Methods	RCT with randomisation of individual wonen.
Participants	UK hospital setting. High-income country. Inclusion criteria: all women expected to give birth vaginally. 1695 women randomised out of a possible 4709 Exclusion criteria: refusal, cardiac disease, APH, non-cephalic presentation, multiple pregnancy, IUFD, if clinician had good reason not to include women After the first 5 months, exclusions included women with ritodrine given 2 hours before birth; anticoagulant treatment; any condition needing a particular management of 3rd stage (e.g. meconium-stained liquor, dural tap)
Interventions	Intervention: active management of 3rd stage (N = 846): prophylactic syntometrine (5 units oxytocin + 0.5 mg ergometrine) (or 10 units syntocinon if mother had raised BP), administered immediately after birth of anterior shoulder; cord clamping and cutting within 30 seconds ofbirth; delivery of the placenta by CCT when uterus contracted. Comparison: expectant management of 3rd stage (N = 849). try not to give oxytocic; try to leave cord attached to baby until placenta delivered; try not to use CCT or any manual interference with uterus at the fundus. Following signs of separation encourage posture aiding delivery by gravity, and maternal effort. Data included in Comparison 1.
Outcomes	Pre-specified outcomes: PPH (and “more objective measures of blood loss”, presumably Hb); length 3rd stage; need for therapeutic oxytocics; manual removal placenta; ERPC; side effects of oxytocics (nausea, vomiting, headaches, hypertension); Apgar scores; PCV; SCBU; jaundice; breastfeeding. Views of a subsample of women
Notes	Actual management used in the active arm: 99% given prophylactic uterotonic before delivery of placenta; 99% cord clamped and cut before delivery of placenta; 99% CCT; 26% upright Actual management used in the expectant arm: 30% received uterotonic for treatment, and 20% prophylactically; cord left unclamped till pulsation ceased 48%; placenta delivered by maternal effort 60% and CCT 40%; 49% upright
Risk of bias
Bias	Authors’ judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	No description of the randomisation given. However, we believe that unclear sequence generation was a consequence of reporting omission rather than methodological inadequacies and therefore does not give rise to bias
Allocation concealment (selection bias)	Low risk	“On admission to the labour ward… Correspondingly numbered, sealed opaque envelopes were placed in the woman’s notes..
Incomplete outcome data (attrition bias) All outcomes	Low risk	Missing data for some outcomes, e.g. 19% of Hb results missing in active arm and 18% in the physiological; 25% of antenatal and/or postnatal Hb results (used to calculate drop in mean Hb) missing in the active arm and 26% in the physiological Apparently no women were excluded after randomisation but 182 are described as having not entered in the trial due to the cord being cut early for fetal safety reasons. The allocation details, however, state that when the clinician was ready to prepare for delivery, the envelope was opened and “all women for whom an envelope was opened were deemed to have entered the trial and were followed up”. The envelope would have been opened before any neonatal need for attention became apparent
Selective reporting (reporting bias)	Unclear risk	Trial protocol not assessed. Reported outcomes not pre-specified: blood transfusions; number of units transfused; days’ stay; serious problems with 3rd stage;neonatal respiratory problems; birth-weight; number with “serious problems in 3rd stage” (not pre-specified), described as PPH, transfusion, manual removal of placenta, ERPC is reported. This is misleading as a number of those problems would co-exist
Other bias	High risk	Protocol was modified after 5 months (425 births), due to high blood loss in expectant management group, to allow women in the control arm who needed some active management to be switched to fully active management. However, data for the first 5 months were still included in analysis Trial was stopped early because of potential harm. Sample size was meant to be 3900 but stopped after 1695 30 women in the control group gave a late maternal refusal, whereas only 1 in the experimental group did so. The outcomes of these women are included in analysis It is questioned whether the midwives had sufficient training in physiological third stage before the trial started. Harding et al found that, of 49 midwives responding to a questionnaire, only one had practised physiological management as defined in the trial. Only 6 (13%) of the mid-wives said that they were very confident of physiological management before the trial and 22 (46%) afterwards (Harding 1989; Prendiville 1988 paper). This trial has been criticised for including all women (including high parity, all age groups, previous PPH, epidural, long labour, operative delivery) and not confining inclusion criteria to women who were low risk. Women at high risk of PPH will have a higher blood loss using expectant management; clinicians experiencing this may respond by anxiety in subsequent births, even oflow-risk women, which may result in higher intervention (mixed management) rates Only 47% (403/849) of women in physiological arm received the full physiological package (a problem with other studies also). But, in particular, 168/849 = 20% had prophylactic oxytocic, which is a large number for a “prophylactic” treatment as opposed to one in response to clinical need; In addition, 252 (30%) had a uterotonic as a treatment, so in total, 50% of the expectant management group received an oxyto-cic However, 99% (838/846) of women in active management group received allocated management