The haemoglobin at 20 to 24 weeks, not the prescription, marks the risk

Design

secondary analysis of a multicentre, three-arm, open-label randomised controlled trial

Population

4,252 singleton pregnancies in India with moderate iron deficiency anaemia (haemoglobin 7.0 to 9.9 g/dL) treated at 14 to 17 weeks

Primary outcome

stillbirth, by haemoglobin response measured at 20 to 24 weeks

Effect

relative risk 0.74 per 1 g/dL rise in haemoglobin (95% CI 0.56 to 0.98); risk rose progressively below 10.5 g/dL (P<0.0001)

A secondary analysis of a multicentre randomised trial in India followed 4,252 singleton pregnancies with moderate iron deficiency anaemia, haemoglobin 7.0 to 9.9 g/dL, treated at 14 to 17 weeks. Women had received single-dose intravenous ferric derisomaltose (1,421), single-dose intravenous ferric carboxymaltose (1,424) or oral iron (1,407). The question asked here was not which drug won but what the haemoglobin, ferritin and transferrin saturation at 20 to 24 weeks went on to predict, with adjustment for age, body mass index, parity, treatment arm, baseline haemoglobin and site.

Each 1 g/dL rise in haemoglobin by 20 to 24 weeks was associated with a lower risk of stillbirth, relative risk 0.74 (95% CI 0.56 to 0.98). Fitted as a curve rather than a straight line, risk of stillbirth climbed progressively once haemoglobin sat below 10.5 g/dL (P<0.0001), and so did early preterm birth before 34 weeks (P=0.01). Small-for-gestational-age birth showed a curved relationship with haemoglobin (P=0.008), but the relative risk at lower haemoglobin was not statistically significant.

This reframes where the decision sits. Anaemia in pregnancy is usually treated once and ticked off, and the follow-up haemoglobin is often ordered for completeness rather than read for a decision. On these data the woman whose haemoglobin has barely moved by the mid-second trimester is the one carrying the risk, and she is identifiable weeks before anything goes wrong.

For Indian practice this is unusually direct evidence: the trial was conducted here, in the population and at the haemoglobin thresholds routinely seen in antenatal clinics. A repeat haemoglobin at 20 to 24 weeks is cheap and already within reach of most antenatal schedules.

• Book the repeat haemoglobin at 20 to 24 weeks when you start treatment, not as an afterthought
• Treat a haemoglobin still below 10.5 g/dL at that visit as an unresolved problem, not a partial success
• Check ferritin and transferrin saturation alongside it — a flat haemoglobin on oral iron may be adherence, absorption or ongoing loss
• Record the treatment start date and route, so a non-responder can be told apart from a woman treated last week
• Add growth surveillance for the persistent non-responder, given the associated early preterm birth signal

Why it matters

The prescription is not the endpoint; the haemoglobin four to eight weeks later is where the risk becomes visible.

Don't overread it

This is an observational analysis inside a trial — it shows that a poor response marks risk, not that pushing the haemoglobin higher prevents stillbirth.

The statistics, in plain English

A relative risk of 0.74 with an upper confidence limit of 0.98 clears 1.0 only just — the association is real but the precision is modest, and this is a secondary analysis of outcomes the parent trial was not designed around. The curved models are the more useful part: they place a threshold at about 10.5 g/dL rather than asserting a fixed percentage reduction per gram. The small-for-gestational-age result is the honest counterexample in the same dataset — a significant shape, no significant risk estimate — and it should temper how far the findings are generalised beyond stillbirth and early preterm birth.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.