Omer 2008.

Methods	Parallel arm cluster‐RCT conducted at 10 sites in Pakistan (Sindh province) between Jun 2000 and April 2001.
Participants	Sample size: 10 clusters (1070 women interviewed, 969 households visited). Clusters: 10 enumeration areas from 3 districts in the Sindh province of Pakistan were chosen. 8 of the 10 areas were rural. Individuals: women who were pregnant or had delivered in the past 3 years were eligible.
Interventions	Target: health system (change in health worker practice) and community (IEC). Arm 1 (5 clusters, 529 women): a LHW showed an evidence‐based tool and embroidered cloth that depicted 3 important maternal practices, viz. attending antenatal check‐ups, giving colostrum after birth and avoiding heavy work. Arm 2 (5 clusters, 541 women): the LHW delivered standard care.
Outcomes	Trial primary outcome: health practices during pregnancy, including ANC. Review outcomes reported: Primary: not reported. Secondary: ANC coverage (at least 1 visit), professional (LHW) ANC. Other: giving colostrum at birth, stopping heavy work, exclusive breastfeeding for 4 months. Follow‐up: after the intervention had been in use for 10 months, field workers completed data collection in 1 week. Fieldworkers also conducted a household survey of pregnant women or those who had delivered in the past 3 years.
Notes	Funders: Canadian International Development Agency International Development Agency’s (CIDA) Canada Fund for Local Initiatives in Pakistan. Data for ANC visits were not reported by randomisation group and not usable.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Sequence from computerised random numbers generator.
Allocation concealment (selection bias)	Unclear risk	Not described.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Blinding not possible due to the nature of the intervention.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Fieldworkers collecting data were blinded to the allocation of the community in which they worked but women may have revealed which group they were in if they mentioned the embroideries.
Recruitment bias (for cluster RCTs)	Low risk	None noted.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	The total number of women visited by LHW is not stated, only households visited. It is not clear how many households refused the survey or whether there were women approached who refused the interview. There seem to me more women with data for visits than there were households visited or women interviewed.
Selective reporting (reporting bias)	Unclear risk	Exclusive breastfeeding for 4 months is mentioned in the abstract, but there are no results for this outcome.
Analysis bias	High risk	It is stated that ITT analysis was undertaken but most results were not reported by randomisation group but rather by whether or not women had seen the LHW. There is no evidence of any adjustment made for correlations within or between clusters. The analysis seems to have been done at the individual level. It was stated that baseline imbalance was taken into account in secondary analysis.
Other bias	Unclear risk	Baseline imbalances are not clearly stated.
Overall risk assessment	High risk	Due to unclear group denominators, attrition and inappropriate analysis methods.