Community‐based maternal and newborn educational care packages for improving neonatal health and survival in low‐ and middle‐income countries

Zohra S Lassi; Sophie GE Kedzior; Zulfiqar A Bhutta

doi:10.1002/14651858.CD007647.pub2

. 2019 Nov 5;2019(11):CD007647. doi: 10.1002/14651858.CD007647.pub2

Community‐based maternal and newborn educational care packages for improving neonatal health and survival in low‐ and middle‐income countries

Zohra S Lassi ¹, Sophie GE Kedzior ², Zulfiqar A Bhutta ^3,^✉

Editor: Cochrane Neonatal Group

PMCID: PMC6828589 PMID: 31686427

Abstract

Background

In low‐ and middle‐income countries (LMICs), health services are under‐utilised, and several studies have reported improvements in neonatal outcomes following health education imparted to mothers in homes, at health units, or in hospitals. However, evaluating health educational strategy to deliver newborn care, such as one‐to‐one counselling or group counselling via peer or support groups, or delivered by health professionals, requires rigorous assessment of methodological design and quality, as well as assessment of cost‐effectiveness, affordability, sustainability, and reproducibility in diverse health systems.

Objectives

To compare a community health educational strategy versus no strategy or the existing approach to health education on maternal and newborn care in LMICs, as imparted to mothers or their family members specifically in community settings during the antenatal and/or postnatal period, in terms of effectiveness for improving neonatal health and survival (i.e. neonatal mortality, neonatal morbidity, access to health care, and cost).

Search methods

We used the standard search strategy of Cochrane Neonatal to search the Cochrane Central Register of Controlled Trials (CENTRAL; 2017, Issue 4), in the Cochrane Library, MEDLINE via PubMed (1966 to 2 May 2017), Embase (1980 to 2 May 2017), and the Cumulative Index to Nursing and Allied Health Literature (CINAHL) (1982 to 2 May 2017). We also searched clinical trials databases, conference proceedings, and the reference lists of retrieved articles for randomised controlled trials and quasi‐randomised trials.

Selection criteria

Community‐based randomised controlled, cluster‐randomised, or quasi‐randomised controlled trials.

Data collection and analysis

Two review authors independently assessed trial quality and extracted the data. We assessed the quality of evidence using the GRADE method and prepared 'Summary of findings' tables.

Main results

We included in this review 33 original trials (reported in 62 separate articles), which were conducted across Africa and Central and South America, with most reported from Asia, specifically India, Pakistan, and Bangladesh. Of the 33 community educational interventions provided, 16 included family members in educational counselling, most frequently the mother‐in‐law or the expectant father. Most studies (n = 14) required one‐to‐one counselling between a healthcare worker and a mother, and 12 interventions involved group counselling for mothers and occasionally family members; the remaining seven incorporated components of both counselling methods.

Our analyses show that community health educational interventions had a significant impact on reducing overall neonatal mortality (risk ratio (RR) 0.87, 95% confidence interval (CI) 0.78 to 0.96; random‐effects model; 26 studies; n = 553,111; I² = 88%; very low‐quality evidence), early neonatal mortality (RR 0.74, 95% CI 0.66 to 0.84; random‐effects model; 15 studies that included 3 subsets from 3 studies; n = 321,588; I² = 86%; very low‐quality evidence), late neonatal mortality (RR 0.54, 95% CI 0.40 to 0.74; random‐effects model; 11 studies; n = 186,643; I² = 88%; very low‐quality evidence), and perinatal mortality (RR 0.83, 95% CI 0.75 to 0.91; random‐effects model; 15 studies; n = 262,613; I² = 81%; very low‐quality evidence). Moreover, community health educational interventions increased utilisation of any antenatal care (RR 1.16, 95% CI 1.11 to 1.22; random‐effects model; 18 studies; n = 307,528; I² = 96%) and initiation of breastfeeding (RR 1.56, 95% CI 1.37 to 1.77; random‐effects model; 19 studies; n = 126,375; I² = 99%). In contrast, community health educational interventions were found to have a non‐significant impact on use of modern contraceptives (RR 1.10, 95% CI 0.86 to 1.41; random‐effects model; 3 studies; n = 22,237; I² = 80%); presence of skilled birth attendance at birth (RR 1.09, 95% CI 0.94 to 1.25; random‐effects model; 10 studies; n = 117,870; I² = 97%); utilisation of clean delivery kits (RR 4.44, 95% CI 0.71 to 27.76; random‐effects model; 2 studies; n = 17,087; I² = 98%); and care‐seeking (RR 1.11, 95% CI 0.97 to 1.27; random‐effects model; 7 studies; n = 46,154; I² = 93%).

Cost‐effectiveness analysis conducted in seven studies demonstrated that the cost‐effectiveness for intervention packages ranged between USD 910 and USD 11,975 for newborn lives saved and newborn deaths averted. For averted disability‐adjusted life‐year, costs ranged from USD 79 to USD 146, depending on the intervention strategy; for cost per year of lost lives averted, the most effective strategy was peer counsellors, and the cost was USD 33.

Authors' conclusions

This review offers encouraging evidence on the value of integrating packages of interventions with educational components delivered by a range of community workers in group settings in LMICs, with groups consisting of mothers, and additional education for family members, for improved neonatal survival, especially early and late neonatal survival.

Plain language summary

Community‐based maternal and newborn educational care packages for improving neonatal health and survival in low‐ and middle‐income countries

Review question

Is community health educational intervention for newborn care effective in improving neonatal health and survival in low‐ and middle‐income countries?

Background

In low‐ and middle‐income countries (LMICs), health service utilisation is low and neonatal mortality and morbidity are high. However, improvements in neonatal outcomes have been documented in several studies with simple health educational interventions. This review assessed the effectiveness of health education strategies imparted to mothers or their family members in community settings of LMICs. It also assessed the impact of health education strategies on neonatal mortality, neonatal morbidity, access to health care, and cost.

Study characteristics

A total of 33 experimental studies were conducted across Africa and Central and South America, with most reported from Asia, specifically India, Pakistan, and Bangladesh. Of the 33 community educational interventions, 16 required involvement of family members, most frequently the mother‐in‐law or the expectant father. Most studies (n = 14) involved one‐to‐one counselling between a range of community healthcare workers and mothers, and 12 involved group counselling consisting predominantly of mothers, with family members included occasionally; the remaining seven had components of both one‐to‐one and group counselling.

Key results

This review found that community health educational interventions significantly reduced newborn death, early newborn mortality, and late newborn mortality, as well as perinatal mortality. These interventions also positively impacted utilisation of any before birth (antenatal), care during pregnancy, and initiation of breastfeeding within an hour after birth. The review shows that educational interventions delivered to both mothers and other family members in a group setting had a greater impact on these outcomes. Educational interventions delivered during antenatal care were more effective for reducing early neonatal deaths, and those delivered during both antenatal and postnatal (after birth) periods were effective for reducing late neonatal deaths and perinatal deaths. Educational interventions during the postnatal period were most effective for improving breastfeeding practices.

Quality of evidence

The quality of evidence is low for newborn mortality outcomes and very low for early, late, and perinatal mortality. This reflects concerns of bias, inconsistency (unexplained variability of results), and imprecision (variation in studies presenting both benefit and harm from the intervention) of the included randomised controlled trials.

Summary of findings

Background

Description of the condition

Globally 2.6 million children die in the first month of life and approximately 7000 newborn deaths occur every day, with about one million dying on the first day and close to one million dying within the next six days of life. Most of these deaths occur in low‐ and middle‐income countries (LMICs) (WHO 2018; UNICEF 2017). Deaths occurring in the neonatal period (aged 0 to 27 days) account for 46% (2.614 million) of all deaths among children younger than five years (UNICEF 2017). Extremely high neonatal mortality rates (over 28 per 1000 live births) are typical of several sub‐Saharan African and South Asian countries (UNICEF 2017); currently 1 in 36 newborns die in sub‐Saharan Africa during the first month of life (UNICEF 2017). The precise contributions of various causes of neonatal death are difficult to ascertain because a vast majority of births and deaths occur in homes and thus are poorly reported and categorised (Black 2010). However, birth asphyxia/intrapartum complications and complications due to preterm birth and infectious causes are recognised as major cause of neonatal death (Black 2010; Lawn 2004; Lawn 2005; UNICEF 2017).

Description of the intervention

In an effort to improve outcomes for both mothers and their newborn infants, the "Mother‐Baby Package" was introduced by the World Health Organization (WHO) in 1994 (WHO 2006a). The Mother‐Baby Package consists of a diverse set of interventions considered essential to maternal and newborn health. These interventions include antenatal registration and care, iron or folate supplementation, tetanus toxoid immunisation, and prevention and management of sexually transmitted infections (STIs) and human immunodeficiency virus (HIV) in endemic areas. They involve treatment for underlying medical conditions such as malaria and hookworm infestation, nutritional advice, ensuring clean delivery, presence of a trained birth attendant at delivery, recognition and management of maternal and neonatal complications, neonatal resuscitation, early and exclusive breastfeeding, and prevention and management of neonatal hypothermia and infections such as ophthalmia neonatorum and cord infection. Implementation and coverage of the "Mother‐Baby Package" vary and the services offered are poorly utilised. Community educational interventions targeting expectant mothers and their family members provide an opportunity to educate the mother and her support network on appropriate care during the antenatal and postnatal periods, especially in populations with minimal access to appropriate antenatal care. Providing mothers with skills and methods to access appropriate care can greatly benefit neonatal and maternal health outcomes. These interventions may be disseminated in homes, at health units, or in hospitals and consist of different counselling strategies from a range of healthcare workers.

In LMICs, almost half of mothers lack adequate antenatal care and home births are extremely common (Benova 2018); however only 13% of women who have birthed at home receive postnatal care within 24 hours (WHO 2018), and in 28 African countries, only 66% of births take place in the presence of a skilled birth attendant (Chukwuma 2017). In many settings, care for mother and baby during the critical first few days after delivery is provided entirely outside the formal healthcare sector. In the least developed regions, contraception prevalence is only 40% (UN 2015a), with the proportion of unmet need for contraception highest among women in sub‐Saharan Africa, at 24% (UN 2015a). Also in the least developed regions, fertility rates (average lifetime number of live births per woman used as current fertility rates) are as high as 4.3 live births per woman (UN 2015), which is especially prevalent in Middle and Western Africa (UN 2015). There is also an urban‐rural gap in contraceptive use in many developing regions. This gap is particularly large in sub‐Saharan Africa, where just 18% of rural women and 31% of urban women are using any method of contraception (UNICEF 2010).

Although reasons for high neonatal mortality rates are multi‐factorial and include shortcomings in supply (such as lack of manpower, poor quality, or dearth of medical supplies and equipment), poor health centre to community linkages, malfunctioning referral systems, non‐existent emergency transportation facilities, and inadequately trained service providers and birth attendants, a major factor is the lack of demand for services provided (Atuoye 2015; Ensor 2004; Lawn 2004; Lawn 2005; Nair 2010; Osrin 2003). This is the result of numerous socioeconomic and cultural factors operating at an individual level and at a collective community level, such as poverty, lack of awareness of services offered, aversion to hospitalisation and formal medical care, lack of awareness of when and how to seek help if desired, and lack of female participation in family decision‐making (Ahmed 2001; Bang 2001; Bhardwaj 1995; Bohren 2014; de Zoysa 1998; Ensor 2004; Riaz 2015).

How the intervention might work

In LMICs, health services may be under‐utilised. Several studies have reported improvements in neonatal outcomes following health education on maternal and newborn care imparted to mothers, in home, at a health unit, or in hospital (Baqui 2008 (a); Pasha 2013; Tripathy 2010). However, the evaluation of any health educational strategy, such as one‐to‐one counselling or group counselling via peer or support groups, through the organisation of men's or women's groups, or delivered by healthcare professionals, requires rigorous assessment of methodological design and quality, as well as assessment of cost‐effectiveness, affordability, sustainability, and reproducibility in diverse health systems.    The "Warmi Project" in rural Bolivia achieved significant reductions in perinatal and neonatal mortality rates (from 11.7% pre‐intervention to 4.4% post‐intervention) through support of women's organisations and community health education (O'Rourke 1998). In three rural districts of Pakistan, local women were trained to deliver primary health care and health education and to facilitate community organisation for health improvement (Barzgar 1997). In a poor urban district of Brazil, significant improvements in maternal knowledge and health behaviour were documented following implementation of the "ProNatal Project" which, among other interventions, provided health education at newly established antenatal clinics and in homes (Emond 2002). One year after initiation, significant reductions in infant mortality and diarrhoea‐related mortality, as well as increased use of contraception, were reported. In Bangalore, India, a one‐to‐one educational session with mothers of children under five years resulted in significant improvements in most aspects of home management of diarrhoea (Mangala 2001). The "Newhints" trial is a cluster‐randomised controlled trial based in Ghana that utilises community‐based surveillance officers to deliver education to improve newborn care practices and care‐seeking during pregnancy and childbirth (Kirkwood 2013). This trial was shown to increase care‐seeking by mothers while reducing the neonatal mortality rate. The "UNEST" cluster‐randomised controlled trial based in Uganda utilised community health workers (CHWs) to provide pregnant women with one‐to‐one counselling to improve newborn practices and showed that the intervention group had a greater proportion of women who initiated breastfeeding within an hour after birth compared to the control group (Waiswa 2015). In Malawi, Africa, the MaiKhanda trial utilised a participatory women's group community intervention and facility quality improvement; 50% of the formed groups developed maternal and neonatal health task forces to enhance antenatal coverage and maternal and neonatal health knowledge, as well as facility delivery (Colbourn 2013). Through this women's group approach, a 22% reduction in neonatal mortality was observed and the facility quality improvement intervention appeared to be most effective in reducing late neonatal deaths.

