Hungary 1988.

Methods	Cluster‐randomised controlled trial.
Participants	997 women were enrolled. Setting: 15 pregnancy clinics run by the city of Szeged, the Medical Faculty of Szeged University, and the Obstetrics and Gynaecology ward of the Municipal Hospital of Szeged from 1984 to 1985. Inclusion criteria: women with singleton pregnancies. Exclusions: multiple pregnancies.
Interventions	Treatment group (n = 490; 485 treated) Women received 1 tablet of 5 mmol magnesium aspartate 3 times per day (15 mmol per day). The tablets were chewable. Control group (n = 507; 500 treated) Women received placebo tablets (the content of the tablets were not reported). Treatment began between 6 to 21 weeks and lasted until birth.
Outcomes	Perinatal mortality (antenatal, during delivery, postnatal)/miscarriage; preterm birth (< 285 days); duration of gestation; birthweight percentiles (e.g. < 10th percentile); low birthweight (< 2500 g); EPH gestosis.
Notes	No sample size calculation presented. We were unable to include the data relating to "EPH gestosis" as this was reported for 1321 women in the magnesium group and 1089 in the placebo group according to the translation, which was considered incorrect given that this far exceed the total number of women in the trial.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Randomisation by centre ‐ 8 "magnesium" centres, 7 placebo centres. Method not stated.
Allocation concealment (selection bias)	Unclear risk	Not reported.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	A placebo was used however the composition of placebo not reported. Only the leader of the study knew the composition of the magnesium/placebo tablet bottles – there were identification marks on the tablets known to the leader of the study.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Blinded outcome assessment (as per translations).
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	7/507 (1.4%) women from the placebo group and 5/490 (1.0%) from the magnesium group were excluded initially (in total 12/997 (1.2%) were excluded). 4 were not pregnant, 7 were multiple pregnancies and 1 pregnancy was terminated due to anencephaly. A further 104/500 (20.8%) women were excluded from the placebo group and 85/485 (17.5%) women were excluded from the magnesium group. 189/985 (19.2%) women were excluded in total due to miscarriage, death of the infant prior to discharge or loss to follow‐up. No significant difference was found in exclusions between the magnesium and placebo groups.
Selective reporting (reporting bias)	Unclear risk	Insufficient information to determine (from translation).
Other bias	Unclear risk	High non‐adherence rate. Difficult to determine other bias due to translation.