Iatrogenic blood loss is one of the few harms in neonatal intensive care that is entirely under our control, and it drives both anaemia and transfusion exposure. This Cochrane review asked what actually works to reduce it. Five randomised trials with 365 preterm infants qualified, all with mean gestational age under 29 weeks, two conducted in India, two in the USA and one in Israel. Three strategies were tested: taking admission samples from cord or placental blood, using devices that return blood to the infant after analysis, and imposing strict sampling protocols.
Only the first worked. Cord or placental blood sampling likely reduced retinopathy of prematurity requiring treatment (risk ratio 0.52, 95% CI 0.32 to 0.85, moderate-certainty evidence, two studies, 152 infants) and may have reduced severe intraventricular haemorrhage in the first week (RR 0.35, 0.14 to 0.89, low certainty). Bronchopulmonary dysplasia trended the same way without reaching significance (RR 0.78, 0.58 to 1.06). Mortality was very uncertain (RR 0.65, 0.22 to 1.89). Heterogeneity was zero throughout, so the trials agreed with each other.
The other two strategies produced nothing usable. Blood-return devices gave risk ratios of 1.17 for severe intraventricular haemorrhage, 1.14 for bronchopulmonary dysplasia and 1.49 for retinopathy needing treatment, all very uncertain and all pointing the wrong way. Strict sampling protocols were equally uninformative. That is a useful corrective: the intuitive interventions did not deliver, while the mundane one did.
What makes this immediately actionable is that cord sampling costs nothing and needs no equipment. The placenta is already delivered, the blood in it is already the baby's, and the admission panel — blood group, haemoglobin, culture, glucose — can come from there instead of from a heel or a line in a 900 gram infant. Two of the five trials were Indian, which is unusual in neonatal evidence and means the practice has been shown workable in units resembling ours. The barrier is coordination between the obstetric and neonatal teams at the moment of delivery, not cost or technology. Set it up as a standing protocol rather than a case-by-case decision, or it will not happen.
The honest limits: 365 infants across five small trials, no study reported cerebral palsy, blindness or deafness at 18 to 24 months, and certainty ranges from moderate down to very low depending on the outcome. This is a reason to change a practice that has no downside, not a reason to claim the neurodevelopmental question is settled.
- Draw admission bloods from cord or placental blood rather than from the infant wherever delivery allows.
- Agree the protocol with the obstetric team in advance; this fails as an ad hoc decision at delivery.
- Do not invest in blood-return devices on this evidence — the trials showed no benefit.
- Strict sampling protocols alone did not change outcomes; the gain came from moving the source, not rationing tests.
- No trial reported neurodevelopmental outcomes at 18 to 24 months, so the long-term case is untested.
The statistics, in plain English
GRADE certainty ratings matter more here than the risk ratios. Moderate certainty for the retinopathy result means the true effect is probably close to what was measured; low certainty for intraventricular haemorrhage means it may be substantially different; very low certainty for mortality means the estimate should not be relied on at all. The intervals reflect the tiny sample sizes — 0.32 to 0.85 for retinopathy is compatible with anything from a two-thirds reduction to a modest one. The zero heterogeneity across the contributing trials is genuinely reassuring: they were small but they agreed.
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.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for paediatrics, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free