Methods	RCT with randomisation of individual women in ratio of 3:1.
Participants	Inclusion criteria Women with incomplete miscarriage and intrauterine fetal death. Women with anembryonic gestation or embryonic or fetal death, also women with incomplete or inevitable miscarriage, were enrolled in the trial after assessment using ultrasound. N = 652 total; ICM - N = 39; IUFD - N = 613. Exclusion criteria Women with anaemia (< 9.5 g/dl); hemodynamic instability; history of clotting disorder; using anticoagulants (not including aspirin); allergic to prostaglandins or non-steroidal antiinflammatory drugs; previously undergone surgical or medical abortion either self induced or induced by other physicians during this pregnancy.
Interventions	Intervention: vaginal misoprostol. 800 ug (4 x 200 ug). N = 30. Comparison: surgery. Vacuum aspiration. N = 9.
Outcomes	Success, Hb, fever, nausea, vomiting, diarrhoea and acceptability Complete miscarriage assessed at 3 days and 8 days using transvaginal ultrasound.
Notes	Setting: 4 university settings in US: Columbia University; University of Miami; University of Pennsylvania; University of Pittsburgh. Authors sent us data which separated the outcomes for women with incomplete miscarriage and those with intrauterine fetal deaths. Additional outcomes assessed but not pre-specified in the review: pain; hospital admission; fever. 1 additional paper (Harwood 2008) compared women’s assessment of quality of life between vaginal misoprostol and surgery. It was reported in the Harwood 2008 publication on this study that despite reporting greater pain and lower acceptability of treatment related symptoms, quality of life and treatment acceptability were similar for medical and surgical treatments. Here women with incomplete miscarriage and intrauterine deaths were assessed together.
Risk of bias
Bias	Authors’ judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	“…centralised, computer-automated telephone response system…” used to ran-domly assign women to groups in a 3:1 ratio
Allocation concealment (selection bias)	Low risk	A centralised, computer-automated telephone response system. It was considered that because it was an automated computer response, then allocation concealment would be good
Blinding (performance bias and detection bias) All outcomes	High risk	Not possible to blind either women or clinicians.
Incomplete outcome data (attrition bias) All outcomes	Low risk	1 woman lost to follow up - surgical group.
Selective reporting (reporting bias)	Unclear risk	Assessed all the pre-specified outcomes from the paper, but the trial protocol was not assessed
Other bias	Unclear risk	No significant difference in baseline data reported on the criteria assessed, but difficult to say anything about all other types of bias

BP: blood pressure

ERPC: evacuation of retained products of conception

GA: general anaesthetic

Hb: haemoglobin

ICM: incomplete miscarriage

ITT: intention to treat

IUFD: intrauterine fetal death

MVA: manual vacuum aspiration

RCT: randomised controlled trial

SD: standard deviation

VAS: visual analogue scale

vs: versus