A Cochrane review included 63 studies, heterogeneous in terms of populations and interventions. 41 studies that randomized abstainers were considered separately from studies that randomized participants prior to their quit date. Most studies were concerned with interventions focusing on identifying and resolving tempting situations.
No benefit was found of brief and 'skills-based' relapse prevention methods for women who had quit smoking due to pregnancy, or for smokers undergoing a period of enforced abstinence during hospitalisation or military training. Also, no significant effects were detected in trials involving smokers who had quit on their own or with a formal programme. In studiess randomising smokers prior to their quit date, no evidence of benefit was found for additional relapse prevention components in any subgroup. Providing training in skills thought to be needed for relapse avoidance was not shown to reduce relapse, but most studies did not use experimental designs best suited to the task, and had limited power to detect expected small differences between interventions.
For pharmacological interventions, extended treatment with varenicline significantly reduced relapse (RR 1.18, 95% CI 1.03 to 1.36, 1 study, n=1210). Extended treatment with bupropion failed to detect a significant effect (RR 1.15, 95% CI 0.98 to 1.35; 6 studies, n=1697). 2 small trials of oral nicotine replacement treatment (NRT) failed to detect an effect but treatment compliance was low and in 2 other trials of oral NRT randomizing short-term abstainers there was a significant effect of intervention.
Comment: The quality of evidence is downgraded by study quality (inadequate or unclear allocation concealment), by inconsistency (heterogeneity in populations and interventions), and by indirectness (most studies which randomized recent abstainers focused on brief or written interventions rather than on more intensive treatments).