Weeks 2015.

Study characteristics
Methods	Individually randomised, parallel, double‐blind, placebo controlled, pilot trial
Participants	The study included 748 pregnant women, at 34 or more weeks of gestation, from one regional referral hospital and three large health centres. All women living in the 200 villages that were served by the participating health facilities and had active village health teams were eligible to participate. Women with a known allergy to misoprostol or other prostaglandins, and unanticipated minors aged less than 18 years, were excluded. Women at risk of experiencing a childbirth complication were not excluded. 374 women were recruited into the intervention arm and 374 women were recruited into the control arm. Only 299 out of 748 women recruited into the study had non‐facility births.
Interventions	Women were given a small purse with a string that could be hung around the neck, containing a packet with three foil‐packed tablets (containing either 600 mcg of misoprostol or identical placebo) along with both pictorial and written instructions on how to take the tablets.
Outcomes	Haemoglobin fall of over 20% lower than the prenatal haemoglobin (primary outcome), mean change in haemoglobin, rate of postnatal anaemia, rate of poor maternal and fetal health, appropriate use of study medication, self‐reported side effects, self‐assessed blood loss, quality‐of‐life measures, and safety outcomes (including transfer to hospital, surgical intervention, blood transfusion, maternal death).
Notes	The study was designed as a pilot trial to assess the feasibility of a larger trial, sample size was not formally calculated, and recruitment lasted two months (23 May 2012 to 17 July 2012). The study was funded by the Bill and Melinda Gates Foundation. The authors declared no competing interests.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Treatment packets were consecutively numbered according to a computer‐generated block randomisation sequence.
Allocation concealment (selection bias)	Low risk	This is not clearly described in the report. The following additional information was obtained through correspondence with the study authors: completely identical misoprostol and placebo tablets were professionally prepared. The tablets were then packed in sequentially numbered, labelled manila coin envelopes with unique packet IDs and sealed in purses. The trial packs (in the purses) were distributed in sequential order to the next randomised woman. Participants were instructed not to open the manila envelope until time of birth.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Participants received a small purse containing three foil‐packed tablets (either misoprostol 600 mcg tablets or identical placebo) which were consecutively numbered and identical. Both women and providers were blinded to study group assignment.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	In addition to blinding of providers, the measurement of all haemoglobin outcomes was unlikely to be affected by blinding of assessors.
Incomplete outcome data (attrition bias) All outcomes	High risk	The proportion of women who did not have primary outcome data (haemoglobin outcomes) measured was 35% in the intervention and 46% in the control arms.
Selective reporting (reporting bias)	Low risk	There was no evidence of selective reporting.
Other bias	Unclear risk	There was no predetermined sample size estimation. This pilot study was not followed by a definitive study.

Hb: haemoglobin
PPH: postpartum haemorrhage