Tripathy 2010.

Methods	A parallel arm cluster‐RCT conducted at 36 sites in India, Jharkhand and Orissa, between Jul 2005 and Jul 2008.
Participants	Sample size: 36 clusters (19030 births). Clusters: not clearly stated. The study area had disproportionately high NMR and an underserved population. Individuals: women aged 15‐49 residing in the project area who gave birth during the study (July 31, 2005‐July 30, 2008). Women who migrated out of the region were excluded from some analyses. 2 women from each arm refused the interview and were excluded.
Interventions	Target: community (IEC). Arm 1 (18 clusters, 9686 births): monthly facilitator‐convened women's groups monthly for a total of 20 meetings. Groups discussed maternal and newborn health problems and practices using pictures, role‐play and storytelling. In addition, health committees were formed to provide village representatives the chance to learn about health services and comment on their design and management. Arm 2 (18 clusters, 9089 births): in control clusters only health committees were formed.
Outcomes	Trial primary outcome: reduction in neonatal mortality rate and maternal depression. Review outcomes reported: Primary: maternal mortality. Secondary: ANC coverage (at least 1 visit), health facility deliveries, tetanus protection, perinatal mortality, neonatal mortality, stillbirth. Follow‐up: data collection took place monthly.
Notes	Funders: Health Foundation, UK Department for International Development, Wellcome Trust, and the Big Lottery Fund (UK). Data for years 1‐3 combined excluding migrants were used for our comparison 4 (Table 2, p. 1188). All ORs were taken directly from the published report (Tripathy 2010). The trial authors adjusted data for clustering, stratification, maternal education, assets and any tribal affiliation. Antenatal care outcome data are found in Table 5, p. 1190. It is unclear whether the OR presented for perinatal mortality (excluding migrants) includes infants who died between 0‐6 days or 0‐28 days (Table 3, p. 1188).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Drawing folded papers with numbers corresponding to clusters from a basket.
Allocation concealment (selection bias)	Unclear risk	The first 4 numbers drawn were assigned to the intervention; the next 4 to the control group. Participants in the randomisation process would have been aware of the next assignment but as the process was transparent it would not have been possible to manipulate the process.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Not blinded.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Not blinded.
Recruitment bias (for cluster RCTs)	Low risk	None noted.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Attrition and exclusions outlined in study flow diagram with limited missing data.
Selective reporting (reporting bias)	Unclear risk	Analysis was presented fully but there were multiple analyses with various adjustments and multiple testing which made results difficult to interpret.
Analysis bias	Unclear risk	Both adjusted and unadjusted data were provided. Adjustment for clustering and other factors did not appear to change the main conclusions. ICC of 0.0005 was mentioned but it's unclear if this was actually used for adjustments of data for neonatal death. Analysis was stated as by ITT.
Other bias	Unclear risk	There were baseline differences in household assets, maternal education, literacy and tribal membership; the intervention clusters were generally poorer. Some analyses adjusted for baseline differences.
Overall risk assessment	Unclear risk	We were uncertain how potential risks above impacted on findings.