Smoking reduction interventions for smoking cessation

Background: The standard way most people are advised to stop smoking is by quitting abruptly on a designated quit day. However, many people who smoke have tried to quit many times and may like to try an alternative method. Reducing smoking behaviour before quitting could be an alternative approach to cessation. However, before this method can be recommended it is important to ensure that abrupt quitting is not more effective than reducing to quit, and to determine whether there are ways to optimise reduction methods to increase the chances of cessation. Objectives: To assess the effect of reduction‐to‐quit interventions on long‐term smoking cessation. Search Methods: We searched the Cochrane Tobacco Addiction Group Specialised Register, MEDLINE, Embase and PsycINFO for studies, using the terms: cold turkey, schedul*, cut* down, cut‐down, gradual*, abrupt*, fading, reduc*, taper*, controlled smoking and smoking reduction. We also searched trial registries to identify unpublished studies. Date of the most recent search: 29 October 2018. Selection Criteria: Randomised controlled trials in which people who smoked were advised to reduce their smoking consumption before quitting smoking altogether in at least one trial arm. This advice could be delivered using self‐help materials or behavioural support, and provided alongside smoking cessation pharmacotherapies or not. We excluded trials that did not assess cessation as an outcome, with follow‐up of less than six months, where participants spontaneously reduced without being advised to do so, where the goal of reduction was not to quit altogether, or where participants were advised to switch to cigarettes with lower nicotine levels without reducing the amount of cigarettes smoked or the length of time spent smoking. We also excluded trials carried out in pregnant women. Data Collection and Analysis: We followed standard Cochrane methods. Smoking cessation was measured after at least six months, using the most rigorous definition available, on an intention‐to‐treat basis. We calculated risk ratios (RRs) and 95% confidence intervals (CIs) for smoking cessation for each study, where possible. We grouped eligible studies according to the type of comparison (no smoking cessation treatment, abrupt quitting interventions, and other reduction‐to‐quit interventions) and carried out meta‐analyses where appropriate, using a Mantel‐Haenszel random‐effects model. We also extracted data on quit attempts, pre‐quit smoking reduction, adverse events (AEs), serious adverse events (SAEs) and nicotine withdrawal symptoms, and meta‐analysed these where sufficient data were available. Main Results: We identified 51 trials with 22,509 participants. Most recruited adults from the community using media or local advertising. People enrolled in the studies typically smoked an average of 23 cigarettes a day. We judged 18 of the studies to be at high risk of bias, but restricting the analysis only to the five studies at low or to the 28 studies at unclear risk of bias did not significantly alter results. We identified very low‐certainty evidence, limited by risk of bias, inconsistency and imprecision, comparing the effect of reduction‐to‐quit interventions with no treatment on cessation rates (RR 1.74, 95% CI 0.90 to 3.38; I2 = 45%; 6 studies, 1599 participants). However, when comparing reduction‐to‐quit interventions with abrupt quitting (standard care) we found evidence that neither approach resulted in superior quit rates (RR 1. 01, 95% CI 0.87 to 1.17; I2 = 29%; 22 studies, 9219 participants). We judged this estimate to be of moderate certainty, due to imprecision. Subgroup analysis provided some evidence (P = 0.01, I2 = 77%) that reduction‐to‐quit interventions may result in more favourable quit rates than abrupt quitting if varenicline is used as a reduction aid. Our analysis comparing reduction using pharmacotherapy with reduction alone found low‐certainty evidence, limited by inconsistency and imprecision, that reduction aided by pharmacotherapy resulted in higher quit rates (RR 1. 68, 95% CI 1.09 to 2.58; I2 = 78%; 11 studies, 8636 participants). However, a significant subgroup analysis (P < 0.001, I2 = 80% for subgroup differences) suggests that this may only be true when fast‐acting NRT or varenicline are used (both moderate‐certainty evidence) and not when nicotine patch, combination NRT or bupropion are used as an aid (all low‐ or very low‐quality evidence). More evidence is likely to change the interpretation of the latter effects. Although there was some evidence from within‐study comparisons that behavioural support for reduction to quit resulted in higher quit rates than self‐help resources alone, the relative efficacy of various other characteristics of reduction‐to‐quit interventions investigated through within‐ and between‐study comparisons did not provide any evidence that they enhanced the success of reduction‐to‐quit interventions. Pre‐quit AEs, SAEs and nicotine withdrawal symptoms were measured variably and infrequently across studies. There was some evidence that AEs occurred more frequently in studies that compared reduction using pharmacotherapy versus no pharmacotherapy; however, the AEs reported were mild and usual symptoms associated with NRT use. There was no clear evidence that the number of people reporting SAEs, or changes in withdrawal symptoms, differed between trial arms. Authors' Conclusions: There is moderate‐certainty evidence that neither reduction‐to‐quit nor abrupt quitting interventions result in superior long‐term quit rates when compared with one another. Evidence comparing the efficacy of reduction‐to‐quit interventions with no treatment was inconclusive and of low certainty. There is also low‐certainty evidence to suggest that reduction‐to‐quit interventions may be more effective when pharmacotherapy is used as an aid, particularly fast‐acting NRT or varenicline (moderate‐certainty evidence). Evidence for any adverse effects of reduction‐to‐quit interventions was sparse, but available data suggested no excess of pre‐quit SAEs or withdrawal symptoms. We downgraded the evidence across comparisons due to risk of bias, inconsistency and imprecision. Future research should aim to match any additional components of multicomponent reduction‐to‐quit interventions across study arms, so that the effect of reduction can be isolated. In particular, well‐conducted, adequately‐powered studies should focus on investigating the most effective features of reduction‐to‐quit interventions to maximise cessation rates.