Five RCTs (n=5,235) were included. Duration of follow-up ranged from 18 to 24 months. All trials were appropriately randomised, used an intention-to-treat analysis and masked the infant examiner to foetal exposure.
There was no difference between magnesium sulphate administered at <32 to 34 weeks gestation and control for the primary outcome of death or cerebral palsy or for death (five RCTs). Prenatal exposure was associated with a significant reduction in the combined outcome of death or moderate-severe cerebral palsy (RR 0.85, 95% CI 0.73 to 0.99; three RCTs), cerebral palsy of any severity (RR 0.70, 95% CI 0.55 to 0.89; five RCTs) and of moderate-severe cerebral palsy (RR 0.60, 95% CI 0.43 to 0.84; three RCTs). Results were similar when the analysis was restricted to women randomised at <30 weeks gestation (three RCTs). Restriction of the analysis to the three studies and one sub-trial that administered magnesium sulphate for neural protection showed a significant reduction in the primary outcome of combined death or cerebral palsy (RR 0.86, 95% CI 0.75 to 0.99), total cerebral palsy (RR 0.71, 95% CI 0.55 to 0.91), death or moderate-severe cerebral palsy (RR 0.85, 95% CI 0.73 to 0.99; three RCTs) and moderate-severe cerebral palsy (RR 0.60, 95% CI 0.43 to 0.84; three RCTs). There was no increase in the risk of death. There was no evidence of statistical heterogeneity for any analyses.
There was no evidence of publication bias for any of the outcomes (p-values ranged from 0.19 to 0.90) and funnel plots appeared symmetrical.