Bujold 2003a.

Methods	RCT.
Participants	Inclusion criteria: women with singleton breech pregnancy at 36 to 40 weeks' gestation Women also given NST and US evaluation for EFW, fetal morphologic features, AFI and placental location. After NST, if women met criteria, clinician verified breech mobility by abdominal palpation. N = 99. Exclusion criteria: IUGR (defined as an EFW (determined by US examination, < 10th percentile for GA), oligohydramnios (defined as AFI ≤ 5 cm), presence of a placenta previa or an abruptio placenta, a previous uterine scar other than a low transverse caesarean delivery, active labour, rupture of membranes, fetal anomalies incompatible with life, a non‐mobile breech by abdominal palpation, any contraindication to vaginal delivery, a medical/allergic contraindication to nitroglycerine.
Interventions	Intervention:tocolysis: nitroglycerine ‐ nitric oxide donor (A3) ‐ sublingual. 2 sublingual sprays of 400 micrograms nitroglycerine given 3 minutes before ECV. N = 50. Comparison:placebo: 2 sublingual sprays of placebo given 3 minutes before ECV. N = 49.
Outcomes	ECV success (at end of procedure); vertex presentation at labour and at birth; vaginal birth; CS; headache; blood pressure; maternal tachycardia; birthweight.
Notes	Sainte‐Justine Hospital, April 1999 to August 2002.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computerised randomisation table.
Allocation concealment (selection bias)	Low risk	Placebo‐controlled trial with identical preparations.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Participants, clinician and assessor were blinded. Intravenous ritodrine and placebo were supplied in identical form; sublingual nitroglycerine and placebo were also supplied in identical form by the hospital pharmacy.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Participants, clinician and assessor were blinded. Intravenous ritodrine and placebo were supplied in identical form; sublingual nitroglycerine and placebo were also supplied in identical form by the hospital pharmacy.
Incomplete outcome data (attrition bias) All outcomes	Low risk	No indication of loss of participants. Not mentioned whether the analysis was intention‐to‐treat; appears that a total of 99 women were randomly assigned and all completed.
Selective reporting (reporting bias)	Unclear risk	We did not assess the trial protocol, but the trial authors said they would compare the rate of vertex presentation at time of birth and the rate of vertex vaginal birth. They reported only that there was no difference and have not provided data. Other outcomes may also be left out.
Other bias	High risk	Halfway through the trial, an interim analysis was performed by the data safety monitoring board. This board decided to stop the trial because of a statistically significant (P value < 0.01) higher rate of side effects and a trend toward a lower rate of successful ECV in 1 group. This decision was based on the likelihood (< 1%) that that group would ultimately show a significant increase in the success rate of ECV, and the likelihood (> 95%) that the subsequently randomly assigned women would be exposed to increased risk of adverse outcomes without potential benefit if the trial was completed. Investigators were informed, and the trial was stopped. No statistically significant differences were observed between the 2 groups with regard to maternal age, GA, EFW, AFI, placental location and type of breech.