Infants born at <34 weeks are at increased risk for cerebral palsy (CP). This risk increases inversely according to gestational age at delivery: approximately 10% at <28 weeks, 6% at 28-29 weeks, and 1.4% at 30-33 weeks. Several retrospective studies evaluated the association between magnesium sulfate (Mag) exposure and the risks of neonatal germinal matrix hemorrhage and CP; some found a beneficial effect whereas others did not. Recently, 4 randomized placebo-controlled trials investigated the benefits of Mag for neuroprotection in women at risk for early preterm delivery. The results of these trials found no differences in the rate of the primary outcome (death/CP) with the use of Mag, but secondary analysis of outcomes in 3 trials found a significant reduction in the rate of neonatal CP among those exposed to Mag. These trials were the subject of several recent metaanalyses and expert opinions with recommendations that have created some uncertainty about whether to offer Mag for neuroprotection and, if offered, about the dose to be used, and the proper gestational age.

When we consider the design of these trials, they were variable in regard to gestational age at randomization (23-32 weeks) and latency period (median of 1.5 hours in 10 days). Two trials used only a loading dose of 4 g of Mag, 1 used 4 g loading and 1 g/h for 24 hours (median 6.5 g), and the other used 6 g loading followed by 2 g/h and allowed retreatment (median dose 31.5 g).

Because the rate of CP increases as gestational age at birth declines, a large number of pregnant women must be exposed to Mag to prevent 1 case of CP at 30-33 weeks compared to those at <28 weeks. If Mag is to be offered to those eligible at <34 weeks, the annual number exposed will be 105,000, whereas it will be 30,000 if offered to <30 weeks, and 12,000 if offered at <28 weeks. Despite the fact that Mag is inexpensive and easy to give, it is associated with high rates of minor side effects, and rare but serious side effects such as cardiorespiratory arrest and death. In addition, its use requires close nursing observation and need for enormous resources in labor and delivery.

Finally, the use of trial sequential analysis to assess the data from the included randomized trials for the effect of antenatal Mag on CP revealed that the apparently conclusive beneficial effect resulting from the recent Cochrane metaanalyses may, in fact, be a false-positive result because of a risk of random error. Thus, it seems that the apparent beneficial effects of Mag on CP remain uncertain.