Waldenstrom 1997.

Methods	Randomized controlled trial; women enrolled at first visit (mean 20 weeks' gestation).
Participants	1860 nulliparous and multiparous women (928 in the experimental group and 932 controls) who were: residents of Greater Stockholm, Sweden, and did not have any disease that might complicate the birth or jeopardize the baby's health, including diabetes, multiple pregnancy, pre‐eclampsia, drug abuse, or smoking during the current pregnancy. A history of low birthweight, preterm birth, perinatal death or a difficult vaginal birth were not exclusion criteria. Women were enrolled between October 1989 and June 1993. The 1230 women who gave birth between October 1989 and January 1992 comprised the sample to assess birth satisfaction and breastfeeding.
Interventions	Experimental: antenatal, intrapartum, and postnatal care in a alternative birth centre located 1 floor below the ordinary labour ward at a Stockholm hospital, with 1:1 midwife‐woman ratio during labour, and discharge within 24 hours of the birth. Control: antenatal care at neighbourhood antenatal clinics, intrapartum care in the hospital labour delivery suite (usually each midwife caring for more than 1 woman), and postnatal care for 3‐4 days in the hospital postnatal ward. Staff working in the alternative birth centre did not work in the delivery suite.
Outcomes	Transfers to and reasons for standard care, intrapartum medical interventions, operative delivery, postpartum hemorrhage, 5‐minute Apgar score < 7, transfer to NICU, perinatal mortality, serious perinatal morbidity, at least 1 postnatal home visit, breastfeeding, stopped breastfeeding within 2 months, sore nipples, engorgement, milk stasis, mastitis, satisfaction with care.
Notes	34% of birth centre group were transferred to standard care either antenatally or intrapartally, and an additional 2% were transferred in the postpartum period.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No mention of the process of sequence generation. 100 envelopes prepared at a time, with a 50/50 split between groups. Envelopes were "mingled" in a box. A new batch of envelopes was added when "a few" remained.
Allocation concealment (selection bias)	Low risk	Opaque envelopes were used and participants picked their own from the box.
Blinding (performance bias and detection bias) All outcomes	Unclear risk	Not noted whether data collectors were blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Only 2 were lost to follow‐up for the main study outcomes. > 90% follow‐up in both groups for the postpartum questionnaire at 2 months.
Selective reporting (reporting bias)	Low risk	All outcomes were reported.
Other bias	Low risk	Earlier reports were of an n of 1230, because funding ended. Subsquently, additional funding permitted additional enrolment to increase statistical power for medical outcomes.

MU: the midwife‐led unit 
 NICU: neonatal intensive care unit 
 NU: the normal unit 
 SU: the special unit