Kottke 1989.

Methods	Setting: family practices, USA Recruitment: physicians recruited for trial; target population ‐ all patients seen during the month (unselected)
Participants	66 physicians, 1653 smoking patients, "2/3rds female, average age slightly over 40 years, just under one pack/day"
Interventions	∙ Physicians attended 6‐hour workshop ∙ Physicians attended workshop and given copies of Quit and Win for their patients ∙ Physicians received no support, but were asked to advise patients during the study period
Outcomes	Abstinence at 1 year Validation: serum cotinine
Notes	2 vs 1, effect of self‐help in addition to advice from a trained physician Including 3 in control group does not affect results (RR for trial becomes 1.02 rather than 0.99)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Cluster‐randomised by physician ‐ not by smoker; method not described; potential for imbalance in participant characteristics, but number of participants per physician low
Allocation concealment (selection bias)	High risk	Researchers attempted to contact all participants seen by physicians during 1 month
Blinding (performance bias and detection bias) All outcomes	Low risk	Biochemical validation used, as with cluster‐randomisation by physician, seems unlikely that control group participants would know what other participants were offered
Incomplete outcome data (attrition bias) All outcomes	Low risk	Over 87% of smokers identified at baseline were reached at 1 year ‐ similar across groups