Why it is important to do this review

This systematic review assessed the effectiveness of community health education on maternal and newborn care for improving neonatal survival in LMICs, and attempted to compare the costs of such strategies. The purpose of this review was to determine whether community health education on maternal and newborn care is an effective and cheap method for reducing neonatal mortality and morbidity, and to identify which strategies resulted in the best neonatal outcomes. A plethora of evidence suggests that community‐based interventions are important for improving healthcare delivery and related outcomes (Lassi 2015; Lassi 2016); however no systematic reviews are currently focusing on targeting mothers and their families with health education to improve neonatal and maternal health outcomes. CHWs and other community facilitators can be at the forefront of interventions that may involve education and health promotion, with some interventions demonstrating that CHWs are able to empower communities to change their health behaviours. Because of the increasing rise in healthcare costs, it has become crucial to focus on developing affordable ways to promote health in community settings. It is therefore important to review and compile recent evidence in the form of randomised controlled trials to determine whether community health education on maternal and newborn care is cost‐effective and has the ability to improve neonatal health and survival.

Objectives

To compare a community health education strategy versus no strategy or the existing approach to health education on maternal and newborn care in LMICs, as imparted to mothers or their family members specifically in community settings during the antenatal and/or postnatal period, in terms of effectiveness for improving neonatal health and survival (i.e. neonatal mortality, neonatal morbidity, access to health care, and cost).

Methods

Criteria for considering studies for this review

Types of studies

Community‐based (home, basic health unit (BHU)† or first‐level health facility‡) randomised controlled, cluster‐randomised, or quasi‐randomised controlled trials. We obtained disaggregated data for neonates from trials conducted on neonates as well as children in older age groups.

†A BHU providing primary level health care is either "a dispensary, health post or Maternal and Child health/Family planning (MCH/FP) clinic which provides basic health services, such as health education, simple laboratory tests and treatment".

‡A first‐level health facility or the first referral level is a "district hospital with around 20 beds, providing inpatient services with staff of one or more physicians and few medical specialists, and equipment necessary to carry out most life‐saving surgical and medical procedures".

Types of participants

Types of participants included the following groups.

Women of reproductive age.
Pregnant women at any period of gestation.
Mothers of neonates (up to 28 days of life).
Their spouses/partners.
Other family members (such as mothers‐in‐law).

All participants resided in LMICs.

Types of interventions

Intervention

Community health education on maternal and newborn care* imparted to mothers or their family members in community settings of LMICs (according to the World Bank list (World Bank 2018)) via:

one‐to one‐counselling;
group counselling (in the form of group sessions);
mass media (radio, television, cellular messages, newspaper, brochures, banners, etc.); we will include only studies in which mass media was introduced as a trial for a certain period or as a pilot before launching at a national level to see clear impacts of this intervention compared to control.
Any combination of the above.

The original protocol described a comparison between one‐to‐one counselling and group counselling; however no included studies compared one‐to‐one versus group counselling.

For this review, maternal and newborn care was defined as an intervention essential to maternal and newborn health such as antenatal care, iron/folic acid supplementation, tetanus toxoid immunisation, prevention and management of STI/HIV in endemic areas, nutritional advice, ensuring clean delivery, presence of a trained birth attendant at delivery, recognition and management of maternal and neonatal complications, neonatal resuscitation, early and exclusive breastfeeding, and prevention and management of neonatal hypothermia, neonatal infections, and immunisation.

We did not include studies with a single intervention such as promotion of breastfeeding or use of family planning methods.

Control

The control did not receive the additional educational intervention or received a conventional level of health education.

Types of outcome measures

Primary outcomes

Neonatal mortality

The number of neonatal deaths from any cause among all live births during the trial period.

Early neonatal mortality: from birth through six completed days of life.
Late neonatal mortality: from seven to 28 completed days of life.

Perinatal mortality

The number of stillbirths and the number of newborn deaths within one week of life among all stillbirths and live births during the trial period.

Secondary outcomes

Neonatal infections
1. Number of neonates diagnosed with infection (as defined by study authors) among all live births during the trial period. Infections included:
  1. sepsis;
  2. pneumonia;
  3. meningitis;
  4. gastroenteritis;
  5. tetanus; or
  6. any combination of the above.
Any antenatal care
1. Number of pregnant women among all pregnant women who were attended for reasons related to pregnancy by skilled health personnel (a doctor, or people with midwifery skills who can manage normal deliveries and diagnose or refer obstetrical complications, or both) at least once during pregnancy. Both trained traditional and untrained traditional birth attendants were excluded (WHO 2006).
Use of any method of contraception
1. Number of women of reproductive age (15 to 49 years) or their spouses who reported that they used any contraceptive method during the trial period (any contraceptive method such as female and male sterilisation, injectable and oral hormones, intrauterine devices, diaphragms, spermicides and condoms, natural family planning, and lactational amenorrhoea) (WHO 2006).
Skilled attendance at delivery
1. Number of births among all live births during the trial period that were attended by skilled health personnel (such as a doctor or a skilled attendant capable of managing normal deliveries and referring obstetrical complications, excluding trained or untrained traditional birth attendants) (WHO 2006).
Delivery attended by unskilled or semi‐skilled birth attendant
1. Number of births among all live births during the trial period that were attended by an unskilled (such as an untrained traditional birth attendant, or a relative) or semi‐skilled birth attendant (such as a trained traditional birth attendant).
Use of clean delivery kit
1. Number of deliveries occurring during the trial period for which a clean delivery kit was used by the birth attendant (typically containing a plastic sheet delivery surface, a clean cutting instrument (a new razor blade), clean ties for the cord, soap for ensuring clean hands of the birth attendant, and instructions) (Beun 2003; PATH 2005).
Care‐seeking
1. Number of mothers among all mothers of neonates who sought or were reported to have sought medical care for their neonate's illness during the trial period.
Use of colostrum
1. Number of women who used colostrum among all women who delivered live born babies during the trial period.
Timely initiation of breastfeeding
1. Number of women who initiated breastfeeding within one hour of birth among all women who delivered live born babies during the trial period.
Mothers' understanding of each of the following "healthy" behaviours, among all women interviewed
1. Ways to prevent neonatal infection (such as handwashing, cord care)
2. Signs of neonatal infection
3. Advantages of breastfeeding
4. Family planning methods
5. Willingness to seek formal medical care for neonatal illness
6. Knowledge of health services offered in the community
Total cost of intervention
1. In US dollars for intervention (including recruiting and training personnel to deliver interventions and conducting sessions) among all recipients of the intervention.
Cost per neonatal life saved
1. Cost in US dollars for each neonatal life saved among all live births during the trial period.

Search methods for identification of studies

We used the criteria and standard methods of Cochrane and Cochrane Neonatal.

Electronic searches

We conducted a comprehensive search including the Cochrane Central Register of Controlled Trials (CENTRAL; 2017, Issue 4) in the Cochrane Library; MEDLINE via PubMed (1 January 2012 to 2 May 2017); Embase (1 January 2012 to 2 May 2017); and the Cumulative Index to Nursing and Allied Health Literatue (CINAHL) (1 January 2012 to 2 May 2017), using the search detailed in Appendix 1. We did not apply language restrictions. This search was run to update searches previously run for the reviews published in October 2010 and October 2012, using the search detailed in Appendix 2.

We searched clinical trials registries for ongoing or recently completed trials (clinicaltrials.gov; the World Health Organization’s International Trials Registry and Platform www.whoint/ictrp/search/en/; and the ISRCTN Registry) on 2 May 2017.

Searching other resources

We also searched the reference lists of any articles selected for inclusion in this review to identify additional relevant articles.

Data collection and analysis

We used the standard review methods of Cochrane and Cochrane Neonatal (Cochrane Handbook for Systematic Reviews of Interventions) (Higgins 2011).

Selection of studies

Two review authors ‐ Zohra Lassi (ZL) and Sophie Kedzior (SK) ‐ independently assessed inclusion of all potential studies identified through the search. We resolved disagreement through discussion, and, if required, we consulted a third review author ‐ Zulfiqar Bhutta (ZB).

Data extraction and management

We designed a form on which to extract data. For eligible studies, two review authors (ZL and SK) independently extracted data using the agreed form. We resolved discrepancies through discussion, or, if required, we consulted a third review author. We entered data into Review Manager software and checked them for accuracy (RevMan 2011). We attempted to contact authors of the original reports to request further details when information regarding any of the above was unclear.

Assessment of risk of bias in included studies

Two review authors (ZL and SK) independently assessed the methodological quality of each included trial using the criteria displayed in Appendix 3.

Measures of treatment effect

We carried out statistical analysis using Review Manager software (RevMan 2011).

Dichotomous data

We presented results as summary risk ratio (RRs) with 95% confidence intervals (CI) for dichotomous data.

Continuous data

We planned to use the mean difference (MD) if outcomes were measured in the same way between trials for continuous data. We used the standardised mean difference (SMD) to combine trials that measure the same outcome but used different methods.

Unit of analysis issues

Cluster‐randomised trials

When trials used clustered‐randomisation, we anticipated that study investigators would have presented their results after appropriately controlling for clustering effects (e.g. variance inflated standard errors, hierarchical linear models). When appropriate controls for clustering were not used, we adjusted for cluster effect using the intracluster coefficient (ICC) from the study similar in context and nature. We included cluster‐randomised/quasi‐randomised trials in the analyses along with individually randomised/quasi‐randomised trials. We incorporated data from cluster‐randomised/quasi‐randomised trials using the generic inverse variance method, in which logarithms of RR estimates were used along with the standard error of the logarithms of RR estimates (Higgins 2011).

Dealing with missing data

We noted levels of attrition for included studies. For all outcomes, we carried out analyses, as far as possible, on an intention‐to‐treat basis, that is, we attempted to include in the analyses all participants randomised to each group. The denominator for each outcome in each trial was the number randomised minus the number of any participants whose outcomes are known to be missing.

Assessment of heterogeneity

We applied tests of heterogeneity between trials, if appropriate, using the I² statistic, and by visual inspection of forest plots. If we identified high levels of heterogeneity among trials and visual inspection of forest plots was suggestive, we explored this by performing pre‐specified subgroup analysis. We pre‐specified the following subgroup analysis to investigate heterogeneity in the primary outcome.

Counselling type: one‐to‐one compared to group counselling, or both.
Neonatal mortality rate at baseline.
Timing of intervention: pre‐conceptual, antenatal versus postnatal.
Who receives intervention: mothers, their spouses, or other family members (such as mothers‐in‐law).
Who provides counselling: support groups or peers, health professionals, traditional birth attendants, village health workers, and so forth.

Assessment of reporting biases

Refer to Appendix 3.

Data synthesis

We carried out statistical analysis using Review Manager software (RevMan 2011). We used fixed‐effect meta‐analysis for combining data when trials were examining the same intervention and the trials' populations and methods were judged sufficiently similar. We used random‐effects meta‐analyses when we suspected clinical or methodological heterogeneity between studies sufficient to suggest that treatment effects may differ between trials. If we identified substantial heterogeneity in a fixed‐effect meta‐analysis, we noted this and repeated the analysis using a random‐effects method.

Quality of evidence

We used the GRADE approach to assess the quality of evidence for the following outcomes: neonatal mortality, early neonatal mortality, late neonatal mortality, and perinatal mortality. Two review authors (SK and ZL) independently assessed the quality of evidence for each of the aforementioned outcomes. Primary outcome data were pooled for randomised controlled trials and evidence was downgraded from "high quality" by one level for serious (or by two for very serious) for study limitations (risk of bias), indirectness of evidence, serious inconsistency, imprecision of effect estimates, or potential publication bias. We used the GRADEpro GDT Guideline Development Tool to create a ‘Summary of findings’ table to report the quality of evidence.

The GRADE approach results in an assessment of the quality of a body of evidence according to one of four grades.

High: we are very confident that the true effect lies close to that of the estimate of the effect.
Moderate: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.
Low: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.
Very low: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

Subgroup analysis and investigation of heterogeneity

We applied tests of heterogeneity between trials, if appropriate, by using the I² statistic and by visually inspecting forest plots. If we identified high levels of heterogeneity among the trials and visual inspection of forest plots was suggestive, we explored this by conducting pre‐specified subgroup analysis. We pre‐specified the following subgroup analysis to investigate heterogeneity in the primary outcome.

Counselling type: one‐to‐one versus group counselling, or both.
Neonatal mortality rate at baseline: 30 or more per 1000 live births versus fewer than 30 per 1000 live births.
Timing of intervention: pre‐conceptional, antenatal versus postnatal.
Who receives intervention: mothers, their spouses, or other family members (such as mothers‐in‐law).
Who provides counselling: support groups or peers, health professionals, traditional birth attendants, village health workers, and so forth.

Sensitivity analysis

We carried out sensitivity analyses to explore the effects of adequate allocation concealment and other 'Risk of bias' components on primary outcomes (Appendix 3).

Results

Description of studies

See Characteristics of included studies, Characteristics of excluded studies, Characteristics of studies awaiting classification, and Characteristics of ongoing studies tables.

