accuracy for HbA1c. When used to screen for previously unrecognised diabetes in pregnancy, appropriate COBAS b101 HbA1c point‐of‐care HbA1c thresholds for a negative and positive result are 7 mmol/mol below and 5 mmol/mol above the clinical threshold, respectively. Values between these limits should be confirmed by laboratory testing.