A randomised controlled trial of intravenous magnesium sulphate versus placebo in the management of women with severe pre-eclampsia


  • E. J. Coetzee,

    Principal Specialist , Corresponding author
    1. Department of Obstetrics and Gynaecology, University of Cape Town, Groote Schuur Hospital, South Africa
      Correspondence: Dr E. J. Coetzee, Department of Obstetrics and Gynaecology, Medical School, Anzio Road, Observatory Cape 7925, South Africa.
    Search for more papers by this author
  • J. Dommisse,

    Chief Specialist
    1. Department of Obstetrics and Gynaecology, University of Cape Town, Groote Schuur Hospital, South Africa
    Search for more papers by this author
  • J. Anthony

    Senior Specialist
    1. Department of Obstetrics and Gynaecology, University of Cape Town, Groote Schuur Hospital, South Africa
    Search for more papers by this author

Correspondence: Dr E. J. Coetzee, Department of Obstetrics and Gynaecology, Medical School, Anzio Road, Observatory Cape 7925, South Africa.


Objective To determine whether the administration of prophylactic intravenous magnesium sulphate reduces the occurrence of eclempsia in women with severe pre-eclampsia.

Design Randomised controlled trial.

Setting A tertiary referral obstetric unit.

Population Eight hundred and twenty-two women with severe pre-eclampsia requiring termination of pregnancy by induction of labour or caesarean section.

Methods The women were randomised to receive either placebo (saline) or magnesium sulphate intravenously. The investigators were blinded to the contents of the pre-mixed solutions.

Main outcome measure The occurrence of eclampsia in the two groups.

Results The data of 699 women were evaluated. Fourteen were withdrawn after randomisation. The overall incidence of eclampsia was 1.8%. Of 345 women who received magnesium sulphate, one developed eclampsia (0.3%); in the placebo group, 11/340 women (3.2%) developed eclampsia (relative risk 0.09; 95% confidence interval 0.01–0.69; P= 0.003).

Conclusion The use of intravenous magnesium sulphate in the management of women with severe pre-eclampsia significantly reduced the development of eclampsia.


In 1990 when this study was planned there was still doubt as to the efficacy of magnesium sulphate in the management of eclampsia. Evidence was based on large descriptive studies such as those of Pritchard1 and the small controlled studies comparing magnesium sulphate to phenytoin2 or diazepam3. The Collaborative Eclampsia Trial4 had not yet begun. There were no published studies comparing magnesium sulphate with placebo or to other therapies in the management of severe pre-eclampsia. There were differences of opinion regarding the use of prophylactic magnesium sulphate, or other anticonvulsants, in the management of women with severe pre-eclampsia. Sibai5 advocated the use of magnesium sulphate, even in cases of moderate pre-eclampsia, while Redmad maintained that anticonvulsants were only indicated if eclampsia occurred. Therefore, a prospective, randomised trial to determine whether intravenous magnesium sulphate was more effective than placebo in the prevention of eclampsia in women with severe pre-eclampsia was commenced in October 1991, after approval by our Ethical committee.


Women with severe pre-eclampsia admitted to the Groote Schuur Hospital Maternity Centre Tertiary Care Unit from secondary level hospitals, outlying clinics (midwife obstetric units) and from the antenatal wards of Groote Schuur Hospital were included in the trial. Only women with severe pre-eclampsia were included where a decision to terminate pregnancy had been made. Severe pre-eclampsia was defined as two or more of the following features: a diastolic blood pressure ≥ 110 mmHg, significant proteinuria, and symptoms of imminent eclampsia. Informed consent was obtained. Women younger than 16 years of age were excluded as there were difficulties in obtaining informed consent. Women already receiving magnesium sulphate or other anticonvulsants were excluded.

The women were allocated using sealed opaque envelopes containing a card instructing the use of solution A or solution B. These cards (but not the envelopes) were consecutively numbered. Envelopes were distributed in mixed batches of 20 and these always had equal numbers of A and B. The sterile solutions were prepared by the hospital pharmacy and contained either 4 g of magnesium sulphate in 200 mL saline or 200 mL saline only. The identity of the solutions marked A or B were changed periodically by the pharmacy without the knowledge of the investigators. The identity of the solutions was revealed only on completion of the study.

An ombudsman was appointed to liaise with the pharmacy, to organise the randomisation of the treatments and to supply personally the batches of envelopes. He monitored the incidence of adverse effects and convulsions and could request an interim analysis if he thought this was indicated. The completed data sheets were collected by one of the investigators (E.J.C.) but only collated and analysed on completion of the study. This investigator was also responsible for recording all cases of eclampsia which occurred in the unit during the period of the study.

