Nimodipine SG 2003.

Methods	Randomisation stratified by centre, blocks of 6. Sealed opaque envelopes. Recruitment 1995‐2000. 100 women (6%) excluded from analysis: 99 did not get allocated treatment, 1 withdrawn. Recruitment stopped early following interim analysis. CFU ‐ B, blinding ‐ C.
Participants	1750 women with PE, planned delivery and no previous MgSO4. BP >/= 140/90 and 1+ proteinuria plus 1 of: headache, clonus, visual disturbance, epigastric pain, oliguria, pulmonary oedema, raised liver enzymes, haemolysis, oligohydramnios, IUGR.
Interventions	Nimodipine: 60 mg 4‐hourly, orally MgSO4: according to local protocol. Either 4 g IV then 1 g/hr, or 6 g IV then 2 g/hr. All continued either for 24 hr total, or until 24 hr after delivery. Serum monitoring not required.
Outcomes	Woman: eclampsia, stroke, coagulopathy, respiratory problems, cardiac failure, antihypertensive drugs, side‐effects, abruption, caesarean section, PPH. Baby: RDS, hypotonia, intubation, hypotension.
Notes	Recruitment at 14 hospitals in 8 countries. Data for stillbirths and neonatal deaths not reported. These data were requested from the investigators, but have been lost.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	“Patients were randomly assigned according to center (Epistat Services) in blocks of six...” does not refer to random number table or use of a computer number generator.
Allocation concealment (selection bias)	Low risk	“Patients were randomly assigned according to center (Epistat Services) in blocks of six, with the use of sealed opaque envelopes...”
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Described as “The study was not blinded, because of logistic and economic constraints. The primary outcome measure (eclampsia) was binary, objective, and not subject to observer or measurement bias”.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Described as “The study was not blinded, because of logistic and economic constraints. The primary outcome measure (eclampsia) was binary, objective, and not subject to observer or measurement bias”.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Data were available for 1650 of 1750 patients (94.3%) – so minimal loss. 99 patients did not receive the study drug mainly because they gave birth before the drug could be administered or because of logistic issues and 1 patient in the MgSO4 group was withdrawn because induction of labour was stopped and conservative management instituted.   However, no baseline details for these 100 patients – so do not know how similar they were the sample as a whole.
Selective reporting (reporting bias)	Low risk	All expected outcomes reported.
Other bias	Unclear risk	Groups appear well balanced for baseline characteristics, apart from SBP.  Study was stopped early because a planned interim analysis showed a significantly higher rate of seizure in the nimodipine group.  Initially planned 1000 patients per group.