Table 2 Characteristics and maternal mortality outcomes of included studies

Manandhar 2004[17]

rural Nepal

Cluster RCT

24 clusters

Facilitator-led women's groups to improve perinatal care practices plus health-service strengthening vs. usual care plus health-service strengthening

28,931 women of childbearing age

6,714 pregnancies

6,165 live births

1:medium;

2: low;

3: medium ;

4: high

69

341

0.28

(0.09 to 0.82)

Jokhio 2005[15]

rural Pakistan

Cluster RCT

7 clusters

TBA training and health service integration, issue of sterile delivery kits vs. usual care

19,557 pregnant women

19,524 deliveries

1:high;

2:high;

3 high;

4: high

268

360

0.74

(0.45 to 1.23)

Munjanja 1996[16]

Harare, Zimbabwe

Cluster RCT

7 clusters

Intervention: fewer, but goal-oriented antenatal visits vs. standard "westernised" antenatal care

15,994 low risk pregnancies

15,532 deliveries

1:medium;

2: medium; 3:medium;

4: low

64

82

0.78

(0.23 to 2.61)

Villar 2001 [20]

Argentina, Cuba, Saudi Arabia, Thailand

Cluster RCT

53 clusters

Intervention: fewer, but goal-oriented antenatal visits vs. standard "westernised" antenatal care

24,526 low risk pregnant women

22,793 single births

1:high;

2:high

3: medium 4:high

60

54

1.11

(0.37 to 3.29)

Majoko 2007[19]

rural Zimbabwe

Cluster RCT

23 clusters

Intervention: fewer, but goal-oriented antenatal visits vs. standard "westernised" antenatal care

13179 pregnant women

1:medium; 2:high;

3: low;

4: high

60

31

1.90

(0.38 to 9.43)

Ackermann-Liebrich 1996[28]

Switzerland

Prospective cohort study with nested matched pairs

Women opting for home vs. hospital birth in "westernised" setting

874 pregnant women

857 deliveries

1:low;

2: low;

3 :not applicable; 4:low

0

0

de Bernis 2000,

Dumont, 2002[29, 30]

Senegal

Prospective survey of two cohorts

Women in Kaolack delivered mainly by TBAs in district birth centres vs. women in St Louis delivered mainly by midwives in hospital

3,777 pregnant women

3,689 deliveries

3,476 live births

1:low; 2:medium;

3:high; 4:unclear

874

151

5·84

(1·66 to 20·53)

Greenwood 1990[31]

rural Gambia

Prospective cohort

TBA training, village health worker support and obstetric pack vs. no additional care

1,963 pregnancies

1:low; 2:medium;

3:medium; 4:unclear

1051

963

1·09

(0·43 to 2·75)

Fauveau 1991; Maine 1996[23, 32]

rural, Bangladesh (Matlab)

Prospective cohort

1987–1989

Midwives working with community health workers and TBAs to attend home births, manage obstetric complications and accompany referral cases to project clinic vs. routine care (not described) plus access to project clinic

9,630 live births

1:low;

2:low;

3:unclear; 4:unclear

136

388

0·35

(0·13 to 0·93)

Ronsmans 1997[24]

rural Bangladesh (Matlab)

Prospective cohort

1990–1993

Access to above Matlab Intervention vs. "routine care"

24,059 live births

1:low;

2:low;

3:unclear; 4:unclear

239

289

0·83

(0·5 to 1·36)

Foord 1995; Fox-Rushby & Foord 1995, 1996. [33–35]

The Gambia

Prospective cohort

Early identification of pregnant women by trained TBAs, mobile antenatal unit to treat anaemia and infections; referral/transfer for obstetric emergency treatment; low-cost insurance scheme to pay for treatment vs. care by TBAs with minimal tertiary facilities

1,059 women delivering

1:low;

2:low;

3:low;

4:unclear

126

693

0·43

(0·02 to 1·55)

Xu 1995[36]

China

Prospective cohort

Reorganisation of maternity care to include better clinical governance, education and training of staff, and some community education

unknown

1:low; 2:unclear;

3:low;

4:unclear

37

93

0·39

Zhang 2004[37]

China

Cohort: Complex stratification of "randomly selected" project and matched non-project areas

Maternal and child health providers at grass roots level given two weeks theory training; some also given one month clinical skills training

unknown

1:low; 2:unclear;

3:low;

4:unclear

53

52

1·06