Results of the search

We ran multiple comprehensive searches for the full review. The PRISMA diagram includes the completed searches and study selections for both 2012 and 2018 (Figure 1). A total of 178 full‐text papers were reviewed for the systematic review; we classified one as ongoing and eight as awaiting classification. Finally, 33 (reported in 62 separate reports) studies met the eligibility criteria for inclusion. All were published journal articles.

Included studies

All 33 included studies were randomised or quasi‐randomised controlled trials.

Comparison 1

"ANY community health educational interventions versus control" (Analysis 1.1; Analysis 1.2; Analysis 1.3; Analysis 1.4; Analysis 1.5; Analysis 1.6; Analysis 1.7; Analysis 1.8; Analysis 1.9; Analysis 1.10; Analysis 1.11; Analysis 1.12; Analysis 1.13)

Study	Family members included	Women/support group	Education given by	ANC, PNC, or both	Baseline NMR (> or < 30 per 1000) I, C	Counselling type (one‐to‐one counselling/group counselling/both)
Ayiasi 2016	Yes	No	CHWs (Village Health Team)	Yes (both)	NA	One‐to‐one
Azad 2010	Yes (mother‐in‐law)	Yes	Local female peer facilitators	Yes (ANC)	21.6, 26.8	Group
Baqui 2008 (a)	Yes (husband)	No	CHWs	Yes (both)	(46.9, 46.7), 48	Both
Baqui 2008 (b)	No	No	CHWs	Yes (both)	NA	Group
Bashour 2008	No	No	Midwives	Yes (PNC)	NA	One‐to‐one
Bhandari 2012	Yes	Yes	CHWs	Yes (PNC)	32.6, 32.4	Group
Bhutta 2008	No	No	LHWs and TBAs	Yes (both)	110.08, 94.64	Both
Bhutta 2011	Yes (additional family members, husband)	Yes	LHWs and TBAs	Yes (both)	48, 51.3	Both
Colbourn 2013	No	Yes	Volunteer facilitators	Yes (both)	(33.3, 29.4, 24) 31.8	Group
Darmstadt 2010	Yes	No	CHWs	Yes (both)	27.9, 25.2	One‐to‐one
Degefie 2017	Yes	No	CHWs, volunteers	Yes (both)	During days 0 to 27: 35, 33.6	One‐to‐one
Fottrell 2013	Yes (men)	Yes	Facilitators	Yes (both)	38.5 (CI 34.8 to 37.4)	Group
Jokhio 2005	No	No	TBAs, LHWs	Yes (both)	NA ‐ No baseline	One‐to‐one
Kirkwood 2013	Yes	No	CBSVs	Yes (both)	32.3, 32.7	One‐to‐one
Kumar 2008	Yes (mother and father‐in‐law, husband)	No	CHWs	Yes (both)	(64.1, 58.9) 54.2	Both
Lewycka 2013	No	Yes	Volunteer peer counsellors	Yes (both)	Assumed baseline MR (34, 27, 76)	Both
Magoma 2013	Yes (male partner)	No	Care providers	Yes (both)	NA	One‐to‐one
Manandhar 2004	No	Yes	Female facilitators	Yes (both)	25.4, 25.1	Group
McConnell 2016	No	No	CHWs	Yes (PNC)	NA	One‐to‐one
Memon 2015	Yes (father, father‐in‐law, mother‐in law)	Yes	CHWs, LHWs	Yes (both)	26, 39.8	Group
Mersal 2013	No	No	NA	Yes (both)	NA	One‐to‐one
Midhet 2010	Yes (husband)	Yes	Female volunteers	Yes (ANC)	NA	Group
More 2012	Yes (other women in the family)	Yes	Female facilitators (Sakhi)	Yes (ANC)	22.3, 18.6	Group
Pasha 2013	No	Yes	TBAs	Yes (PNC)	23.8, 22.5	Both
Penfold 2014	Yes (father, mothers‐in‐law)	No	Community volunteers	Yes (both)	35, 47	One‐to‐one
Persson 2013	No (only if details of neonatal death were required)	Yes	Volunteers from Women's Union, VHWs	Yes (both)	> 15 for both groups	Group
Soofi 2017	No (male community mobilisers ‐ no specifics on relation)	Yes	LHWs	Yes (both)	43.7, 44.6	Both
Srinivasan 1995	No	No	Female ANMs	Yes (both)	18.7, 15.5, 18	One‐to‐one
Ijumba 2015	Yes (participants' mothers, grandmothers, sisters, husband/boyfriend)	No	CHWs	Yes (both)	NA	One‐to‐one
Tripathy 2010	No	Yes	Facilitators	Yes (both)	61.8, 53.6	Group
Tripathy 2016	No	Yes	ASHAs	Yes (both)	63.4, 51.7	Group
Waiswa 2015	No	No	CHWs	Yes (both)	NA	One‐to‐one
Wu 2011	No (no intervention; for survey, yes if mother was not available)	No	Midwives	Yes (both)	37.9, 30.8	One‐to‐one

Community health educational interventions compared to control in developing countries
Patient or population: developing countries  Setting: low‐middle‐income countries, community  Intervention: community health educational interventions  Comparison: control
Outcomes	Relative effect  (95% CI)	№ of participants  (studies)	Certainty of the evidence  (GRADE)	Comments
Neonatal mortality	RR 0.87  (0.78 to 0.96)	553111  (26 RCTs)	⊕⊕⊝⊝  Low^a,b	25/26 studies have unsure risk or high risk for more than 1 type of bias, with the most common high risk being performance bias. However because the outcome is mortality, blinding is objective and therefore is unlikely to be affected by blinding of outcome assessment. Not all confidence intervals from the studies overlap; there is also inconsistency in direction across studies
Early neonatal mortality	RR 0.74  (0.66 to 0.84)	321588  (15 RCTs)	⊕⊝⊝⊝  Very low^a,b,c	There was a combination of high and unclear risk of bias regarding blinding; this can mainly be attributed to the cluster‐randomised design. The study with the greatest risk of bias concerns had the smallest weighting. Most of the studies overlap and show the same direction of effect; however there is 1 major outlier that is also in the opposite direction. The statistical measure for heterogeneity is high, suggesting inconsistency. The confidence interval is wide; however the sample size is sufficient
Late neonatal mortality	RR 0.54  (0.40 to 0.74)	186643  (11 RCTs)	⊕⊝⊝⊝  Very low^a,b,c	High risk of bias was present for blinding across studies; however studies were cluster‐randomised trials, and this is justifiable. I² (88%) was considerably large; however most of the confidence intervals overlap, and direction of effect is consistent. The confidence interval is wide; however the sample size is sufficient
Perinatal mortality	RR 0.83  (0.75 to 0.91)	262613  (15 RCTs)	⊕⊝⊝⊝  Very low^b,c	I² (81%) was considerably large; some studies (although with small weighting) support the control, and others support the intervention. Most of the confidence intervals overlap; however some CIs are large. The confidence interval is wide; however the sample size is sufficient
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).    CI: confidence interval; LMICs: low‐ and middle‐income countries; RCT: randomised controlled trial; RR: risk ratio.
GRADE Working Group grades of evidence.  High certainty: we are very confident that the true effect lies close to that of the estimate of the effect.  Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.  Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.  Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

Community health educational one‐to‐one and group and both counselling (subgroup) compared to control in LMICs
Patient or population: developing countries  Setting: low‐middle‐income countries, community  Intervention: community health educational one‐to‐one and group and both counselling (subgroup)  Comparison: control
Outcomes	Relative effect  (95% CI)	№ of participants  (studies)	Certainty of the evidence  (GRADE)	Comments
Neonatal mortality ‐ One‐to‐one counselling	RR 0.92  (0.71 to 1.20)	105,735  (8 RCTs)	⊕⊝⊝⊝  Very low^a,b,c	Not all confidence intervals overlap, and there is some inconsistency in the direction of effect. However the study with the greatest confidence interval reported weights of only 0.2%. This was further downgraded by 1 as the confidence interval includes both important benefit or harm and no effect. Most studies had concerns in areas of bias that may have affected the outcome; however those with high risk of performance bias should not have impacted the outcome of neonatal mortality
Neonatal mortality ‐ Group counselling	RR 0.83  (0.74 to 0.92)	211,164  (12 RCTs)	⊕⊕⊝⊝  Low^a,b	Around half of the studies raised concerns in areas of bias that may impact the direction of effect; therefore this was downgraded by one level. Not all confidence intervals from the studies overlap; there is also inconsistency in direction across studies
Neonatal mortality ‐ Both group and one‐to‐one counselling	RR 0.90  (0.76 to 1.06)	236,212  (6 RCTs)	⊕⊕⊕⊝  Moderate^b	Not all confidence intervals from the studies overlap; there is also inconsistency in direction across studies
Early neonatal mortality ‐ Group counselling	RR 0.70  (0.61 to 0.80)	122,151  (9 RCTs)	⊕⊕⊝⊝  Low^a,b	There was a combination of high and unclear risk of bias regarding blinding; this can be attributed mainly to the cluster‐randomised design. The study with the greatest risk of bias concerns had the smallest weighting. Most studies are consistent in the direction of effect; however not all confidence intervals overlap
Early neonatal mortality ‐ One‐to‐one counselling	RR 1.30  (1.01 to 1.67)	18,747  (1 RCT)	⊕⊕⊝⊝  Low^a,c	There was high risk of bias for selection; however this was a cluster‐randomised trial with a large confidence interval
Early neonatal mortality ‐ Both one‐to‐one and group counselling	RR 0.78  (0.65 to 0.93)	180,690  (5 RCTs)	⊕⊕⊕⊝  Moderate^b	There was considerable heterogeneity (I² = 85%), and not all confidence intervals overlapped
Late neonatal mortality ‐ Group counselling	RR 0.50  (0.31 to 0.81)	118,239  (7 RCTs)	⊕⊝⊝⊝  Very low^a,b,c	High risk of bias was present for blinding across studies; however these were cluster‐randomised trials, and this is justifiable. There was considerable heterogeneity (I² = 91%), and all studies are consistent in their direction; however not all of them overlap. The confidence interval is wide; however the sample size is sufficient
Late neonatal mortality ‐ Both group and one‐to‐one counselling	RR 0.72  (0.57 to 0.91)	68,404  (4 RCTs)	⊕⊕⊕⊝  Moderate^c	Confidence interval is wide
Perinatal mortality ‐ One‐to‐one counselling	RR 0.88  (0.57 to 1.34)	23,829  (2 RCTs)	⊕⊝⊝⊝  Very low^a,b,c	One of the studies that had greater weighting had considerable risk of bias that may impact the direction of effect Confidence intervals do not overlap, and the direction of effect is inconsistent Confidence interval include both important benefit or harm and no effect
Perinatal mortality ‐ Group counselling	RR 0.85  (0.77 to 0.94)	156,505  (8 RCTs)	⊕⊕⊝⊝  Low^a,b	The 2 heaviest weighted studies have concerns regarding selection bias Most confidence intervals overlap, and only 1 study (that has less weighting) is going in the opposite direction of effect
Perinatal mortality ‐ Both group and one‐to‐one counselling	RR 0.78  (0.67 to 0.90)	82,279  (5 RCTs)	⊕⊕⊝⊝  Low^b,c	The confidence interval is wide; however the sample size is sufficient. Most confidence intervals overlap; however 1 of the studies is inconsistent in its direction of effect (although it is weighted the least)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).    CI: confidence interval; LMICs: low‐ and middle‐income countries; RCT: randomised controlled trial; RR: risk ratio.
GRADE Working Group grades of evidence.  High certainty: we are very confident that the true effect lies close to that of the estimate of the effect.  Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.  Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.  Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

