Schofield 1999.

Methods	Setting: hospital, Australia Recruitment: smokers discharged from hospital (unselected)
Participants	2465 smokers or recent quitters (excludes 1693 randomised but lost at 12‐month follow‐up) No differential dropout; 59% followed up in each arm; no demographic data
Interventions	∙ Self‐help 31‐page SoC‐based booklet + personally addressed letter from consultant stating health risks and urging to quit ∙ Usual care
Outcomes	Abstinence at 12 months and at 6 months Validation: urine cotinine ≤ 50 ng/mL or CO ≤ 8 ppm for sample Refusers (22% in each group) classified as smokers
Notes	Self‐help; no contact Study authors reported benefit for subgroup for whom quitting was highly relevant to diagnosis
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomised: "alternately allocated to intervention or control conditions by computer"
Allocation concealment (selection bias)	Low risk	Smokers identified at time of admission and allocation determined at that time Mailing of materials done by medical records office
Blinding (performance bias and detection bias) All outcomes	Low risk	Control group does not appear to have been aware of intervention condition; biochemical validation used
Incomplete outcome data (attrition bias) All outcomes	Low risk	Some people discovered to be ineligible at follow‐up and excluded Loss to follow‐up 41% ‐ identical in each group Meta‐analysis based on eligible respondents