Author's reply

EDITOR,—To answer the questions of Sarah Vause and colleagues in reverse order, I can reassure them that the data given in our paper referred only to singleton pregnancies. This information was included in our original submission but was inadvertently omitted when we revised the paper to answer points made by referees. I am grateful for the chance to put the record straight.

The St Mary's maternity information system records the lowest haemoglobin concentration but not the gestation at which the lowest value occurred. The references we quoted in our paper make it clear that most of the expansion in plasma volume (which is reflected in the fall in haemoglobin concentration) has occurred by 20-24 weeks' gestation and there is comparatively little change thereafter. Thus, estimation of haemoglobin concentration at any gestation after 24 weeks is likely to show a similar relation with birth weight.

Finally, although we referred mainly to the relation of the lowest haemoglobin concentration to birth weight, I confess that we assumed several times that the lowest concentrations showed the largest falls from prepregnancy values. This assumption is consistent with the longitudinal studies of plasma volume in pregnancy that we cited in our paper. I accept that without measuring prepregnancy haemoglobin concentrations it cannot be proved. If the assumption were not true, however, the average fall in haemoglobin concentration would be the same (or greater) with a high pregnancy haemoglobin concentration as with a low (95-105 g/l) concentration. This would lead to the intriguing conclusion that women with higher than average haemoglobin concentrations when not pregnant would have smaller babies than their peers with lower values. I am not aware of any obvious physiological basis for such a hypothesis. To test it would require a large number of women to have their haemoglobin concentrations measured before they became pregnant; such a study would probably present insuperable logistical problems.