Vallikkannu 2014.

Methods	RCT 2‐arm (then cross‐over for second attempt; data following cross‐over have not been included in the review).
Participants	Dates of data collection: 18 Jan 2011 to 23 Dec 2012. Setting: University Hospital, Kuala Lumpur, Malaysia. 6000 to 7000 births a year. Inclusion criteria: women scheduled for ECV (≥ 36 weeks' gestation). Scheduled ECV, breech presentation or transverse lie, singleton gestation, gestational age ≥ 36 weeks, intact membranes, non‐anomalous fetus, reassuring fetal status on cardiotocogram. N = 95. Exclusion criteria: regular contractions were present, estimated fetal weight < 2 kg, oligohydramnios (amniotic fluid index < 5 cm), severe hypertension, recent antepartum haemorrhage, uterine scar, related allergy and any potential contraindication to vaginal birth.
Interventions	Experimental intervention: powder. Commercially available baby talcum powder was applied to the woman’s abdomen by the operator. N = 48. 250 mcg terbutaline was administered subcutaneously 5 to 10 minutes before ECV was attempted. In the first round, a maximum of 2 attempts at ECV were permitted. An attempt comprised a continuous manoeuvre typically lasting not longer than 2 to 3 minutes. Fetal presentation and heart rate were then checked by ultrasound. If ECV was unsuccessful but fetal heart rate was normal and the woman was agreeable, a second attempt was made with the same allocated aid. After completion of the first round of a maximum of 2 attempts, the participant was asked to record her ECV‐related pain score, and the operator was asked to provide a satisfaction score with use of the allocated aid, using a 10 point visual numerical rating scale (VNRS ‐ scored from 1 to 10, marked as higher score more desirable result). Following an unsuccessful first round of ECV, if fetal status was reassuring on cardiotocogram (i.e. until at least 2 fetal heart rate accelerations were observed in the context of a normal baseline, baseline variability and absence of decelerations), and both the provider and the woman were willing, a second round of up to 2 ECV attempts was permitted with cross‐over to the opposing aid, i.e. powder to gel, gel to powder. A further terbutaline dose was given for the second round, which was conducted in similar fashion to the first round. Control/Comparison intervention: gel. Ultrasound aqueous gel was applied to the woman’s abdomen by the operator. N = 47. 250 mcg terbutaline was administered subcutaneously 5 to 10 minutes before ECV was attempted. In the first round, a maximum of 2 attempts at ECV were permitted. An attempt comprised a continuous manoeuvre typically lasting not longer than 2 to 3 minutes. Fetal presentation and heart rate were then checked by ultrasound. If ECV was unsuccessful but fetal heart rate was normal and the woman was agreeable, a second attempt was made with the same allocated aid. After completion of the first round of a maximum of 2 attempts, the participant was asked to record her ECV‐related pain score, and the operator was asked to provide a satisfaction score with use of the allocated aid, using a 10 point visual numerical rating scale (VNRS ‐ scored from 1 to 10, marked as higher score more desirable result). Following an unsuccessful first round of ECV, if fetal status was reassuring on cardiotocogram (i.e. until at least 2 fetal heart rate accelerations were observed in the context of a normal baseline, baseline variability and absence of decelerations), and both the provider and the woman were willing, a second round of up to 2 ECV attempts was permitted with cross‐over to the opposing aid, i.e. powder to gel, gel to powder. A further terbutaline dose was given for the second round, which was conducted in similar fashion to the first round.
Outcomes	Self‐reported pain; success of ECV; operator’s VNRS satisfaction score (identical scale to the pain VNRS described above) with the agent used; significant post‐ECV cardiotocogram anomaly; cephalic presentation at birth; caesarean (and indication); neonatal outcomes of Apgar score, umbilical cord arterial blood pH and base deficit and neonatal admission; gestational age at birth; blood loss at birth and birthweight; fetal or neonatal death; neonatal hypoxic‐ischaemic encephalopathy and major abruptio placenta. ECV was considered a success if cephalic presentation was demonstrated on ultrasound immediately after an attempt.
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	“…computer generated randomisation sequence obtained from http://www.random.org...”.
Allocation concealment (selection bias)	Unclear risk	“…randomisation envelopes were prepared by an author (NV who was not involved in recruitment) in a single block of 100…sequential opening of the lowest numbered sealed opaque envelope remaining just before the start of ECV”. 5 envelopes were not accounted for.
Blinding of participants and personnel (performance bias) All outcomes	High risk	“Blinding of providers and patients to the intervention was not attempted as it was considered unachievable.” It was not clear whether staff were using their usual or preferred method (it was stated that talcum powder had mainly been used, although some staff had started to use gel for ECV).
Blinding of outcome assessment (detection bias) All outcomes	High risk	Blinding was not attempted, as it was considered unachievable.
Incomplete outcome data (attrition bias) All outcomes	Low risk	48 randomly assigned to powder and 47 to gel. Recruitment ceased when all 100 numbered envelopes were used. 5 numbered envelopes could not be accounted for (2 allocated to powder and 3 allocated to gel). All participants received powder or gel as allocated for their first round of ECV. Primary analysis was per protocol.
Selective reporting (reporting bias)	Unclear risk	We did not assess the trial protocol.
Other bias	Unclear risk	No baseline imbalances in age; gestation; parity; nulliparous; weight; height; BMI; ethnicity; etc. When possible, we have used the data related to the first attempt only, but the assessment of pain seems to be pooled in the published paper.