As many women had to be transferred from outlying clinics, they were given clonazepam (1 mg intravenously) before transfer. Women who were not transferred were therefore also given 1 mg of clonazepam intravenously after randomisation. Women then received 200 mL of the unknown solution by slow intra- venous infusion over 20 minutes. This was followed by 200 mL by slow intravenous infusion over the following four hours, and repeated every four hours until 24 hours after delivery unless the urine output dropped below 30 mL per hour. All women were managed on the assumption that they were receiving 1 g magnesium sulphate hourly.

Serum magnesium levels were not measured, but routine electrolyte, urea, creatinine, urate and haematological tests were performed. The women were monitored in a special care unit with particular attention to blood pressure levels, urine output and the development of symptoms. Where necessary the blood pressure was controlled to diastolic levels between 100 and 110 mmHg by the graduated intravenous infusion of 25 mg dihydralazine in 200 mL saline after plasma volume expansion with 200–300 mL of isotonic solution, as described by Anthony and Johanson7. Once stabilised women not in labour either had their labour induced or were delivered by caesarean section.

The sample size was calculated on the assumption that the incidence of eclampsia would be 2% and that magnesium sulphate would reduce this by 75%. The estimated sample size on planning the study was 1980 women (α= 0.05; 1 –β= 0.80).

Categorical data were analysed using the χ2 test. The relative risk and 95% confidence limits were calculated for the incidence of seizures in the two study groups. Normally distributed continuous variables were analysed with Student's t test.


An interim analysis was carried out when eight women had developed eclampsia, but although all eight women were receiving placebo, statistical significance had not been achieved and the study was continued. The study was discontinued after four years before the originally planned sample size was attained.

On completion of the study 123 random envelopes and data sheets could not be retrieved from the patient records, and it was not possible to determine who had been randomised to the treatment or the placebo groups. A detailed record of all cases of eclampsia in Groote Schuur Hospital during the period of the study showed that no cases of eclampsia occurred in these 123 women; therefore it was decided to exclude them from the study.

On completion of the study data were available for 699 women. Fourteen women were withdrawn from the trial after randomisation (four were delivered before treatment was given; in three instances the trial solution was unavailable; four women had received magnesium sulphate; consent was unobtainable in two women, and one woman was anuric). None of these women developed eclampsia. Women who received magnesium sulphate totalled 345 and those given placebo 340.

The entry criteria before randomisation or treatment are shown in Table 1. Apart from the higher incidence of twin pregnancies in the placebo group there were no significant differences between the two groups. No cases of eclampsia occurred in the women with multiple pregnancies. The mean blood pressure was 173/116 mmHg, the modal proteinuria was 4+ and 58% of the women had symptoms of imminent eclampsia.

Table 1.  Maternal characteristics and birthweight. Values are given as mean (range) [SD] or n.
 Magnesium sulphate (n= 345)Placebo (n= 340)
Maternal age (years)24 (16–43) [5.9]25 (16–42) [5.9]
Parity0 (0–5)0 (0–6)
No. twin pregnancies314
Gestational age (weeks)34.3 (18–42) [4.3]34.8 (21–44) [1.3]
Birthweight (g)2115 (260–4420) [861]2236 (460–5200) [855]
Systolic blood pressure (mmHg)173 (130–246) [21]173 (122–240) [22]
Diastolic blood pressure (mmHg)117 (90–155) [13]116 (90–155) [12]
Modal proteinuria (+)4 (1–4)4 (1–4)
Symptoms of imminent eclampsia190205

Twelve women developed eclampsia (1.8%). Eleven of these received placebo (3.2%) and one magnesium sulphate (0.3%). The relative risk of eclampsia was 11 times less in women receiving magnesium sulphate (relative risk 0.09, 95% confidence limits 0.01–0.69; P= 0.003). The maternal and fetal outcome was good in all twelve women with eclampsia. One woman was inadvertently given 200 mL of the intravenous solution in one hour instead of 200 mL in four hours due to the use of an incorrect administration set. She developed absent tendon reflexes associated with respiratory depression. She was given intravenous calcium gluconate and recovered fully. It was subsequently determined that she received magnesium sulphate.

One woman died. This patient, treated with placebo during the study, presented 10 days after discharge from hospital with signs of pelvic sepsis and a depressed level of consciousness. She developed progressive flaccid paralysis and died in the intensive care unit. Autopsy showed infarction in the region of the right middle cerebral artery, oedema and herniation of the cerebellum.

