Table 2.

Risk of bias of the included studies

Keltner[20]	Author’s judgment	Support of judgment
Allocation sequence (potential selection bias)	Unclear risk	Method not specified
Allocation concealment (potential selection bias)	Low risk	Allocation was performed independently by a person not associated with the project
Baseline differences in outcomes	High risk	Imbalanced for the marital status and not adjusted
Baseline differences in characteristics	Low risk	No significant difference see Table 1
Incomplete outcome data (potential attrition bias)	Low risk	No attrition
Knowledge of interventions (potential detection bias)	Low risk	Assessed blindly: assessors blinding: local professionals who would likely know some or all of the families participating were not involved in the assessments
Contamination	-	-
Selective outcome reporting	Low risk	No other outcomes mentioned
Other risks of bias	Low risk
Scafidi[22]	Author’s judgment	Support of judgment
Allocation sequence	Low risk	Random number table
Allocation concealment	Low risk	Randomised by researcher not associated with the project
Baseline differences in outcomes	Low risk	Imbalanced but adjusted
Baseline differences in characteristics	Low risk	Adjusted
Incomplete outcome data	Low risk	No drop out
Knowledge of interventions	Low risk	Assessors were blind to the group intervention
Contamination	-	-
Selective outcome reporting	Low risk	No other outcomes mentioned
Other risks of bias	Low risk
Barbosa[21]	Author’s judgment	Support of judgment
Allocation sequence	High risk	By clinic attendance day
Allocation concealment	Unclear risk	No information
Baseline differences in outcomes	Unclear risk	No information
Baseline differences in characteristics	High risk	No comparison made between the groups
Incomplete outcome data	Unclear	No information
Knowledge of interventions	Unclear	No information
Contamination	-	-
Selective outcome reporting	Low risk	Only reported items measured
Other risks of bias	Low risk