Table 1 Summary of characteristics of the eleven studies included in the present study
Author (year) | Country | Design and data sources | Facility level, location, obstetric care and managing authority | Type of comparison group | MWH admission criteria and (% MWH use) | Community outreach | Health worker monitoring of users | Study qualityd |
|---|---|---|---|---|---|---|---|---|
Braat et al (2018) | Ethiopia | Hospital-based retrospective cohort. Hospital records and hospital-based survey | Rural NGO hospital providing CEmOC | Direct admissionsa,b | Risk factors, distance (34%) | Well established program set up by Attat hospital | None at MWH, users expected to attend routine ANC clinic at hospital | Selection bias possibly present. No adjustment for confounders. Sample size adequacy uncertain. Findings limited to women using hospital-level care |
Chandramohan et al. (1995) | Zimbabwe | Hospital based cohort. Labour ward logbook | Rural government referral hospital with doctors on staff to handle complicated births | Direct admissions | Focus on risk factors, but any pregnant woman permitted to stay (35%) | No information | Weekly examination by midwife at MWH | Sample size adequacy uncertain. Findings limited to women using hospital-level care |
Fogliati et al. (2017) | Tanzania | Cross-sectional hospital study. Survey and patient records | Rural NGO hospital providing CEmOC | Direct admissions | No information (31%) | No information | No information | No adjustment for confounders. Sample size adequacy uncertain. Uncertainty around comparability of risk profile between groups |
Kelly et al. (2010) | Ethiopia | Hospital-based retrospective cohort. Delivery records | Rural NGO hospital providing CEmOC | Direct admissions | Risk factors (average: 28%, range 25%-56%) | Hospital program to train TBAs and CHWs to identify and refer pregnant women | None at MWH, users expected to attend routine ANC clinic at hospital | Descriptive comparison of perinatal outcomes. Uncertainty around comparability of groups. No information on data source completeness |
Gaym et al. (2012)b | Ethiopia | Hospital-based retrospective cohort. No information | Rural NGO hospital, no information on obstetric care capacity | Direct admissions | No information (10%) | No information | No information | Descriptive comparison of perinatal outcomes between users and non-users that was not the primary aim. Selection bias, sample size adequacy and group comparability all uncertain |
Meshesha et al. (2017) | Ethiopia | Hospital-based study with unclear design. Delivery records | Rural government general hospital. No information about obstetric care capacity | Direct admissions | No information (17%) | No information | No information | Descriptive results limited to women able to access facility-based obstetric care. Inadequate reporting makes it difficult to assess the extent of misclassification bias and selection bias |
Millard et al. (1991) | Zimbabwe | Hospital-based retrospective cohort. Hospital records | Rural NGO hospital with C-section services | Direct admissions | Focus on risk factors and distance, but any pregnant woman permitted to stay (60%) | No information | No information | Older study with inadequate reporting presenting descriptive results of hospital-based sample. Comparability of group risk profile uncertain |
Singh et al. (2016) | Malawi | Hospital-based cross-sectional study. Interviews and intake/discharge forms | 2 sites—rural government hospital and urban health centre. No information on obstetric care capacity | Direct admissions | Focus on risk factors and distance, but any pregnant woman permitted to stay (N/A) | Safe Motherhood project in study district includes community education to increase use of services including MWHs | No information | Descriptive results limited to women able to access facility-based obstetric care. Inadequate reporting makes it difficult to assess the extent of misclassification bias and selection bias |
Tumwine et al. (1996) | Zimbabwe | Hospital-based retrospective cohort. No information | Rural NGO hospital with C-section services | Direct admissions | No information (27%) | No information | Unclear | Older study with inadequate reporting presenting descriptive results of hospital-based sample. Comparability of group risk profile uncertain |
van Lonkhuijzen et al. (2003) | Zambia | Hospital-based retrospective cohort. Surveys | Rural NGO hospital with C-section services | Direct admissions | No information | No information | None at MWH, users expected to attend routine ANC clinic at hospital | Descriptive results limited to women able to access facility-based obstetric care. Inadequate reporting makes it difficult to assess the extent of misclassification bias and selection bias |