Community health educational ANC period and PNC period and both periods (subgroup) compared to control in LMICs
Patient or population: developing countries  Setting: low‐middle‐income countries, community  Intervention: community health educational ANC period and PNC period and both periods (subgroup)  Comparison: control
Outcomes	Relative effect  (95% CI)	№ of participants  (studies)	Certainty of the evidence  (GRADE)	Comments
Neonatal mortality ‐ Education provided during ANC period only	RR 0.84  (0.64 to 1.09)	47,849  (3 RCTs)	⊕⊝⊝⊝  Very low^a,b	Studies were rated evenly, and the quality of evidence was downgraded as all studies had biases that were of unclear or high risk that may have affected the outcome. Not all confidence intervals from the studies overlap; there is also inconsistency in direction across studies
Neonatal mortality ‐ Education provided during PNC period only	RR 1.02  (0.84 to 1.24)	172,882  (3 RCTs)	⊕⊕⊝⊝  Low^b	There is inconsistency in the direction of effect
Neonatal mortality ‐ Education provided in both ANC and PNC periods	RR 0.85  (0.76 to 0.96)	332,380  (20 RCTs)	⊕⊝⊝⊝  Very low^a,b	Because the outcome is mortality, blinding is objective and therefore is unlikely to be affected by blinding of outcome assessment Quality of evidence was downgraded 2 points due to inconsistent direction of effect and because not all confidence intervals overlapped
Early neonatal mortality ‐ Education provided during ANC period only	RR 0.64  (0.43 to 0.95)	33,209  (2 RCTs)	⊕⊝⊝⊝  Very low^a,b,c	Quality assessment was downgraded both both studies had reasonable amounts of bias that may have impacted the effect Heterogeneity is high according to I², and there is consistency in the direction of effect, although the larger study's confidence intervals do not overlap the others Confidence interval shows large spread
Early neonatal mortality ‐ Education provided during PNC period only	RR 1.03  (0.94 to 1.12)	111,529  (1 RCT)	⊕⊕⊕⊝  Moderate^c	The confidence interval includes benefit, harm, and no effect
Early neonatal mortality ‐ Education provided during both ANC and PNC periods	RR 0.76  (0.68 to 0.84)	176,850  (12 RCTs)	⊕⊝⊝⊝  Very low^a,b	There was a combination of high and unclear risk of bias regarding blinding; this can be attributed mainly to the cluster‐randomised design. The study with greatest risk of bias concerns had the smallest weighting. Quality of evidence was downgraded 2 points due to inconsistent direction of effect, and not all confidence intervals overlap
Late neonatal mortality ‐ Education provided during ANC period only	RR 0.87  (0.54 to 1.40)	30,952  (1 RCT)	⊕⊝⊝⊝  Very low^a	This was downgraded by 1 level due to possible selection bias
Late neonatal mortality ‐ Education provided during both ANC and PNC periods	RR 0.52  (0.38 to 0.72)	155,691  (10 RCTs)	⊕⊝⊝⊝  Very low^a,b,c	High risk of bias was present for blinding across studies; however studies were cluster‐randomised trials, and this is justifiable I² (88%) was considerably large; however most of the confidence intervals overlap and direction of effect is consistent This was downgraded by 1 level, as there was a sufficient number of events; however the confidence interval was wide
Perinatal mortality ‐ Education provided during PNC only	RR 0.89  (0.78 to 1.02)	60,480  (1 RCT)	⊕⊕⊕⊕  High	There were no concerns regarding certainty assessment; however this is for only 1 study
Perinatal mortality ‐ Education provided during ANC period only	RR 0.90  (0.59 to 1.39)	33,513  (2 RCTs)	⊕⊝⊝⊝  Very low^a,b	Quality assessment was downgraded because both studies had reasonable amounts of bias that may have impacted the effect. Heterogeneity is large; direction of effect shows inconsistency, and overlapping of confidence intervals is minimal
Perinatal mortality ‐ Education provided during both ANC and PNC periods	RR 0.81  (0.72 to 0.91)	168,620  (12 RCTs)	⊕⊕⊝⊝  Low^b	Direction of effect shows inconsistency; there is also inconsistency in confidence intervals overlapping each other
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).    ANC: antenatal care; CI: confidence interval; LMICs: low‐ to middle‐income countries; PNC: post‐natal care; RCT: randomised controlled trial; RR: risk ratio.
GRADE Working Group grades of evidence.  High certainty: we are very confident that the true effect lies close to that of the estimate of the effect.  Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.  Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.  Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Neonatal mortality	26	553111	Risk Ratio (Random, 95% CI)	0.87 [0.78, 0.96]
2 Early neonatal mortality	15	321588	Risk Ratio (Random, 95% CI)	0.74 [0.66, 0.84]
3 Late neonatal mortality	11	186643	Risk Ratio (Random, 95% CI)	0.54 [0.40, 0.74]
4 Perinatal mortality	15	262613	Risk Ratio (Random, 95% CI)	0.83 [0.75, 0.91]
5 Neonatal infection	2	42043	Risk Ratio (Random, 95% CI)	0.88 [0.72, 1.08]
6 Any antenatal care	18	307528	Risk Ratio (Random, 95% CI)	1.16 [1.11, 1.22]
6.1 Any ANC	7	65811	Risk Ratio (Random, 95% CI)	1.19 [1.07, 1.33]
6.2 ≥ 1 ANC	5	67050	Risk Ratio (Random, 95% CI)	1.17 [1.04, 1.31]
6.3 ≥ 3 ANCs	4	48738	Risk Ratio (Random, 95% CI)	1.07 [0.85, 1.35]
6.4 ≥ 4 ANCs	8	124701	Risk Ratio (Random, 95% CI)	1.14 [1.01, 1.28]
6.5 > 5 ANCs	1	1228	Risk Ratio (Random, 95% CI)	1.31 [1.16, 1.46]
7 Use of any method of contraception	3	22237	Risk Ratio (Random, 95% CI)	1.10 [0.86, 1.41]
8 Skilled attendance at delivery	10	117870	Risk Ratio (Random, 95% CI)	1.09 [0.94, 1.25]
9 Delivery attended by unskilled or semi‐skilled birth attendant	3	40456	Risk Ratio (Random, 95% CI)	1.02 [0.70, 1.49]
10 Use of clean delivery kit	2	17087	Risk Ratio (Random, 95% CI)	4.44 [0.71, 27.76]
11 Care‐seeking for neonatal illness	7	46154	Risk Ratio (Random, 95% CI)	1.11 [0.97, 1.27]
12 Use of colostrum	5	28631	Risk Ratio (Random, 95% CI)	1.16 [0.83, 1.61]
13 Timely initiation of breastfeeding	19	126375	Risk Ratio (Random, 95% CI)	1.56 [1.37, 1.77]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Neonatal mortality	22	497258	Risk Ratio (Random, 95% CI)	0.88 [0.79, 0.98]
2 Early neonatal mortality	11	264672	Risk Ratio (Random, 95% CI)	0.71 [0.62, 0.82]
3 Late neonatal mortality	9	150876	Risk Ratio (Random, 95% CI)	0.51 [0.36, 0.72]
4 Perinatal mortality	12	224107	Risk Ratio (Random, 95% CI)	0.84 [0.75, 0.94]

Methods	Randomised community intervention trial conducted in Masindi and Kiryandongo districts, Uganda
Participants	Pregnant women were recruited during their first antenatal visit; in total 1644 were enrolled (control = 893, intervention = 751) 16 health centres were randomly and equally assigned to 1 of 2 arms: the intervention arm or the control arm 48 Village Health Teams (VHTs) were selected for training
Interventions	Intervention arm: received a package consisting of (1) VHT making home visits to provide educational messages about maternal and newborn care, and (2) VHT members equipped with mobile phones to make unlimited phone call consultations with professional health workers for further clarification and advice Control arm: women in control group had access to group education that was routinely offered by health centres. The control group did not receive follow‐up visits by VHTs, and VHTs in control groups did not receive mobile devices
Outcomes	Primary outcomes: health facility delivery (compared to home delivery) Secondary outcomes: antenatal consultations, birth preparedness, newborn care practices (cord care, thermal care, initiation of exclusive breastfeeding, care‐seeking for newborn illness)
Notes	The funding agency (Directorate General for Development Cooperation of Belgium) had no role in study design, data management, or manuscript writing
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "random allocation was conducted by writing the names of HCs, VHTs or villages on small pieces of papers which were then folded to conceal the names. Two persons, each representing a study arm, but not associated to the study were asked to randomly pick folded papers from the pool of names. The process was repeated until allocation was completed for HCs, villages and VHTs" Comment: probably done
Allocation concealment (selection bias)	Low risk	Quote: "random allocation was conducted by writing the names of HCs, VHTs or villages on small pieces of papers which were then folded to conceal the names" Comment: probably done
Blinding of participants and personnel (performance bias)   All outcomes	Low risk	Quote: "blinding was not necessary since randomisation was at the level of Health Centres" Comment: blinding was not possible due to the nature of the intervention. Probably no
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Comment: insufficient information to make any judgement
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Comment: attrition was 11.2% in the intervention arm and 9.7% in the control arm Reasons for attrition were similar across the 2 arms
Selective reporting (reporting bias)	Low risk	Comment: this is a registered trial, and the study has reported all outcomes mentioned in the protocol
Other bias	High risk	Quote: "the control arm had four HCs of level II and four HCs of level III. The intervention arm had two HCs of level II and six HCs of level III. Therefore, pre‐intervention differences between control and intervention arms can be explained by variation in the distribution of HC levels in each arm"; "the difference in utilisation between HCs of level II and III explains why there were significantly more women recruited in HCs of level II among the control group compared to HCs of level II in the intervention group. The same argument can be extended to explain significant difference in antenatal care attendance in control arm compared to the intervention arm. These differences could have affected our results" Comment: probably not done

Methods	Cluster‐randomised controlled trial conducted in Bangladesh
Participants	Target population included women aged 15 to 49 years residing in the project area who gave birth during the study period, adolescents, and mothers‐in‐law. 18 clusters in 3 districts were randomly assigned to either intervention or control. There were 36,139 births in an overall population of 478,611 during 3 years
Interventions	Intervention clusters: a facilitator convened 18 groups every month to support participatory action and learning for women, and to develop and implement strategies to address maternal and neonatal health problems. Other intervention involved training traditional birth attendants in bag‐valve‐mask resuscitation of neonates with symptoms of birth asphyxia Facilitators received 5 training sessions that covered participatory modes of communication and maternal and neonatal health issues There were 3162 live births in the intervention arm Control group: received the usual care provided in the area (n = 3227 total births, n = 3069 live births)
Outcomes	Primary outcomes: neonatal mortality rate, maternal deaths, stillbirths Secondary outcomes: uptake of antenatal and delivery services, home care practices during and after delivery, infant mortality, health care‐seeking behaviour, perinatal mortality
Notes	In all clusters (intervention and control), health services were strengthened and traditional birth attendants were trained Sponsors of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "unions were randomly allocated to either intervention or control groups by district in the presence of four project staff and two external individuals" Comment: insufficient information to permit judgement
Allocation concealment (selection bias)	Low risk	Comment: since this was a cluster‐randomised trial, allocation concealment should not be an issue, as in this design, all clusters are randomised at once Comment: probably done
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Quote: "neither the study investigators nor the participants were masked to group allocation" Comment: probably not done
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Comment: insufficient information to permit judgment
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Exclusion (0%) and attrition (14.3%) were reported along with reasons
Selective reporting (reporting bias)	Low risk	Comment: this is a registered trial, and this study has reported all outcomes mentioned in the protocol
Other bias	High risk	Quote: "the purposive selection of districts and upazilas, and the stratification of sampling, together with the restricted number of clusters, might have limited the effectiveness of randomisation procedures" Comment: probably not done

Methods	Quasi‐experimental study with 1 intervention district and 1 comparison (control) district of rural Uttar Pradesh, India
Participants	Each district had 15 rural blocks; 9 of these made up the intervention group and 8 made up the comparison district. One sector from each block was selected for baseline and end of line surveys. Women who had experienced a live birth in the reference period were included in the analysis. At baseline, a total of 6196 women were included in the comparison group compared to 8756 in the intervention group. For the end line survey, 6014 women were included in the comparison and 7812 in the intervention district Community‐based workers: home visitation during neonatal and postnatal periods
Interventions	Intervention: community‐based workers (CBWs) conducted home visits during antenatal and postnatal periods as the main strategy for behaviour change communication about healthy maternal and newborn care practices, including recognition of danger signs and care‐seeking. CBWs also promoted recognition of maternal and newborn complications during pregnancy, during delivery, or postpartum Control: received the standard government programme Baseline and end line surveys were conducted to determine effects of the intervention regarding rates of programme coverage, maternal and newborn care practices, and healthcare utilisation
Outcomes	Antenatal care: antenatal home visit, birth preparation, emergency preparation, tetanus toxoid immunisation, iron/folic acid supplementation, antenatal care visit Delivery care: medically trained birth attendant Newborn care: clean cord care, newborn dried and wrapped, newborn bath delayed, immediate breastfeeding, postnatal home visit, newborn check‐up
Notes	Funding: project was funded by USAID, India Mission, through Global Research Activity Award # HRNA‐00‐96‐90006‐00 to the Johns Hopkins Bloomberg School of Public Health
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "each district had 15 rural blocks; 9 blocks in the intervention district and 8 in the comparison district were randomly selected using a computer program. WIthin each block, one sector,..., was randomly selected" Comment: probably done. However the intervention group had a significantly larger sample size (8756) compared to the control group (6196)
Allocation concealment (selection bias)	Unclear risk	Comment: insufficient information to permit judgement
Blinding of participants and personnel (performance bias)   All outcomes	Low risk	Comment: since this is a cluster‐randomised trial, allocation concealment should not be an issue, as in this design, all clusters are randomised at once
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Comment: insufficient information to permit judgement
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	Comment: insufficient information to permit judgement
Selective reporting (reporting bias)	Unclear risk	Comment: insufficient information to permit judgement
Other bias	Low risk	Study seems to be free from other biases

Methods	Randomised controlled trial conducted in Damascus, Syria
Participants	Women who delivered a healthy newborn whether by vaginal delivery or Caesarean section, who lived within 30 km from the hospital, and who were available for follow‐up for the coming 6 months were included (n = 876)
Interventions	Intervention groups: the intervention consisted of home visits aimed to examine, follow up, educate, support, and counsel women who had recently given birth. Registered midwives undertook 5 days of special training and implemented postpartum home visits. Training included a review of postnatal care, the role of home visits, the content of each visit, including the physical examination as well as educational messages, and communication skills. Two intervention groups received 4 postnatal home visits (group A = 285) or 1 visit (group B = 294). Women in group B received 1 home visit on day 3, similar in content to visits to group A Control: women (group C = 297) received the current standard of care in Syria, that is, no visit following hospital discharge
Outcomes	Primary outcomes: maternal postpartum morbidities; postnatal care uptake; contraceptive uptake and type; infant morbidities; infant immunisation according to the national schedule at 3 months; infant feeding, namely, exclusive breastfeeding during the first 4 months of life Secondary outcomes: women's perceptions of their health and their impressions about the home visit and perceptions of its quality
Notes	We included data for group B vs C in this analysis Funding: study was supported by the American University of Beirut Award (Regional Changing Childbirth Research Program at Faculty of Health Sciences supported by Wellcome Trust)
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "randomisation was in blocks of 7 where a caseload of 21 eligible deliveries per day was assumed" Comment: insufficient information to permit judgement
Allocation concealment (selection bias)	Low risk	Quote: "the method that was used to implement the random allocation relied on numbered opaque and sealed envelopes" Comment: probably done
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Quote: "the midwives who carried out the home visits were not blinded to the group assignments, but the outcome assessors were" Comment: probably not done
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Quote: "the midwives who carried out the home visits were not blinded to the group assignments, but the outcome assessors were" Comment: probably done
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Exclusion (1.2%) was given and reasons mentioned
Selective reporting (reporting bias)	Unclear risk	Comment: insufficient information to make an assessment, as study protocol is not available
Other bias	Low risk	Comment: study seems to be free from other biases

