- Design
- updated Cochrane systematic review and meta-analysis of randomised and cluster-randomised trials with GRADE assessment
- Population
- six trials, 5,917 women at risk of preterm birth before 34 weeks and their 6,759 fetuses, all in high-income countries
- Primary outcome
- cerebral palsy, and death or cerebral palsy, up to two years' corrected age
- Effect
- cerebral palsy RR 0.71 (95% CI 0.57-0.89, NNT 60); death or cerebral palsy 0.87 (0.77-0.98, NNT 56); death 0.96 (0.82-1.13); maternal adverse effects stopping treatment average RR 3.21 (1.88-5.48)
Antenatal magnesium sulphate before preterm birth has been recommended internationally since the 2009 Cochrane review. This update adds new trials and longer follow-up: six randomised trials, 5,917 women and 6,759 fetuses alive at randomisation, all in high-income countries, comparing magnesium sulphate with placebo below 34 weeks.
Up to two years' corrected age the benefit holds and is graded high-certainty. Cerebral palsy fell (RR 0.71, 95% CI 0.57-0.89), with a number needed to treat of 60 (41-158), as did death or cerebral palsy (0.87, 0.77-0.98; NNT 56). Death alone was unchanged (0.96, 0.82-1.13), as were major neurodevelopmental disability and severe intraventricular haemorrhage.
Two findings deserve more attention than they usually get. At early school age, no difference was detectable in death, cerebral palsy, or death or major neurodevelopmental disability — all on low-certainty evidence from one or two trials, which means the follow-up is too thin to conclude the benefit disappears, but also that it has not been shown to persist. And magnesium sulphate probably tripled maternal adverse effects severe enough to stop the infusion (average RR 3.21, 1.88-5.48), while severe maternal outcomes were not increased. Continue to give it — the two-year evidence is as good as obstetric evidence gets — and tell women that the flushing, nausea and malaise are common, expected and the commonest reason treatment is abandoned.
- Continue magnesium sulphate for neuroprotection below 34 weeks; the two-year evidence is high-certainty
- Number needed to treat is 60 for cerebral palsy — say that plainly when consenting
- Warn about flushing, nausea and malaise: side effects severe enough to stop treatment were three times commoner
- School-age data are too sparse to say whether the benefit persists; do not claim that it does
- All six trials were in high-income countries; regimens varied, and generalisability to Indian practice is untested
The statistics, in plain English
A risk ratio of 0.71 means about 29% fewer children with cerebral palsy in relative terms; the number needed to treat of 60 translates that into how many women must be treated for one child to benefit, and its interval (41 to 158) is the honest range. The school-age results are graded low or very low certainty because they come from one or two trials with few children — an interval like 0.69 to 1.41 for cerebral palsy is compatible with a substantial benefit and with a modest harm, so it is uninformative rather than negative. The maternal side-effect estimate is an average risk ratio across trials with differing regimens, so the exact figure depends on the dose used.
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 obstetrics & gynaecology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free