Graham 2014.

Methods	Randomised blinded, parallel arm controlled trial.
Participants	Setting and population: 30 women from a tertiary referral hospital in Australia, with cephalic, singleton pregnancies. 15 women were randomised to each arm (intervention and control). Inclusion criteria: women who had completed 37 weeks of gestation with a cephalic singleton pregnancy, planned a vaginal delivery, had a cervix at full dilatation with an OP position of the fetal head confirmed by a mobile transabdominal ultrasound scan, aged > 16 years old and had given written informed consent. A vaginal examination at full dilatation was performed. A bedside mobile ultrasound scan was performed to confirm fetal head position by the labour ward registrar or consultant. If the position was OP, a study investigator performed either the manual rotation or sham procedure. Fetal OP position was established by obtaining a transverse view of the fetal orbits. OP was defined as fetal occiput within 45 degrees of the midline. At the first urge to push or 1 hour after full dilatation (whichever came first), the study investigator repeated the bedside ultrasound and if the OP position persisted, randomised the participant to either the control or intervention arm of the study. Exclusion criteria: clinical suspicion of cephalo‐pelvic disproportion, a history of previous uterine surgery (caesarean section, open myomectomy), a brow or face presentation, a pathologic CTG according to the RCOG guidelines plus either abnormal baseline or reduced variability for more than 90 minutes, suspected fetal compromise, an anatomical fetal abnormality, suspected or known chorioamnionitis, any condition requiring immediate delivery, any condition requiring an elective caesarean section, an intrapartum haemorrhage > 50 mL, a temperature more than 38.4 °C in the first stage of labour or a suspected fetal bleeding diathesis.
Interventions	Intervention: manual rotation using the digital technique, with the operator's fingers placed along the lambdoid sutures and rotating the posterior fontanelle towards the pubic symphysis during contractions with maternal expulsive efforts over 3 contractions. The fetal occiput was then held in the OA position for a further 1 or 2 contractions. Control: sham manual rotation that involved a vaginal examination over 4 or 5 contractions during which the attending midwife was asked not to observe.
Outcomes	Primary outcomes: operative delivery (defined as caesarean, forceps or vacuum delivery), and the feasibility of running a large trial.  Other outcomes: duration of the second stage of labour, the time between intervention and delivery, degree of perineal trauma, incidence of postpartum haemorrhage (estimated blood loss > 500 mL), cervical laceration, cord prolapse, admission to the NICU, perinatal mortality and serious neonatal morbidity. Neonatal outcomes: birthweight, sex of the infant, 5‐minute Apgar score, cord umbilical artery lactate at delivery, neonatal trauma and admissions to the NICU Study investigator used a repeat mobile transabdominal ultrasound, blinded to the treating clinicians, to confirm and record the final position of the fetal occiput. Digital rotation was considered to be successful if the fetal occiput was OA and within 45 degrees of the midline.
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomisation was performed using computer‐generated random sized blocks with variable block sizes of 4, 6 or 8.
Allocation concealment (selection bias)	Low risk	Women were consented, enrolled and then randomised. Treatment allocations were placed into opaque, sequentially numbered, sealed envelopes stratified by parity (nulliparous or multiparous).
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Women in the control group underwent a sham manual rotation that involved a vaginal examination over 4 or 5 contractions during which the attending midwife was asked not to observe. However, it is unclear how effective the blinding was (author communication).
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Data collection performed blind to intervention allocation by independent researcher.
Incomplete outcome data (attrition bias) All outcomes	Low risk	No losses reported.
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes.
Other bias	Unclear risk	No statistically significant baseline imbalances. Clinical baseline difference in thick meconium‐stained liquor (manual rotation 40% vs. sham 7%; P value = 0.08).

CTG: cardiotocograph; NICU: neonatal intensive care unit; OA: occiput anterior; OP: occiput posterior; RCOG: Royal College of Gynaecologists.