Methods	Cluster‐randomised trial conducted in catchment areas of 18 primary health centres in Haryana, India
Participants	Women including pregnant women, neonates, and children (60,480 births)
Interventions	Intervention arm (n = 29,667 births): IMNCI training for AWWs, ANMs, TBAs, and private practitioners These sessions covered neonatal conditions requiring referral, pre‐referral treatment, problems that can be managed at home, and components of essential newborn care. Management of diarrhoea, pneumonia, conditions requiring referral, and pre‐referral treatment, as well as appropriate complementary feeding practices for older children, was also covered. Improved availability of medicines was achieved by establishing medicine stores with the VLCs, which were replenished every 3 months by a study supply officer. Supervision of health workers and community health workers was strengthened in the intervention areas. The study recruited 2 supervisors per 30,000 population in consultation with the local government; they conducted monthly visits to ASHAs and AWWs, observed women's group meetings, and attended monthly government review meetings Control arm (n = 30,813): AWWs and ANMs were in place in control areas and continued to provide the usual services. However, they were not trained to conduct IMNCI activities such as home visits for newborns and community management of sick infants and children. Management of sick infants and children remained in the hands of private providers and physicians at government facilities
Outcomes	Primary outcomes: perinatal deaths, neonatal deaths, postneonatal deaths, infant deaths Secondary outcomes: danger signs of severe illness, local infections, diarrhoea, general danger signs, pneumonia
Notes	Funding: study was funded by the World Health Organization, Geneva (through an umbrella grant from USAID); the United Nations Children’s Fund, New Delhi; and the GLOBVAC Program of the Research Council of Norway through grant No. 183722. Individual scientists at WHO and Unicef contributed importantly to planning, analysis, and reporting of this study. However, the central bodies of these agencies and the Research Council of Norway had no influence on how data were collected, analysed, or presented
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "an independent epidemiologist generated 10 stratified randomisation schemes to allocate the clusters to intervention or control groups"; "We selected one of the remaining seven allocation schemes by a computer generated random number" Comment: probably done
Allocation concealment (selection bias)	Low risk	Comment: since this is a cluster trial, allocation concealment should not be an issue, as in this design, as all clusters are randomised
Blinding of participants and personnel (performance bias)   All outcomes	Unclear risk	Comment: insufficient information to permit judgement
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Quote: "the surveillance team was not told the intervention status of the community they were visiting"
Incomplete outcome data (attrition bias)   All outcomes	Low risk	12.3% were lost to follow‐up in intervention clusters, and 14% were lost to follow‐up in control clusters
Selective reporting (reporting bias)	Low risk	Comment: this is a registered trial, and the study has reported all outcomes mentioned in the protocol
Other bias	Low risk	Study seems to be free from other biases

Methods	Pilot clustered‐randomised trial from Hala and Matiari subdistricts of Sind, Pakistan
Participants	Target population consisted of pregnant women. 24 village clusters were identified at a primary care facility. Out of those, 8 clusters were randomly selected for this pilot study. 4 districts chosen to receive intervention were matched with 4 control clusters for population size and birth and neonatal mortalities rates (5134 births)
Interventions	Intervention: promotion of maternal nutrition and rest, early breastfeeding (within first hour), colostrum administration (avoidance of pre‐lacteal feeds), thermoregulation, home care for low birth weight infants, treatment of pneumonia with oral trimethoprim‐sulphamethoxazole (TMP‐SMX), recognition of danger signs, training in group counselling and communication strategies (n = 2672 births) Control: routine care (n = 2462 births) The following activities were conducted in both intervention and control groups: promotion of antenatal care, iron and folate use during pregnancy, immediate newborn care, cord care, promotion of exclusive breastfeeding
Outcomes	Primary outcomes: stillbirths, early neonatal deaths, late neonatal deaths, total neonatal deaths, perinatal deaths
Notes	Local health workers (LHWs) were encouraged to visit all pregnant women twice during pregnancy, within 24 hours of birth, and 4 times in the first postnatal month, and were encouraged to link up with local Dais (traditional birth attendants). LHWs were supported by the creation of voluntary community health committee, which helped in conducting community education group sessions Funding: the Hala project is supported by a collaborative grant from WHO and the Saving Newborn Lives (SNL) programme of Save the Children (USA), funded by the Bill & Melinda Gates Foundation
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Quote: "eight clusters were randomly selected for this pilot study"; "subsequently, the four clusters chosen to receive the intervention were matched with four control clusters" Comment: probably not done
Allocation concealment (selection bias)	Low risk	Comment: since this is a cluster‐randomised trial, allocation concealment should not be an issue, as in this design, all clusters are randomised at once
Blinding of participants and personnel (performance bias)   All outcomes	Unclear risk	Comment: insufficient information to permit judgement
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Quote: "cross‐sectional surveys of all households were conducted by a separate team in both the middle (June–July 2004) and at the end (August –September 2005) of the pilot study to collect data on births, deaths and care‐seeking behaviour in the preceding 12 months" Comment: done
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	Comment: insufficient information to permit judgement
Selective reporting (reporting bias)	Low risk	Comment: study reported all outcomes outlined in the protocol
Other bias	Low risk	Comment: study seems to be free from other biases

Methods	Cluster‐randomised trial of community‐based interventions conducted in 2 Talukas (subdistricts) of Naushero Feroze, rural districts of Sind, Pakistan
Participants	Pregnant women. 16 clusters were assigned to intervention and control groups. Total number of births in intervention group was 12,391, and in control group 11,443 (n = 23,834)
Interventions	There were 2 study groups  Intervention arm (n = 12,391): local health workers (LHWs) = along with basic training (for control group), they received additional training on recognition of high‐risk pregnancies and referral and management of birth asphyxia, serious bacterial infections, and low birth weight infants  Traditional birth attendants (TBAs) = along with basic training (for control group), they received additional training on promotion of LHW attendance at births and resuscitation (mouth‐to‐mouth) of newborns  Control arm (n = 11,443): trained LHWs on community mobilisation through building support groups, promotion of use of clean delivery kits, recognition of neonatal illness, and referral for care; TBAs were linked with LHWs; they were trained on promotion and use of clean delivery kits
Outcomes	Primary outcomes: neonatal mortality rates, perinatal mortality rates  Secondary outcomes: birth asphyxia‐related neonatal mortality rates, neonatal mortality rates in low birth weight infants, neonatal mortality rates due to sepsis
Notes	To create awareness in the community and at the household level in control and intervention clusters, female and male support groups (health committees) were formed/strengthened. The LHW formed a female health committee, and male activists formed male health committees, in the LHW catchment area. Meetings of both groups were arranged with the assistance of the community health committee, and LHWs met on a monthly basis for dissemination of health messages and education related to maternal and newborn health and problems. Separate community group educational sessions for mothers, mothers‐in‐law, married women especially with pregnancy, fathers, fathers‐in‐law were conducted for health education of communities through support groups in the LHW catchment area using educational material as flip charts on antenatal care, identification of danger signs related to pregnancy, recognition of simple risk factors for high‐risk pregnancies and births (these include severe maternal malnutrition, illness, short stature, previous perinatal deaths, etc.), and birth preparedness (transport, money, skilled birth attendant, facility) Funding: WHO; Saving Newborn Lives Program of Save the Children USA, funded by the Bill & Melinda Gates Foundation
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "twenty‐six such clusters with available LHWs were identified in the district, 8 of which were involved in the pilot study. Two further clusters were excluded as they had very few LHWs. The full‐cluster RCT was thus implemented in the remaining 16 clusters"; "used restricted, stratified randomisation to allocate clusters to the intervention and control arms (21). Three strata (comprising 2, 6 and 8 clusters) were identified based on their size and the number of LHWs per 1000 population. We identified 126 random allocations which resulted in similar population sizes in the 2 arms.........From this list of "balanced" allocations we selected one scheme at random"; "From this list of balanced allocations, we selected one scheme using a computer generated random number"  Comment: probably done
Allocation concealment (selection bias)	Low risk	Comment: since this is a cluster‐randomised trial, allocation concealment should not be an issue, as in this design, all clusters are randomised at once
Blinding of participants and personnel (performance bias)   All outcomes	Unclear risk	Comment: insufficient information to permit judgement
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Quote: "data collectors were masked to cluster allocation; those analysing data were not" Comment: data collectors were independent of implementers
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Attrition was 12.4% in intervention clusters and 10.8% in control clusters, and reasons for exclusion were mentioned
Selective reporting (reporting bias)	Low risk	Comment: study reported all outcomes outlined in the protocol
Other bias	Low risk	Comment: study seems to be free from other biases

Methods	Two‐by‐two factorial cluster‐randomised controlled trial evaluating community‐ and facility‐based interventions to reduce deaths in 3 districts of Malawi
Participants	Randomly sampled approximately 4000 people per cluster and set up community surveillance to track pregnancies, births, and deaths of consenting women Baseline (n = 14,576 births) Intervention (n = 20,576 births)
Interventions	Community mobilisation: 81 volunteer facilitators each formed 9 women's groups, which followed an "action cycle" to identify and prioritise maternal and neonatal health problems. Fifty per cent of groups had maternal and neonatal task forces added to them to enhance antenatal coverage, maternal/neonatal health knowledge, and facility delivery Quality improvement: facility level Control: 17 clusters (out of 62) were assigned to control, where these clusters underwent no interventions
Outcomes	Primary outcomes: maternal, neonatal and perinatal mortality Secondary outcomes: % of deliveries at a health facility, % of maternal deaths subjected to maternal death audit, case‐fatality rates, practice of signal obstetrical care at community level, number of women's groups mobilised annually, % of pregnant women attending women's groups
Notes	Funding: project was funded by The Health Foundation, London, UK. Project funders commissioned the randomised controlled trial to evaluate the 2 interventions but had no direct input into the design of the study, data collection, nor data analysis
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "clusters were health centre catchment areas assigned using stratified computer generated randomisation" (random number sequence generated in Strata 7) Comment: probably done
Allocation concealment (selection bias)	Low risk	Quote: "to ensure concealment of intervention allocation, identification numbers were assigned for each cluster and a random number generated for each" Comment: probably done
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Quote: "neither participants nor those administering the interventions were blinded to group assignment" Comment: probably not done
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Comment: insufficient information to permit judgement
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Quote: "during the baseline period, the recorded loss to follow‐up to birth outcomes was 19% and during the intervention it was 29%, with higher rates in later months. Given that observed birth rates in the study matched those expected from the crude birth rate 1.27 to within 3%, and that in‐migration probably broadly matched out‐migration, many of the pregnancies recorded by KIs as ‘lost to follow‐up’ may have been recorded as pregnancies by mistake and true loss‐to‐follow‐up was probably much lower; there was little difference in loss‐to‐follow‐up between arms" Comment: probably not
Selective reporting (reporting bias)	Low risk	Comment: this is a registered trial, and this study has reported all outcomes mentioned in the protocol
Other bias	Low risk	Study seems to be free from other biases

Methods	Two‐arm cluster‐randomised trial in Rural Ethiopia, evaluating the impact of making newborn infection management available at health posts when referral to health centres were not possible
Participants	22 geographical clusters (11 in intervention and 11 in control) Number of women of reproductive age (15 to 49) interviewed: comparison (control), n = 58,497; intervention, n = 56,733
Interventions	Trained 3500 female volunteers on what to do during house visits, counselling families about importance of antenatal care, danger signs for women and newborns that should prompt care‐seeking, birth preparedness, clean delivery, and healthy newborn practices that prevent infection. 270 HEWs were also trained on home visits, volunteer support, iCCM, and case management. HEWs in the intervention arm also received training on the treatment algorithm, administration of medicine, and injection safety Control: HEWs in the control group did not receive additional training related to the algorithm, administration of medicine, injection safety, or treatment of PSBI at health posts. However, home visits, referral of newborns to health centres, and project officers were available at health posts for both arms
Outcomes	Primary outcome: post day 1 neonatal mortality (2 to 27 days after birth) Secondary outcomes: socioeconomic status, knowledge, practices, and care‐seeking from women who had delivered in previous 60 days. Home visit coverage by volunteers and HEWs, PSBI cases, HEW performance, PSBI signs, mortality impact
Notes	Funding: funding agency had no role in study design, data collection, data analysis and interpretation, writing of the report, or the decision to submit for publication
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Quote: "clusters were randomised 1:1 stratified on region and using restriction to ensure arms were balanced in population size, annual number of births, baseline neonatal mortality rate, and proportion of HEWs resident in their village" Comment: probably not done
Allocation concealment (selection bias)	Low risk	Comment: since this is a cluster‐randomised trial, allocation concealment should not be an issue, as in this design, all clusters are randomised at once
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Comment: "allocation was not masked"; "training was an element of the intervention, therefore blinding of female volunteers and Health Extension Workers was not possible" Comment: not done
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Quote: "although survey teams were blinded to minimize interviewer bias" Comment: probably done
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Comment: attrition was similar across groups and reasons were provided
Selective reporting (reporting bias)	Low risk	Comment: this is a registered trial, and the study has reported all outcomes mentioned in the protocol
Other bias	Low risk	Study seems to be free from other biases

