Table 1. Characteristics of cluster-randomised controlled trials included in the systematic review and meta-analysis.

	Study population and setting	Intervention	Control	Outcomes
Manandhar et al,¹² 2004 (Nepal)	24 clusters; population of about 7000 per cluster Closed cohort of married women of reproductive age (15–49 years) living in Makwanpur district, rural Nepal; pregnancies registered during Nov 1, 2001, to Oct 31, 2003, were followed up	12 clusters (2972 births) Each cluster had a local literate female facilitator who was given a brief training in perinatal health issues and a facilitation manual; facilitators supported women’s groups through ten monthly meetings using a participatory learning and action cycle and a picture card game that addressed prevention and treatment for typical problems in mothers and infants; one supervisor supported three facilitators Health service strengthening and training of traditional birth attendants were as in the control group	12 clusters (3303 births) Health service strengthening activities and training of traditional birth attendants: primary health centres given resuscitation equipment, phototherapy units, and warm cots; essential newborn-care training for local health staff and traditional birth attendants; and newborn-care kits given to community-based workers	Primary: neonatal mortality rate Secondary: stillbirth rate, maternal mortality ratio, uptake of maternity services, care practices at home, neonatal morbidity, and health-care seeking
Tripathy et al,¹³ 2010 (India)	36 clusters; mean population 6338 per cluster (SD 2101) Open cohort of women aged 15–49 years, living in rural areas of three districts of Jharkhand and Orissa, eastern India, who gave birth between July 31, 2005, and July 30, 2008	18 clusters (9770 births) A local woman facilitated 20 monthly meetings with women’s groups after 7 days of training; each facilitator convened 13 groups per month; groups followed a four-phase participatory learning and action cycle and were open to all members of the community though primarily targeting pregnant women and new mothers Facilitators and group members used stories, participatory games, and picture cards to facilitate discussions about prevention and care-seeking Health service strengthening was as in the control group	18 clusters (9260 births) Health service strengthening activities: health committees formed so community members could express opinions about local health services; committees met every 2 months to discuss maternal and newborn health entitlement issues; and workshops using appreciative inquiry provided to frontline government health staff	Primary: neonatal mortality rate and maternal depression scores Secondary: stillbirths, maternal mortality ratio, and perinatal mortality, uptake of maternity services, care practices at home, and health-care seeking
Azad et al,¹⁴ 2010 (Bangladesh)	18 clusters; mean population 27 953 per cluster (SD 5953) Open cohort of women aged 15–49 years living in three rural districts of Bangladesh, who gave birth between Feb 1, 2005, and Dec 31, 2007	Nine clusters (15 695 births) A local woman facilitated groups using a participatory learning and action cycle after receiving five training sessions that covered communication, maternal and neonatal health issues; she visited every tenth household in the intervention clusters and invited married women of reproductive age to join the groups; mothers-in-law, adolescent girls, and other women joined at a later date Health service strengthening and training of traditional birth attendants were as in the control group	Nine clusters (15 257 births) Health service strengthening activities and training of traditional birth attendants: improvements to referral systems and links between communities and health services; and provision of basic and refresher training in essential maternal and newborn care	Primary: neonatal mortality rate Secondary: maternal mortality ratio, stillbirths, perinatal mortality rate, uptake of maternity services, care practices at home, neonatal morbidity, and health-care seeking
More et al,²⁴ 2012 (India)	48 clusters; mean population 5865 per cluster (SD 1077) Women were recruited between Oct 1, 2006, and Sept 30, 2009, in urban Mumbai slums; women from transient communities and areas for which resettlement was being negotiated were excluded	24 clusters (9155 births) A facilitator (local woman with secondary education and leadership skills) set up ten groups in a cluster of 1000 households; groups met fortnightly, and the facilitator met weekly with other facilitators and her supervisor; women’s groups followed a cycle of 36 meetings and were open to all women. Participatory methods with seven phases, based on the principles of appreciative inquiry, were used in the meetings	24 clusters (9042 births); no details were provided about control clusters	Primary: stillbirths, neonatal mortality rate and extended perinatal mortality rate, perinatal care, and maternal morbidity Secondary: maternal mortality ratio, antenatal care, institutional delivery, breastfeeding, and care- seeking for newborn illness
Lewycka et al,²⁵ 2013 (Malawi)^*	48 clusters; mean population 3958 per cluster (SD 404) A cohort of women aged 10–49 years in Mchinji district, rural Malawi, who delivered a child between Feb 1, 2006, and Jan 31, 2009	24 clusters and 9374 births in factorial analysis, 12 clusters and 3129 in stratified analysis for women’s groups Women’s groups were supported by a female facilitator through a participatory learning and action cycle of 20 meetings Facilitators were local, literate women aged 20–49 years; they were trained for 11 days, with refresher training every 4 months, and supported by one supervisor per six facilitators Meetings followed a four-phase participatory learning and action cycle; group membership was restricted to women, but expanded to men in later stages Health service strengthening was as in the control group	24 clusters and 9749 births in overall analysis; 12 clusters and 3329 births in stratified analysis for women’s groups Health service strengthening activities: health workers received training in essential newborn care and safe motherhood; neonatal resuscitation equipment donated to all facilities; a project for prevention of mother-to-child transmission of HIV introduced in 2005 was scaled up to all facilities by 2008	Primary: neonatal, perinatal, and infant mortality rates, and maternal mortality ratio Secondary: maternal and infant morbidity, use of skilled maternity services, immunisation, malaria prophylaxis, use of prevention of mother-to-child transmission services, and breastfeeding
Colbourn et al,²⁶ 2013 (Malawi)^†	32 clusters; mean population of 3934 per cluster (SD 1332) An open cohort of pregnant women was recruited from three rural districts of Malawi between Oct 1, 2008, and Dec 31, 2010; women were excluded if they were living in urban areas, or areas with facilities providing comprehensive emergency obstetric care or non-functioning facilities	15 clusters (10329 births); 81 volunteer facilitators supported by nine MaiKhanda study staff, each formed a women’s group that followed a participatory learning and action cycle to improve maternal and neonatal health	17 clusters (10 247 births): no details reported	Primary: maternal mortality ratio, and perinatal, and neonatal mortality rates Secondary: institutional delivery, percentage of maternal deaths subjected to audit, case fatality rates, practice of signal obstetric- care functions
Fottrell et al,²⁷ 2013 (Bangladesh)	Clusters were the same as in Azad et al¹⁴ An open cohort of women residing in three rural districts of Bangladesh, who were permanent residents of the union in which their delivery was identified from January, 2009, to June, 2011; temporary residents were excluded	Nine cluster (9106 births) In addition to the 162 women’s groups already set up previously (Azad et al¹⁴), 648 new groups were formed by newly recruited facilitators to increase population coverage; from January, 2009, the new groups followed a participatory learning and action cycle with monthly meetings about maternal and newborn health Health service strengthening was as in the control group	Nine clusters (8834 births) Health service strengthening: provision of basic medical equipment to local facilities; training of traditional birth attendants in essential newborn care; and refresher training in essential newborn care for physicians	Primary: neonatal mortality rate Secondary: stillbirth, perinatal mortality rate, pregnancy- related mortality, institutional delivery, home-care practices, and health-care seeking

2×2 factorial, cluster-randomised controlled trial of volunteer peer counselling support for breastfeeding and infant care.

^†

2×2 factorial, cluster-randomised controlled trial of quality improvement of health facilities.