Factors moderating the relative effectiveness of varenicline and nicotine replacement therapy in clients using smoking cessation services

AimsTo assess how far the greater effectiveness of varenicline over nicotine replacement therapy (NRT) is moderated by characteristics of the smokers or setting in clinical practice. DesignWe used observational data from 22472 treatment episodes between 2013 and 2016 from smoking cessation services in England to assess whether differences between varenicline and NRT were moderated by a set of smoker and setting characteristics: these included level of social deprivation, age, gender, ethnic group, nicotine dependence and treatment context. From the above, 15640 episodes were analysed in relation to 4-week quit and 14273 episodes at 12weeks. All two-way interactions involving pharmacotherapy were fitted in addition to the main effects and a parsimonious model identified using a backwards stepwise selection procedure. SettingEngland ParticipantsClients of smoking cessation service (number of individuals in 4-week quit analysis=15640). MeasurementsFour-week carbon monoxide-validated (primary outcome) and 12-week self-reported (secondary outcome) quit success/failure. FindingsAt both follow-up points, varenicline was associated with higher success rates overall [P<0.001 at both 4 and 12weeks; adjusted odds ratio (OR) varenicline versus NRT=1.82 (95% confidence interval (CI)=1.61, 2.06) and 2.58 (95% CI=2.26, 2.94) at 4 and 12weeks, respectively]. At 12weeks, the relative benefits of varenicline were found to be influenced by the setting in which advice was provided [P<0.001 for interaction pharmacotherapy x setting; adjusted odds ratio for varenicline x pharmacy setting=0.53 (95% CI=0.42, 0.69) and for varenicline x general practice (GP) setting=0.79 (95% CI=0.64, 0.98) against a baseline of 1 for varenicline x community setting]. The same trends were evident at 4weeks, but this did not translate to statistical significance. There was inconclusive evidence for moderating effects of other variables. ConclusionsVarenicline use was associated with higher smoking cessation rates than nicotine replacement therapy in routine clinical practice, irrespective of a wide range of smoker characteristics, but the difference was less in certain intervention settings, most notably pharmacy but also GP practice, compared with community setting.