Methods	Cluster‐randomised controlled trial in Bangladesh to assess the effect of participatory women's group intervention with higher population coverage on neonatal mortality
Participants	Women who were permanently residing in 18 unions, in 3 districts, and accounting for 19,301 births during the final 24 months of the intervention. In intervention areas, 648 new women's groups were formed in addition to 162 women's groups set up as a part of an earlier trial (combined women's groups: n = 810). All unions, regardless of allocation, received health system‐strengthening initiatives
Interventions	Women's group utilising participatory learning and action cycle, where they prioritise issues that affected maternal and neonatal health, and design and implement strategies to address said issues Control: both intervention and control clusters received health system‐strengthening activities. A small number of women from control areas participated in the women's group meetings; however contamination is fairly limited due to the geographical location of the clusters
Outcomes	Primary outcome: neonatal mortality rates Secondary outcomes: facility deliveries, birth attendant home delivery practices (e.g. washed hands with soap), thermal care of newborns, early infant‐feeding practices, health service utilisation (antepartum, intrapartum, postpartum)
Notes	Funding: implementation and evaluation of women’s groups was funded by a Big Lottery Fund International Strategic Grant. This study was supported with funds from a Wellcome Trust Strategic Award (085417ma /Z/08/Z). Sponsors did not participate in design and conduct of the study; in collection, analysis, and interpretation of data; or in preparation, review, or approval of the manuscript
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "6 unions (clusters) in each district (stratum) were randomly allocated to either intervention or control by blindly pulling pieces of paper, each representing 1 union, from a bottle. The allocation sequence had been decided on before drawing the papers" Comment: probably done
Allocation concealment (selection bias)	Low risk	Quote: "the nature of the intervention means that allocation was not masked" Comment: since this is a cluster‐randomised trial, allocation concealment should not be an issue, as in this design, all clusters are randomised at once
Blinding of participants and personnel (performance bias)   All outcomes	Low risk	Comment: blinding of participants not possible due to nature of study
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Quote: "the board, as well as the implementation and in country monitoring and evaluation teams, were blind to the allocation arm on both occasions" Comment: probably yes
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Attrition was 1% and reasons were reported
Selective reporting (reporting bias)	Low risk	Comment: this is a registered trial, and the study has reported all outcomes mentioned in the protocol
Other bias	Low risk	Comment: study seems to be free from other biases

Methods	Community randomised trial in an urban informal settlement, KwaZulu‐Natal, South Africa
Participants	Participants included all consenting pregnant women < 17 years of age and their newborns residing in clusters. In the intervention (n = 15) and control (n = 15) clusters (total = 30), community health workers (CHWs) identified all pregnant women living in their cluster. CHWs were women who resided in the neighbourhoods and were literate. Data were available for 1659 intervention and 1902 mother‐infant pairs at 12 weeks postpartum
Interventions	Intervention: CHWs were trained and implemented the Goodstart intervention package through a structured home visiting programme consistent with existing programmes to "prevent mother‐to‐child transmission" of HIV, Integated Management of Childhood Illness, lactation counselling, and newborn care guidelines. Motivational interviewing techniques were used for breastfeeding counselling ‐ to assist women to engage in social supports (e.g. family members, community resources). Women in the intervention arm were scheduled to receive 7 home‐based visits (1 within 48 hours of delivery, during days 3 to 4 and 10 to 14, and during weeks 3 to 4 and 7 to 8) (n = 1659) Control: women were provided with information and support by CHWs for accessing social welfare grants and also conducted 3 home‐based visits (1 during pregnancy, 2 during weeks 4‐6 and 10‐12 post delivery). CHWs provided home‐based education and support for families to understand how to obtain the grant, the necessary papers and procedures. (n = 1,902)
Outcomes	Primary outcomes: exclusive and appropriate breastfeeding at 12 weeks postnatally, HIV‐free infant survival Secondary outcomes: uptake of postnatal clinic visit within 7 days of life, coverage of care and behavioural indicators, post‐intervention levels of maternal depressed mood at 12 weeks postpartum
Notes	Funding: funded through a grant to Save the Children’s Saving Newborn Lives programme, from the Bill & Melinda Gates Foundation. No funding bodies had any role in study design, data collection and analysis, or decision to publish, nor in preparation of the manuscript
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "...neither stratification nor matching was performed. We used simple computer‐generated randomisation, with clusters assigned in a 1:1 allocation ratio"  Comment: probably done
Allocation concealment (selection bias)	Low risk	Comment: this study was a cluster‐randomised effectiveness trial; allocation concealment should not have been an issue due to the randomisation process
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Comment: could not blind participants or personnel to the intervention due to training and receipt of treatment
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Comment: insufficient information to permit judgement
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Comment: there was loss to follow‐up in 2 periods of the study: first in pregnant women (attrition rate 2.1%) and then following birth (attrition rate 9.5%)
Selective reporting (reporting bias)	Low risk	Comment: this is not a registered trial, but outcomes mentioned in the methods are reported in the results
Other bias	Low risk	Study seems to be free from other biases

Methods	Cluster‐randomised controlled trial in Pakistan
Participants	All pregnant women were eligible for inclusion. Seven subdistricts (talukas) were randomly assigned to intervention (n = 3) or control (n = 4). Total numbers of participants were 10,114 and 9443 in intervention and control arms, respectively (total = 19,557). Completed follow‐up data of 42 days resulted in 10,093 women in intervention and 9432 in control (total = 19,525)
Interventions	Intervention: traditional birth attendants (TBAs) were trained by obstetricians and female paramedics and issued disposable delivery kits. Lady Health Workers linked TBAs to established services and documented processes and outcomes. Obstetrical teams provided outreach clinics for antenatal care. TBAs were asked to visit each woman at least 3 times during pregnancy for danger signs Control: women received standard care by Lady Health Workers; TBAs did not receive additional training nor disposable delivery kits
Outcomes	Primary outcomes: perinatal and maternal mortality Secondary outcomes: major complications of pregnancy (haemorrhage, obstructed labor, puerperal sepsis, eclampsia, abortion), referral by TBA for emergency obstetrical care, types and places of delivery, delivery attendant
Notes	Funding: supported by a grant from Family Health Project of the Sindh government’s health department (for capital costs) and by the University of Birmingham (for data entry). There is no statement relating to involvement of these funding parties in the design and conduct of the study
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "with a simple cluster‐randomisation sampling scheme, and with a computer‐generated procedure, Larkana’s seven talukas were allocated to intervention or control groups" Comment: probably done
Allocation concealment (selection bias)	Low risk	Comment: since this is a cluster‐randomised trial, allocation concealment should not be an issue, as in this design, all clusters are randomised at once
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Quote: "although the traditional birth attendants and Lady Health Workers could not be blinded to the intervention, observer bias is unlikely to have affected the reporting of the primary outcomes of perinatal and maternal mortality" Comment: probably not done
Blinding of outcome assessment (detection bias)   All outcomes	High risk	Quote: "Lady Health Workers who recorded outcomes could not be blinded to the intervention status of the women but were not made aware of the main study objective or the outcome measures for the planned comparison" Comment: probably not done
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Comment: acknowledged loss to follow‐up but no reasons were given; attrition rate was 2% in the intervention arm and 1.2% in the control arm, and reasons were not provided
Selective reporting (reporting bias)	Low risk	Comment: not a registered trial; however, outcomes reported in the methods section were reported in the results section
Other bias	Low risk	Study seems to be free from other biases

Methods	Cluster‐randomised trial (NEWHINTs) undertaken in Ghana
Participants	49 zones were randomly assigned to intervention and 49 to control. From the control arm, n = 10,096 pregnant women were recruited, and n = 9885 pregnant women were recruited from the NEWHINTs zones (intervention). Data analysis included all pregnancies that result in live birth or stillbirth (total of 16,329 deliveries during trial period)
Interventions	Community‐based surveillance volunteers (CBSVs) in intervention zones were trained to identify pregnant women in their community and to counsel and solve problems related to key essential newborn care behaviours. CBSVs were trained to make 2 home visits during pregnancy, and 3 in the first week of life, to promote newborn care practices, to weigh and assess babies for danger signs, and to refer if necessary (n = 8035 eligible deliveries) Control: routine maternal and child health care, including antenatal clinics, access to free fertility delivery, postpartum checkups, infant welfare clinics, and routine CBSV activities.(n = 8294 eligible deliveries)
Outcomes	Primary outcome: all‐cause neonatal mortality rate (all deaths that happen in the first 28 days of life), % of mothers practising the Newhints recommended behaviours Secondary outcomes: age‐specific and cause‐specific neonatal mortality rates
Notes	Funding: WHO; Save the Children’s Saving Newborn Lives Programme from the Bill & Melinda Gates Foundation; and UK Department for International Development provided funding. Funders had no role in data gathering, data analysis, or writing of the report
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐generated restricted randomisation using stratified sampling (1‐to‐1 ratio) Comment: probably done
Allocation concealment (selection bias)	Low risk	Comment: since this is a cluster‐randomised trial, allocation concealment should not be an issue, as in this design, all clusters are randomised at once
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Comment: could not blind community‐based surveillance volunteers as they were trained
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Comment: insufficient information to permit judgement
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Comment: attrition (4.9%) and reasons were reported
Selective reporting (reporting bias)	Low risk	Comment: this is a registered trial, and this study has reported all outcomes mentioned in the protocol
Other bias	High risk	Comment: some pregnant women swapped groups (intervention or control) during the trial

Methods	Three‐arm cluster‐randomised trial done in Shivgarh, a rural area in Uttar Pradesh, India
Participants	Pregnant women, mothers‐in‐law, other female members who played supportive role, male members including fathers‐in‐law and husbands, family's immediate support group including neighbours and relatives who influenced family behaviours and helped with delivery (n = 3762 births)
Interventions	Intervention 1 (n = 1537): essential newborn care only Control (n = 1115): usual services of governmental and non‐governmental organisations in the area Essential preventive newborn care includes home visits and group meetings of stakeholders about birth preparedness, hygienic delivery, and immediate newborn care, including clean umbilical cord, and skin care and thermal care including skin‐to‐skin care, breastfeeding, and care‐seeking from trained providers. The second intervention group received essential newborn care plus use of a liquid crystal sticker that indicates hypothermia by changing colour. All messages were designed to promote newborn care practices to align with existing cultural values and traditions of the local area. Saksham Sahayak (community health worker) were recruited and received classroom‐based and apprenticeship‐based field training on knowledge, attitudes, and practices about essential newborn care, behaviour change management, and trust building. Saksham Sahayaks first engaged with community stakeholders in community meetings to seek their approval, sensitise them toward the importance of their role in newborn survival, encourage shared learning, and create a supportive environment in community. Pregnant women were identified by Saksham Sahayak, self‐reporting of pregnant women, and other community health workers. Intervention was delivered by 2 antenatal visits (60 days and 30 days before expected delivery) and 2 postnatal visits (first within 24 hours of birth and second on day 3). No treatment was offered to sick neonates; however, they were advised to seek care at nearest health facility
Outcomes	Primary outcomes: miscarriages, stillbirths, live births, neonatal deaths Secondary outcomes: other pregnancy outcomes, neonatal care outcomes
Notes	We included in this analysis data for Intervention 1 (i.e. essential newborn care) vs control Funding: study was funded by the United States Agency for International Development, Delhi Mission, and the Saving Newborn Lives program of Save the Children US through a grant from the Bill and Melinda Gates Foundation. There is no statement about the influence these funding sources had on design and cohort of the study
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "stratified cluster randomisation was done at Johns Hopkins University using Stata 7.0 (StataCorp, College Station, TX, USA) to allocate the 39 cluster units randomly to the three study groups, yielding three allocation sequences of 13 clusters each" Comment: probably done
Allocation concealment (selection bias)	Low risk	Comment: since this is a cluster‐randomised trial, allocation concealment should not be an issue, as in this design, all clusters are randomised at once
Blinding of participants and personnel (performance bias)   All outcomes	Unclear risk	Comment: because of the nature of the intervention, blinding did not occur
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Comment: insufficient information to permit judgement
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Attrition (3.9%) was given along with reasons
Selective reporting (reporting bias)	Low risk	Comment: this is a registered trial, and this study has reported all outcomes mentioned in the protocol
Other bias	Low risk	Study seems to be free from other biases

PERMALINK

Community‐based maternal and newborn educational care packages for improving neonatal health and survival in low‐ and middle‐income countries

Zohra S Lassi

Sophie GE Kedzior

Zulfiqar A Bhutta

Abstract

Background

Objectives

Search methods

Selection criteria

Data collection and analysis

Main results

Authors' conclusions

Plain language summary

Summary of findings

Background

Description of the condition

Description of the intervention

How the intervention might work

Why it is important to do this review

Objectives

Methods

Criteria for considering studies for this review

Types of studies

Types of participants

Types of interventions

Intervention

Control

Types of outcome measures

Primary outcomes

Neonatal mortality

Perinatal mortality

Secondary outcomes

Search methods for identification of studies

Electronic searches

Searching other resources

Data collection and analysis

Selection of studies

Data extraction and management

Assessment of risk of bias in included studies

Measures of treatment effect

Dichotomous data

Continuous data

Unit of analysis issues

Cluster‐randomised trials

Dealing with missing data

Assessment of heterogeneity

Assessment of reporting biases

Data synthesis

Quality of evidence

Subgroup analysis and investigation of heterogeneity

Sensitivity analysis

Results

Description of studies

Results of the search

1.