Table 2 presents recorded outcomes: there were no significant differences in women receiving placebo or magnesium sulphate solution intravenously, apart from the incidence of eclampsia. Half of the women were delivered by caesarean section.

Table 2.  Outcomes. Values are given as n (%).
 Magnesium sulphate (n= 345)Placebo (n= 340)
  1. *Relative risk 0.09, 95% confidence interval 0.01–0.69; P= 0.003.

Convulsions1 (0.3)11 (3.2)*
Maternal death1
Adverse reaction1
Anti-hypertensive therapy244 (71)257 (76)
Caesarean Section173 (50)165 (49)
Live births310 (89)326 (96)
Stillbirths38 (11)28 (8)


This study shows that intravenously administered magnesium sulphate in the dosage used is effective in reducing the incidence of eclampsia in women with severe pre-eclampsia. However, the routine use of magnesium sulphate in all cases of pre-eclampsia is not justified as the incidence of eclampsia is likely to be lower in milder cases than in those with severe disease, and also in view of the adverse effects of magnesium sulphate. In this study there was one adverse effect due to over dosage. However, this is a rare occurrence if the women are carefully monitored. The dosage (1 g per hour) was considerably less than the 2–3 g per hour advocated by Sibai5 but this higher dosage may reduce the safety margin. It is of note that there were no adverse effects attributable to the use of magnesium sulphate in the Collaborative Eclampsia Trial4.

An analysis after the trial had been in progress for four years indicated a significantly reduced incidence of eclampsia in those women treated with magnesium sulphate (P= 0.003). In view of this, and the fact that the recently published Collaborative Eclampsia Trial4 had demonstrated the efficacy of magnesium sulphate, it was decided that it was not in the best interests of our women to continue a placebo controlled trial which could be considered to be unethical in the light of the above evidence. Burrows and Burrows8 stated that as the incidence of eclampsia in their study of women with severe pre-eclampsia was low (1.3%) a placebo controlled trial was ethically justified. However in our study the incidence of eclampsia in the placebo group was 3.2%.

Chien et al.9 carried out an overview of the evidence from randomised trials of magnesium sulphate in the treatment of eclampsia and pre-eclampsia and concluded that while there was strong support for the use of magnesium sulphate in preventing recurrent seizures in eclampsia, larger trials are required to evaluate whether anticonvulsant therapy is necessary in women with pre-eclampsia. In a recent review article Duley and Johan son10 agreed with this view and stated that to estimate reliably the effects of magnesium sulphate in halving the risk of convulsions would require a trial of over 12,000 women. However, the evidence from the Collaborative Eclampsia Trial and this randomised trial may alter obstetricians’ views such that they may consider further placebo controlled trials to be unnecessary. Lucas et al.11 conducted a prospective study comparing magnesium sulphate to phenytoin in the prevention of eclampsia. Although this study involved 2138 women, the inclusion criteria varied from severe to mild disease, with or without proteinuria. In their study no woman receiving magnesium sulphate developed eclampsia, while 10 women randomised to the phenytoin group had convulsions (P= 0.0004). In a recent review article Roberts12 suggested that there may be other beneficial effects of magnesium therapy but concluded that as the seizure rate is very low, a strategy of treating all pregnant hypertensive women with magnesium sulphate would result in substantial overtreatment. Our study did not evaluate other possible benefits of magnesium sulphate therapy.

It should be noted that there were factors in addition to the administration of magnesium sulphate or placebo which may have affected the incidence of eclampsia in women with severe pre-eclampsia. All the women received intensive care in a tertiary hospital, all had received clonazepam before or on admission and anti- hypertensive treatment (dihydralazine) administered after volume expansion where appropriate. This study can therefore only report the risk of eclampsia in women treated in this way and cannot be extrapolated to all women with untreated severe pre-eclampsia.


Eclampsia occurs 11 times less often in women with severe pre-eclampsia treated in a special care tertiary unit when magnesium sulphate is administered (P= 0.003).

Despite the relatively low overall incidence of seizures in women given placebo (3.2%) and their satisfactory outcome in this series, consideration should be given to the administration of intravenous magnesium sulphate to all women with severe pre-eclampsia, together with adequate antihypertensive therapy and early delivery.


The authors would like to thank the obstetric registrars and nursing staff who cared for the women; the ombudsman Dr R. Rush; and the Groote Schuur Hospital Pharmacy who were responsible for the randomisation and the preparation of the solutions.