Gonzales 2006

Methods	Country: USA Setting: 19 research centres Aim: To test the efficacy and safety of varenicline for smoking cessation Dates conducted: June 2003 ‐ April 2005 Study Design: Double‐blind placebo‐controlled parallel‐group RCT Analysis: Power calculation (90%, alpha = 0.05); ITT denominators and logistic regression analysis (step‐down procedure)
Participants	1025 healthy adult volunteers, recruited through media advertising. Allocated to varenicline (352), bupropion (329) or placebo (344). 54% men, 79% white, mean age 42.4, mean CPD 21, mean FTND score 5.3. No significant differences between groups at baseline Exclusion criteria: Standard pharmacotherapy trial criteria, + use of tobacco products other than cigarettes; use of NRT, clonidine, nortriptyline within last month; BMI < 15 or > 38 or weight < 45.5 kg; any prior use of bupropion or varenicline
Interventions	1. Varenicline 1 mg x 2/day 2. Bupropion 150 mg x 2/day 3. Placebo inactive tablets, same regimen Treatment period was 12 wks. All participants received Clearing the Air self‐help booklet at baseline, and brief counselling (≤ 10 mins) at each clinic visit. Weekly visits throughout treatment phase, plus a phone call 3 days post‐TQD In follow‐up phase, clinic visits at wks 13, 24, 36, 44 and 52, plus brief phone calls at wks 16, 20, 28, 32, 40 and 48
Outcomes	Primary outcome: CO‐validated CAR at 9 ‐ 12 wks Secondary outcomes: CO‐validated CAR at 9 ‐ 24 wks and 9 ‐ 52 weeks; 7‐day PPA at wks 12, 24 and 52 Other outcomes: Weight change, withdrawal symptoms (using MNWS, QSU‐brief and mCEQ), adverse events Validation was by expired CO ≤ 10 ppm Dropouts and losses to follow‐up were included in the analyses as continuing smokers (ITT analysis) Attrition in treatment phase was 31.5%, losses to follow‐up 16% of treatment completers
Treatment type	Medication: VARENICLINE / BUPROPION
Notes	This trial had the same aims and study design as Jorenby 2006 The trial was funded by Pfizer Inc.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"predefined ... computer‐generated randomization sequence", 1:1:1, using block size of 6, stratified by centre
Allocation concealment (selection bias)	Low risk	Central allocation
Blinding (performance bias and detection bias) All outcomes	Low risk	"Participants and investigators were blinded to drug treatment assignments[, and] ... were not encouraged to guess their treatment assignment"
Incomplete outcome data (attrition bias) All outcomes	Low risk	Considered abstinent if, at next non‐missed visit, they reported no smoking... Missing CO but otherwise OK considered abstinent, except at end of study, where all criteria had to be present
Selective reporting (reporting bias)	Low risk	All expected and predicted outcomes covered
Other bias	Unclear risk	None noted