Methods	RCT with randomisation of individual women.
Participants	Inclusion criteria • Women seeking medical attention due to signs of miscarriage in 1st trimester. • To be included women had to be circulatory stable (stable blood pressure and haemoglobin > 90 g/L) and without any signs of genital infection. Only women with a gestational residue (A-P diameter) between 15 and 50 mm were included. The non-viability of the concepts had to be confirmed and accepted by both the physician and woman. Only women above the age of 18 were included. • Vaginal ultrasound confirmed the miscarriage diagnosis. • N = 126 women. Exclusion criteria • Women who were not able to understand the information provided regarding the study and women with a possible allergy or medical contraindication for analgesics or misoprostol were not included.
Interventions	Intervention: vaginal misoprostol - 400 ug. • 2 tablets of 200 ug, each self administered at home. • N = 64. Comparison: placebo. • tablets identical with the misoprostol tablets. • N = 62.
Outcomes	Complete miscarriage assessed at 6-7 days; infection; bleeding; gastrointestinal side effects; subjective pain; use of analgesics and length of sick leave • Assessed at 7 days. • Successful miscarriage was defined as A-P diameter for the gestational residue was < 15 mm.
Notes	1. Setting: University Hospital, Goteborg, Sweden. 2. Confirmed with the author that the women had incomplete miscarriages diagnosed by ultrasound and there were no IUF. 3. Additional outcomes assessed but not pre-specified in the review: serum haemoglobin; reduction in serum haemoglobin and days of sick leave.
Risk of bias
Bias	Authors’ judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	“…random table system…”
Allocation concealment (selection bias)	Low risk	“Patients were randomised…by drawing a sealed envelope from a box…tablets were delivered to the independent pharmacy where they were inserted by the pharmacy staff into numbered envelopes in blocks of 10…the randomisation list was retained by the hospital pharmacy and was not broken until after completion of the study when statistical analyses were performed”. However, no mention of the envelopes being opaque - so concealment allocation unclear but because tablets are identical, it seems unlikely there is a problem here
Blinding (performance bias and detection bias) All outcomes	Low risk	The placebo tablets “…were identical in appearance to the active misoprostol tablets” and clinicians “…unaware of the randomisation sequence”
Incomplete outcome data (attrition bias) All outcomes	Low risk	There was no loss to follow up and women received their appropriate allocation. the analysis appears to be ITT
Selective reporting (reporting bias)	Unclear risk	Seems to be free of bias here, though the secondary outcome of ‘total number of days of bleeding’ was not reported. However, we did not assess the trial protocol
Other bias	Unclear risk	1. There was an imbalance in baseline data for gestational age: misoprostol: 72.8 (SD 12.2) and placebo 77.8 (SD 12.9). This might favour better outcomes for the placebo group, but probably no important bias here. 2. Women chose whether they wanted a D&C if miscarriage not complete after 1 week or whether to wait longer. So we used the outcome of complete miscarriage at 1 week which excludes problems with choice after that time, but the problem is present for the outcome of surgical evacuation.