To investigate the accuracy of a non-invasive device to measure haemoglobin and compare it with the laboratory gold standard in pregnant women, and to assess if the results vary by skin pigmentation.

Women attending pre-operative assessment before a scheduled caesarean section were invited, and informed consent was taken. Haemoglobin (Hb) estimation was performed using a non-invasive point-of-care device (Masimo Pronto pulse co-oximeter). Venous blood sample was collected for full blood count. The skin pigmentation was classified using the Fitzpatrick scale. Bias and 95% limits of agreement were examined for haemoglobin estimation. Accuracy of the pulse co-oximeter device was assessed in the six Fitzpatrick scales of skin tone.

A total of 122 women were recruited. Twelve of these women (9.8%) had their haemoglobin values < 105 g/L. The Pronto device showed a significant positive bias. The mean difference was 23.4 (SD: 11.97) g/L. The 95% limits of agreement were − 0.06 to 46.9 g/L. The difference between the two methods was unrelated to the skin pigmentation. The area under the ROC curve was 0.705 (95% CI: 0.817–0.593, p = 0.013) for the detection of anaemia.

The point-of-care device (Masimo Pronto pulse co-oximeter) overestimates the Hb readings in pregnant women, and a correction is needed to estimate the real Hb value. The accuracy is unaffected by skin pigmentation. The device may be a potentially useful screening tool in the detection of anaemia in pregnancy in limited resource setting, but the 95% limits of agreements are relatively wide.