Crowther 1990.

Methods	Randomisation and allocation concealment: block randomisation with no stratification and use of numbered, opaque, sealed envelopes. Researchers involved in treatment allocation were not involved in making the randomisation schedule. Blinding of outcome assessment: outcomes were assessed by staff who were unaware of participants' allocation to intervention and control group Documentation of exclusion: no loss to follow‐up
Participants	118 women with an uncomplicated twin pregnancy between 28 and 30 weeks' gestation attending a special antenatal clinic for multiple pregnancy were recruited into the study. Women were not eligible for inclusion if they had a cervical suture, hypertension, a caesarean section scar, an antepartum haemorrhage or uncertain gestational age.
Interventions	Women allocated to the intervention group received bed rest in hospital and were encouraged to rest in bed as much as possible, although voluntary ambulation was allowed. Women in the control group were encouraged to continue their normal activities at home. They were admitted to the hospital only if complications arose such as hypertension, preterm labour or prelabour rupture of the membranes. All women received weekly antenatal assessment.
Outcomes	Primary outcomes Gestational age at delivery (using Dubowitz scoring) Preterm birth (< 37 weeks) Small‐for‐gestational age birthweight (< 10th centile of local singleton standards) Birthweight Admission to the NICU and length of stay Secondary outcomes Very preterm birth (< 34 weeks) Neonatal and maternal morbidity
Notes	Chorionicity: not reported Sample‐size calculation: reported, 100 women for a reduction of preterm delivery from 50% to 28% and 200 women for a reduction in the incidence of small‐for‐gestational age from 35% to 17.5%, with 80% power Intention‐to‐treat analyses: performed Compliance: 4 women (7%) in the hospitalisation group did not attend for hospitalisation and 11 women (19%) required leave of absence from the hospital. In the control group, 22 women (37%) subsequently required hospitalisation because of pregnancy complications. Location: single centre in Harare, Zimbabwe Timeframe: 1984‐1986
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Block randomisation with no stratification and use of numbered, opaque, sealed envelopes. p 873
Allocation concealment (selection bias)	Low risk	"..the researchers involved in treatment allocation were not involved in making the randomisation schedule." p 873
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Blinding of participants and personnel was not possible due to the type of intervention (hospitalisation and bed rest compared with no activity restriction at home)
Blinding of outcome assessment (detection bias) All outcomes	Low risk	"The gestational age assessments were made by a neonatologist who was unaware of the allocation treatment group." "The other primary measures of outcomes were measurements or decisions taken by staff who were unaware of the group to which the mother had been allocated." p 873
Incomplete outcome data (attrition bias) All outcomes	Low risk	No loss to follow‐up and analysis on an intention‐to‐treat basis
Selective reporting (reporting bias)	Unclear risk	No information provided about trial protocol or predefined outcomes
Other bias	Low risk	The study appears to be free of other sources of bias