Hajek 2015

Methods	Country: UK Setting: Specialist stop‐smoking clinic in London Aim: To determine whether increasing varenicline dose in people who show no response to the drug improves treatment efficacy in terms of tobacco withdrawal relief and abstinence rates Study Design: Double‐blind placebo‐controlled RCT Dates conducted: July 2011 ‐ February 2013 Analysis: ANOVA for continuous end points and Χ² for categorical end points. Sample size of 200 for 80% power to detect a difference in 4‐wk abstinence between 60% placebo and 80% varenicline. 2‐tailed P < 0.05
Participants	200 non‐responders to varenicline at day 12, from an initial cohort of 503 given varenicline while still smoking, randomised to varenicline (100) or placebo (100) add‐on treatment. Treatment‐seeking smokers, aged 18+; 28% women, 65% white, mean age 45.8 yrs, 20.5 cigs in previous wk, mean FTND 5.5
Interventions	503 eligible consented volunteers began taking varenicline at standard dosage; at day 12, 204 were rated as non‐responders (no strong nausea, no reduction in smoking enjoyment, < 50% reduction in baseline smoking), and 200 were then randomised to additional varenicline or placebo. All participants received phone calls on days 15 and 18, with TQD at day 21 + phone call 24 hours later, and 4 x weekly supportive visits (as per standard NHS stop‐smoking treatment protocol). participants were invited to a 12‐week final visit for assessment 1. Varenicline: standard dose + initial increase of 0.5 mg x 2/day which could be increased by 0.5 twice daily up to a total of 5 mg/day. Dosage used at TQD was maintained for 3 wks, with an option to reduce it if necessary. From 4 wks, only standard dose was used 2. Placebo: same regimen, but with identical placebo pills
Outcomes	Smoking enjoyment and withdrawal symptoms weekly for 1st 4 wks; CAR at wks 1, 4, 12 after TQD Validation: CO < 9 ppm
Treatment type	Medication: VARENICLINE
Notes	Funding: Pfizer Although this study did not assess abstinence beyond 3m, we have included it for assessment of variation in dosing, and for safety data. We have not pooled the efficacy findings with the other included studies, apart from sensitivity analysis 13.2 (CAR at 9 ‐ 12 wks) New for 2016 update
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Participants were randomized to treatment arms using sequentially numbered prepackaged medication containers boxed according to a computer‐generated randomization list prepared by an independent statistician"
Allocation concealment (selection bias)	Low risk	See above
Blinding (performance bias and detection bias) All outcomes	Low risk	"The authors were unblinded only after the data analysis was completed"
Incomplete outcome data (attrition bias) All outcomes	Low risk	All losses during treatment and follow‐up reported; ITT analyses conducted
Selective reporting (reporting bias)	Low risk	None noted
Other bias	Low risk	None noted