Included studies

Comparison 1

1.1. Analysis.

1.2. Analysis.

1.3. Analysis.

1.4. Analysis.

1.5. Analysis.

1.6. Analysis.

1.7. Analysis.

1.8. Analysis.

1.9. Analysis.

1.10. Analysis.

1.11. Analysis.

1.12. Analysis.

1.13. Analysis.

Comparison 2

2.1. Analysis.

2.2. Analysis.

2.3. Analysis.

2.4. Analysis.

2.5. Analysis.

2.6. Analysis.

2.7. Analysis.

2.8. Analysis.

Methods	Factorial, cluster‐randomised controlled trial in rural Mawali (Mai Mwana trial)
Participants	A cohort of women that of child‐bearing age (10 to 49 years) was defined at a baseline survey (n = 43,719), and those who consented were enrolled to participate. 48 clusters were defined, with populations of approximately 3000 people, and these clusters were randomly allocated to 1 of the 4 study groups. Monitored outcomes of 26,262 births between 2005 and 2009
Interventions	Four study groups were included: (1) women's groups and volunteer peer counselling, (2) only women's groups, (3) only volunteer peer counselling, (4) no intervention Facilatator guided women's groups through a community action cycle, to tackle maternal and child health problems (n = 207 groups) Trained volunteer peer counsellors made home visits at 5 time points during pregnancy and after birth to support breastfeeding and infant care. (n = 75 female volunteer peer counsellors)
Outcomes	Primary outcomes: women's groups: maternal, perinatal, neonatal, and infant mortality rates; peer counsellors: infant mortality rates, exclusive breastfeeding rates Secondary outcomes: maternal and infant morbidity; skilled antenatal, delivery, and postnatal care; tetanus toxoid immunisation; use of malaria prophylaxis; insecticide‐treated bed nets during pregnancy; PMTCT services; infant immunisations; early BF; reduced use of prelacteal feeds; neonatal mortality and infant morbidity rates; caretaker practices; timing of initiation of breastfeeding; family planning (condom use)
Notes	Funding: Saving Newborn Lives, UK Department for International Development, and Wellcome Trust. There is no statement related to involvement of these funding parties in design and conduct of the study
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "clusters of villages, rather than individual villages, were used as the unit of randomisation to further reduce rates of travel across cluster boundaries and the possibility of contamination"; "allocated clusters with a random number sequence generated in Stata"  Comment: probably done
Allocation concealment (selection bias)	Low risk	Quote: "the interventions made masking of allocation impossible at the participant level, allocation was masked for data analysis" Comment: probably done
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Comment: the women knew their intervention allocation due to activities that they undertook
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Comment: insufficient information to permit judgement
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Comment: attrition (14.5%) was given with reasons
Selective reporting (reporting bias)	Low risk	Comment: this is a registered trial, and this study has reported all outcomes mentioned in the protocol
Other bias	Low risk	Study seems to be free from other biases

Methods	Cluster‐randomised trial in Ngorongoro district, Arusha region in Tanzania, involving 16 health units to determine the effectiveness of birth plans
Participants	16 health units were randomly assigned to control or intervention (8 health units per arm). 905 pregnant women at > 24 weeks' gestation (intervention arm, n = 404) were enrolled in the study
Interventions	Intervention arm: introduction and promotion of birth plans by care providers during antenatal care to prepare women and their families for birth and complication readiness. This involved discussions on topics of delivery (planned place, importance of skilled delivery, transport arrangements to delivery site or during an emergency, funding arrangements/emergency care services, identification of possible blood donors, identification of birth companion, support for looking after the household) and strategies for overcoming barriers and recognising danger signs. Providers in the intervention arm were given a birth plan intervention guide with instructions on how to assist women to develop and achieve their birth plans. These providers also received practical and didactic training for 2.5 days Control arm: providers in the control units were trained for a half‐day on how to collect required information (n = 501 women)
Outcomes	Primary outcome: delivery by individual woman in a health unit Secondary outcomes: postnatal care attendance, satisfaction of women and providers with care received and provided, respectively
Notes	Funding: World Health Organization/HRP (UNDP/UNFPA/WHO/World Bank Special Programme of Research, Development and Research Training in Human Reproduction). There is no statement related to involvement of these funding parties in design and conduct of the study
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "the facilities were randomly assigned to the intervention or control arms in a 1:1 ratio using computer‐generated random numbers" Comment: probably done
Allocation concealment (selection bias)	Low risk	Comment: since this is a cluster‐randomised trial, allocation concealment should not be an issue, as in this design, all clusters are randomised at once
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Quote: "the study did not allow blinding of either providers or women who participated in the study to the treatment allocation" Comment: probably not done
Blinding of outcome assessment (detection bias)   All outcomes	High risk	Quote: "the data collectors were people from the same villages as the study participants, which facilitated the identification of women’s location for the follow‐up interview" Comment: probably not done
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	Comment: 0% attrition rate reported, perhaps due to use of health units as clusters
Selective reporting (reporting bias)	Low risk	Comment: not a registered trial, but outcomes mentioned in the methods were reported in the results
Other bias	Low risk	Study seems to be free from other biases

Methods	Randomised trial of community health workers using a postnatal checklist to increase health‐seeking behaviours and knowledge in new mothers in Kiambu County, Kenya
Participants	Women were recruited after a normal delivery just before discharge home. A total of 104 women were enrolled in the study; 41 were assigned to the community health worker (CHW) phone call group, 32 to the CHW home visit, and 31 to standard care
Interventions	Randomised into 1 of 3 groups: 1. Early postnatal care 3 days after delivery provided in‐person with a CHW using a simple checklist 2. Care provided by phone with a CHW using the same checklist 3. Standard care (control) CHWs in the intervention groups were trained to screen for maternal and neonatal danger signs, to deliver targeted postnatal health education, and to refer mothers and their newborns to facility‐based care guided by a checklist. For all referrals, a nurse conducted a follow‐up by phone call to ensure that mother and newborn received appropriate care
Outcomes	Primary outcomes: 3 days postpartum (maternal or newborn complications detected or referred), day 10 and 9‐week survey outcomes (care‐seeking behaviours for mothers and newborns, knowledge and practice of infant care, nutrition, feeding, recognition of danger signs)
Notes	Funding: this research was supported by Grand Challenges Canada, funding award number 0166–03. There is no statement related to involvement of these funding parties in the design and conduct of the study
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "patients were individually randomised prior to enrolment using numeric patient identifiers assigned by Jacaranda Health"; "Randomization was conducted by assignment of each of these unique identifiers to one of the three central treatment groups with equal probability"; "Random assignment of patient identifiers was done using a randomisation sequence generated by the principal investigators with STATA 11" Comment: probably done
Allocation concealment (selection bias)	Low risk	Comment: since this is a cluster‐randomised trial, allocation concealment should not be an issue, as in this design, all clusters are randomised at once
Blinding of participants and personnel (performance bias)   All outcomes	Unclear risk	Comment: insufficient information to permit judgement
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Comment: insufficient information to permit judgement
Incomplete outcome data (attrition bias)   All outcomes	High risk	Quote: "our study also had significant attrition, both in the implementation of the interventions and tracking of respondents at the ten‐day follow‐up survey and nine week follow‐up survey"
Selective reporting (reporting bias)	Low risk	Comment: this is a registered trial, and this study has reported all outcomes mentioned in the protocol
Other bias	High risk	Comment: mention of reporting bias due to self‐reporting method related to cord care and exclusive breastfeeding

Methods	Study of exploratory quasi‐experimental design to determine the impact of community‐based perinatal and newborn preventive care package on perinatal and neonatal mortality in Northern Parkistan
Participants	Overall population for the study district was 283,324, comprising 35,641 households; the intervention area comprised 16,802 households and a population of 137,781; the control area covered 18,659 households with a population of 145,543. A total 165 LHWs and CHWs were trained with additional curriculum on essential newborn care to deliver the intervention package
Interventions	Intervention package: Lady Health Workers and community health workers were trained on causes of perinatal and newborn mortality and risky maternal and newborn care practices, and they were trained to educate families on these topics. They were trained to deliver the intervention package that involves developing awareness related to positive maternal and newborn healthcare practices at a household level, involving the importance of seeking antenatal care, adequate nutrition during pregnancy and lactation, and skilled birth attendance. The intervention package was implemented through monthly household visits, one‐to‐one counselling sessions with pregnant women, and video sessions in communities (group education) (n = 849 total births in intervention areas post intervention)  Control: these areas received the routine services of government and non‐government‐provided services (n = 863 total births in control areas post intervention)
Outcomes	Outcome measures: change in maternal and newborn practices, perinatal and neonatal mortality rates
Notes	Funding: jointly funded by University Research Council, Aga Khan University, Pakistan, and Saving Newborn Lives Initiative, Save the Children, US. There is no statement related to involvement of these funding parties in design and conduct of the study
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Quote: "the overall population of 283,324 comprising 35,641 households located in the study district was allocated to intervention and control areas based on geographical proximity to avoid contamination and manage logistics and undertake the study with limited resources available"; "forty‐eight villages were randomly selected for this phase" Comment: probably not done
Allocation concealment (selection bias)	High risk	Comment: this is a quasi‐experimental design
Blinding of participants and personnel (performance bias)   All outcomes	Unclear risk	Comment: insufficient information to permit judgement
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Comment: insufficient information to permit judgement
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	Attrition of pre‐intervention and post‐intervention populations 4.4%; no explanation for losses to follow‐up Comment: insufficient information to permit judgement
Selective reporting (reporting bias)	Low risk	Comment: this is a registered trial, and this study has reported all outcomes mentioned in the protocol
Other bias	Low risk	Study seems to be free from other biases

Methods	Experimental randomised controlled study evaluating effects of prenatal counselling on compliance with health promotion and teenage pregnancy outcomes in Egypt
Participants	A sample of 86 teenage pregnant women attending the maternal and child health centre, with 43 mothers and their newborns in the control group and the other 43 mothers and newborns in the educational intervention group
Interventions	Educational intervention: a prenatal counselling programme was designed by researchers and 3 to 4 sessions were conducted per pregnant woman. A pre‐test on knowledge about health promotion was conducted before counselling and a post‐test was conducted on all participants 3 months after the pre‐test. Pregnany outcome was assessed in 3 to 6 days at home post delivery (n = 47 pregnant teenagers) Control: participants were recruited from the same health centre; further details of their care were not provided (n = 46)
Outcomes	Outcomes: knowledge regarding health promotion pre‐counselling and post‐counselling, pre‐counselling and post‐counselling compliance regarding aspects of health promotion (e.g. antenatal follow‐up, hygiene), pregnancy outcomes (gestational age at labour, type of delivery, full‐term/pre‐term baby, general condition of baby, birth weight, breastfeeding, umbilicus)
Notes	Funding: no mention of funding
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Comment: this is a randomised controlled study; however the randomisation process is not described
Allocation concealment (selection bias)	Unclear risk	Comment: insufficient information to permit judgement
Blinding of participants and personnel (performance bias)   All outcomes	Unclear risk	Comment: insufficient information to permit judgement
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Comment: insufficient information to permit judgement
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Quote: "4 were excluded from the study group and 3 from the controls owing to incomplete data" Comment: explanation was included; attrition 7.5%
Selective reporting (reporting bias)	Unclear risk	Comment: this is not a registered trial; outcomes mentioned in the methods section are reported in the results section
Other bias	Low risk	Study seems to be free from other biases

Methods	Cluster‐randomised controlled trial implemented during 1998 to 2002 in 32 village clusters in Khuzdar, a rural district of Balochistan province in Pakistan
Participants	Women of reproductive age and their husbands (n = 2561)
Interventions	Information and Education for Empowerment and Change (IEEC) (n = 836) for women was designed to increase awareness of safe motherhood and neonatal health. Each facilitator initially invited 10 to 12 women from close villages to participate in a support group. Local traditional birth attendants (TBAs) ‐ who deliver over 90% of all births in the project area ‐ were trained in clean home delivery and in recognising common obstetrical and newborn emergencies. The project also facilitated timely referral and transportation of obstetrical and newborn emergencies to the district hospital. A typical support group started with a discussion of the problems faced by women during pregnancy and childbirth. Participants were then asked to look at their booklets while listening to a cassette tape that guided them through the pictures in the booklet. The pictures formed part of the dramatised stories recorded on the tape, thus creating an audiovisual effect. The booklet covered the following topics: family planning; nutrition; preparation for pregnancy and delivery; and danger signs during pregnancy, at delivery, and postpartum. Typically, the booklet was finished in 6 sessions of 1 to 2 hours each, after which participants were entitled to have their personal copy of the booklet and audiocassette. The husbands' IEEC (n = 703)was implemented in 8 village clusters randomly selected from the 16 intervention clusters. Husbands' booklets and audiocassettes were designed after formative research with married men. Then in each village cluster, 20 to 30 male community volunteers were identified who distributed the materials among husbands of the women who had participated in the support groups The control group received standard care (n = 1022)
Outcomes	Primary outcomes: neonatal mortality, perinatal mortality Secondary outcomes: iron supplementation, tetanus immunisation
Notes	Funding: funded by NICHD, USAID, UNICEF, World Health Organization, British Council, Government of Japan, and The Asia Foundation, and implemented by The Asia Foundation’s Islamabad office. There is no statement related to the involvement of these funding parties in design and conduct of the study
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "randomisation took place separately within each of the three zones; equal numbers of village clusters were randomly allocated to the intervention or control sites (by blindly drawing village cluster names written on folded chits)" Comment: probably done
Allocation concealment (selection bias)	Low risk	Quote: "randomisation took place separately within each of the three zones; equal numbers of village clusters were randomly allocated to the intervention or control sites (by blindly drawing village cluster names written on folded chits)" Comment: probably done
Blinding of participants and personnel (performance bias)   All outcomes	Unclear risk	Comment: insufficient information to permit any judgement
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Comment: insufficient information to permit any judgement
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	Comment: insufficient information to permit any judgement
Selective reporting (reporting bias)	Unclear risk	Comment: this is not a registered trial, but outcomes mentioned in the methods section are reported in the results section
Other bias	Low risk	Study seems to be free from other biases

