Bobo 1998.

Methods	Study design: Cluster‐randomized controlled trial Setting: 12 residential centres for alcohol/drug treatment, USA Recruitment: inpatient volunteers
Participants	(50 participants in each of 12 sites) 67% men, av. age 33 50% smoked > 1 pack/day
Interventions	Therapists: centre staff for 1st session, trained counsellors for telephone sessions 1. 4 x 10 ‐ 15 min sessions. 1st during inpatient stay. 3 by telephone, 8, 12, 16 wks post‐discharge 2. No intervention
Outcomes	Abstinence at 12 m post‐discharge (7 day PP) Validation: saliva cotinine, but validated quit rates not reported (A primary outcome for the study was alcohol abstinence)
Notes	Cluster‐randomized, so individual data not used in primary meta‐analysis. Adjusted OR 1.02 (CI 0.50 to 2.49). Inclusion would not materially change results of analysis 1.1.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Matched pairs of centres allocated by coin toss, 2 centres declined participation after allocation
Allocation concealment (selection bias)	High risk	Cluster‐randomized with participant recruitment (by research team) after centre allocation so potential for selection bias
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Biochemical validation of abstinence but validated results not reported
Incomplete outcome data (attrition bias) All outcomes	Low risk	22% lost to follow‐up. Including them as smokers made little difference to estimates