Methods	3-armed randomised controlled trial of home visiting during pregnancy by paraprofessionals and nurses to improve maternal and child health, and included advice about smoking The study was conducted in 21 prenatal clinics in Denver (USA) from March 1994 to June 1995
Participants	Inclusion criteria: Pregnant women with no previous live births and either qualified for Medicaid or had no private medical insurance Exclusion criteria: Not further specified. Recruitment: By written invite, and were not required to respond. 735/1135 eligible women participated in the study, 70 of whom were smokers (C = 25, I1 = 21, I2 = 24). Baseline characteristics: Not reported among smoking subgroup.
Interventions	Control: Developmental screening and referral services for children at 6, 12, 15, 21 and 24 months old Intervention 1 (Paraprofessional): Screening and referral plus paraprofessional home visiting for first 2 years of infants life. Aimed to improve maternal and fetal health, improve health and development of child, and enhance parents personal development Intervention 2 (Nurse): Screening and referral plus nurse home visiting for first 2 years of infants life. Aimed to improve maternal and fetal health, improve health and development of child, and enhance parents personal development Main intervention strategy: Social support. Not coded or compared in this review as outcomes unable to be included
Outcomes	Outcomes not able to be included in meta-analysis, as only mean reduction in cotinine reported. See Table 1 for outcome summary.
Notes
Risk of bias
Bias	Authors’ judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer-generated random sequence.
Allocation concealment (selection bias)	Unclear risk	Allocation conducted in separate data centre.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Unclear whether all randomised smokers were included in cotinine analysis
Selective reporting (reporting bias)	High risk	Smoking cessation rates not reported, but are not a primary outcome of this study
Other bias	Low risk	No other bias detected.
Biochemical validation of smoking abstinence (detection bias)	Unclear risk	Unclear whether all randomised women included in cotinine analysis
Blinding of participants and personnel (performance bias) All outcomes	High risk	Providers and women not able to be blinded as social support intervention
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Outcome assessors blinded to allocation. Study team unaware of allocation, unless the participant told them
Incomplete implementation	Low risk	Paraprofessionals completed an average of 6.3 visits and nurses an average of 6.5 visits
Equal baseline characteristics in study arms	Unclear risk	Baseline characteristics of smokers not reported. But treatment groups similar with ‘few exceptions’
Contamination of control group	Low risk	Home visits.