Methods	Cluster‐randomised controlled trial in 24 intervention and 24 control settlements covering a population of 283,000 in India
Participants	Key participants were women who joined groups in 24 intervention clusters to be compared with 24 control clusters (n = 15,192 births)
Interventions	Intervention (n = 7656 births): study recruited 1 full‐time facilitator in each intervention cluster of about 1000 households. This sakhi (friend) was a local woman with secondary education and leadership skills, preferably married with children. Her role was to conduct meetings with women, attend planning and supervision of meetings, and support group action. After training, she began by profiling her settlement and building rapport with local stakeholders. She also attended regular training on a range of healthcare topics. Over about 6 months, she set up 10 women's groups, formative work having shown that women's mobility tended to be confined to their own alley. The groups met fortnightly, and she met weekly with other sakhis and her supervisor. The intervention followed a 36‐meeting cycle that was pre‐determined in general but developed iteratively in detail. There was no set point at which women had to join a group, and women of all ages, pregnant and non‐pregnant, were welcome to participate. Study took a participatory approach with emphasis on sharing and peer learning, rather than on the sakhi as an expert resource, and used the change method of appreciative inquiry to focus on the positive and to build energy for action through identification of the strengths of participants and their families and neighbourhoods Control (n = 7536): standard care
Outcomes	Perinatal care, maternal morbidity, extended perinatal mortality
Notes	Funding: interventions involved in the City Initiative for Newborn Health were funded by the ICICI Foundation for Inclusive Growth – Centre for Child Health and Nutrition. Evaluative aspects of the trial were funded from 2007 by The Wellcome Trust. DO was funded by a Wellcome Trust Fellowship (081052/Z/06/Z). The funding agency had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "in a transparent process, social workers external to the trial drew lots to select 48 in blocks of eight per ward, and then to allocate four clusters per block to the intervention" Comment: probably done
Allocation concealment (selection bias)	Low risk	Quote: "the nature of the intervention precluded allocation concealment" Comment: since it was a cluster randomised trial, allocation concealment should not be an issue, as in this design, all clusters are randomised at once
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Quote: "emphasis on participation and demystification of research" Comment: probably not done
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Comment: insufficient information to permit judgement
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Loss to follow‐up was around 18% in both intervention and control arms, and reasons for loss to follow‐up were similar across both arms
Selective reporting (reporting bias)	Low risk	Comment: this is a registered trial, and this study has reported all outcomes mentioned in the protocol
Other bias	Low risk	Comment: study seems to be free from other biases

Methods	Community‐based, 2‐arm, cluster‐randomised trial that tested whether a community‐ and facility‐based approach can improve pregnancy outcomes in low‐resource settings where there is limited access to quality obstetrical and neonatal care
Participants	This trial included all pregnancies of residents in 106 clusters, ranging from 6 clusters in Argentina to 24 in Pakistan, which were randomised with 53 in each treatment group. Outcomes were recorded for all women with newborns at ≥ 1000 grams and/or > 28 weeks who consented to be included in the study. A total of 70,351 pregnant women were screened in the intervention clusters and 66,830 in the control clusters (59,189 and 57,929 women in the intervention and control clusters, respectively, were eligible and consented)
Interventions	Intervention: each cluster included a cluster team that comprised healthcare providers, local residents, and study personnel to develop and implement comprehensive interventions to improve the quality of obstetrical and neonatal care. These teams worked within their community and local healthcare system to facilitate a multi‐faceted intervention that involved 2 components. The first component involved community mobilisation to establish village core groups and strengthen community capacity to identify needs and barriers related to obstetrical and neonatal care. Each village core group was trained to move through a cycle to organise, plan, explore, act on, and evaluate perinatal and maternal outcomes in their community. The second component consisted of Home‐based Life Saving Skills (HBLSS) training for birth attendants and families to recognise danger signs (e.g. prolonged labor, infection) and appropriate stabilisation methods that can be employed in homes and in first‐level care facilities. This component also involved improvement of quality care through a combination of facility staff Emergency Obstetric and Newborn Care (EmONC) training for clinical care for the major causes of perinatal and maternal mortality. This EmONC trial utilised a train‐the‐trainer model, the aforementioned 2 components of intervention training were combined, and this trial emphasised the Community Action Cycle and relevant HBLSS modules Control: some clusters were assigned to control with no intervention, and data were collected for comparison (no extensive detail for methods)
Outcomes	Primary outcome: perinatal mortality (defined: composite of stillbirth and 7‐day neonatal mortality per 1000 births among births occurring at > 28 weeks' gestation or > 1000 g) Secondary outcomes: rate of stillbirth (fresh and macerated), 7‐day neonatal mortality, 28‐day neonatal mortality, maternal death
Notes	Funding: this trial was funded by grants from the US National Institutes of Health (NIH). The NIH programme officers participated in protocol development and study monitoring and reviewed the manuscript
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "randomization was performed at the cluster level, stratifying by rates of the primary outcome (stillbirth and early neonatal death) and number of deliveries. The data coordinating centre produced a computer‐generated randomisation algorithm which assigned clusters at a 1:1 ratio within each stratum"  Comment: probably done
Allocation concealment (selection bias)	Low risk	Comment: since this was a cluster‐randomised trial, allocation concealment should not be an issue
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Quote: "because of the nature of the intervention, there was no masking" Comment: probably not done
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Comment: insufficient information to permit judgement
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Quote: "an additional 3,139 (4.5%) and 2,862 (4.3%) deliveries in intervention and control clusters were excluded at delivery due to missing data or other ineligibility criteria"
Selective reporting (reporting bias)	Low risk	Comment: this is a registered trial, and this study has reported all outcomes mentioned in the protocol
Other bias	Low risk	Comment: study seems to be free from other biases

Methods	Cluster‐randomised controlled trial investigating the effect of home‐based counselling on newborn care practices based in southern Tanzania
Participants	6 districts that consisted of 132 wards; 65 of these wards were allocated to the home‐based counselling intervention with routine care, and 67 received routine care only. After randomisation, there were no exclusion criteria for clusters, households, or women. All villages in the intervention wards recruited volunteers to implement the counselling intervention. Over 800 women who volunteered were recruited from and by their communities and were trained to make contact with every pregnant woman in their village A household survey sample size was based on the number of women age 13 to 49 who had given birth in the year before the survey and had received counselling or routine care. It was assumed that 1 woman of reproductive age resided in each household. 131 of the 132 wards were visited, and a total of 5217 households were visited. Of these households 4989 (96%) household heads agreed to participate ‐ 2491 in intervention areas and 2498 in comparison areas. For the final analysis, there were 512 women (n = 257 in intervention, n = 255 in comparison) from 128 subvillages who had delivered 521 live births during the study period and who answered the detailed survey questions
Interventions	Intervention: for every pregnant woman identified, a village volunteer was expected to make 3 visits to her home during pregnancy and 2 in the early neonatal period, with additional visits for small babies born. Counselling involved one‐on‐one interaction between volunteer and mother; there were also discussions with other family members (fathers and mother‐in‐laws) involving decision‐making around childbirth and newborn care. Counselling included behaviour change messages, specifically, hygiene during delivery (e.g. birth assistants using gloves), initiation of and exclusive breastfeeding, and identification of and care for small babies. Additional behaviour messages included birth preparedness (importance of health facility delivery, cleanliness of materials and money), delayed bathing of baby, and putting nothing on the cord. Postnatal counselling visits focused on reinforcing and supporting mothers to implement appropriate neonatal care. Volunteers also utilised Mtunzi counselling tools during their visits, including picture‐based cards that illustrated the counselling messages for each visit; these were left with the family to spread the message to other family members and to assist with retention. Volunteers also used a locally made doll to demonstrate the breastfeeding position and skin‐to‐skin care  Control: routine care
Outcomes	Primary outcomes: breastfeeding within an hour of delivery, birth attendants for home deliveries washing hands before childbirth or wearing gloves, babies fed only breast milk for the first 3 days Secondary outcomes: other behaviours promoted during counselling visits to maximise newborn health (e.g. birth preparedness, skilled attendance for childbirth, immediate drying and wrapping of baby, clean cord care, delayed bathing of the baby)
Notes	Funding: funded by the Bill & Melinda Gates Foundation through the Saving Newborn Lives programme of Save the Children, Unicef, the Laerdal Foundation, and the Batchworth Trust. Funders had no role in design and conduct of the study; in collection, analysis, and interpretation of data; nor in preparation, review, or approval of the manuscript
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "114 wards in 5 districts with baseline data...were randomised using implicit stratification to maximise balance in intervention and comparison groups... allocated the wards in each 'pair' to intervention of control using random numbers generated by Microsoft Excel"; "We used multi‐stage sampling to select households" Comment: probably done
Allocation concealment (selection bias)	Low risk	Comment: since this is a cluster‐randomised trial, allocation concealment should not be an issue, as in this design, all clusters are randomised at once
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Quote: "the nature of the intervention prevented blinding researchers, community members or health staff to the allocation" Comment: probably not done
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Comment: insufficient information to permit judgement
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	Comment: insufficient information to permit judgement
Selective reporting (reporting bias)	Low risk	Comment: this is a registered trial, and this study has reported all outcomes mentioned in the protocol
Other bias	Low risk	Comment: study seems to be free from other biases

Methods	Community‐based cluster‐randomised trial in rural Pakistan
Participants	All households in the selected districts were included in this study. There were 27 randomised clusters, including 35 healthcare facilities and their affiliated Lady Health Workers (LHWs), with the entire population of the clusters included in the trial. 13 clusters were allocated to the intervention group (n = 242,740) and 14 to the control group (n = 256,985). During the trial intervention period, 51,436 pregnancies were identified
Interventions	Intervention: package focused on immediate household management of intrapartum events, recognition of low birth weight and suspected serious neonatal infection, and prompt referral to public sector hospitals LHWs from the intervention clusters were trained at their affiliated health facility and were provided with a bag and mask for neonatal resuscitation and oral amoxicillin. LHWs were also given pictorial guides depicting the management of asphyxia, thermal care, co‐bedding, breastfeeding of low birth weight babies, and recognition of suspected pneumonia and administration of oral amoxicillin before referral. LHWs were also encouraged to maintain links with traditional birth attendants, to keep records of expected births, and to attend home deliveries. Pregnant women in the intervention clusters were provided with clean delivery kits during health education sessions delivered by LHWs and the importance of provision of urgent neonatal care at birth. LHWs were also trained as per national and project guidelines to do additional postnatal visits (days 3, 5, 14, 28 after birth). With regard to traditional birth attendants in the intervention arm, a programme in basic immediate maternal and newborn care was organised, and attendants were trained in the use of clean delivery kits. Separate training sessions on health education and community mobilisation were held for male community members, during which meetings aimed to promote antenatal and postnatal care and facility births (n = 19,984 deliveries) Control: received routine care through the existing national programme, with the LHW programme functioning as usual (n = 18,325 deliveries)
Outcomes	Primary outcome: all‐cause neonatal mortality Secondary outcomes: cause‐specific neonatal mortality due to intrapartum events, prematurity, and sepsis; stillbirth rate
Notes	Funding: Saving Newborn Lives, Save the Children USA. There is no statement related to involvement of these funding parties in design and conduct of the study
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "to ensure reasonable balance between the two arms, we used stratified, restricted randomisation to allocate clusters"; "1 million random allocation schemes were generated by the study statistician (SC), who used a computer algorithm" Comment: probably done
Allocation concealment (selection bias)	Low risk	Comment: as this is a cluster‐randomised trial, allocation concealment should not be an issue because the clusters are randomised at once
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Quote: "delivery of the intervention was not blinded for practical reasons" Comment: probably not done
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Quote: "verbal autopsies of all stillbirths and neonatal deaths were done by a separate team of trained anthropologists within 2–16 weeks of the event with standard WHO recommended instruments" Comment: probably done
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Reasons given (migrated or lost to follow‐up). Attrition (%) was 9.5% in each of intervention and control
Selective reporting (reporting bias)	Low risk	Comment: this is a registered trial, and this study has reported all outcomes mentioned in the protocol
Other bias	Low risk	Comment: study seems to be free from other biases