Longitudinal patterns in smoking abstinence in trials of e-cigarettes for smoking cessation: secondary analysis of data from a systematic review, with meta-analyses.

INTRODUCTION: We set out to better understand patterns of smoking abstinence and relapse in trials of e-cigarettes for smoking cessation. METHODS: Secondary analysis of studies from a Cochrane review. Studies had to test any type of e-cigarette intervention for smoking cessation. They had to follow-up for at least 6 months and report either: abstinence at multiple time points; abstinence using multiple definitions; relapse. We narratively synthesized data and conducted meta-analyses. RESULTS: We included 15 studies (n=7,233 participants). Using the Cochrane risk of bias tool v1, 5 were judged to be at high risk of bias, 8 were at low risk, and 2 at unclear risk. Absolute continuous abstinence rates tended to decline over time, but varying slopes. For absolute point prevalence abstinence, three studies demonstrated a shallow decline over time, two a steep decline, and three the opposite - an increase in abstinence over time. Data on relative abstinence rates (e-cigarettes versus control) were mixed. There were multiple instances of differences between point prevalence and continuous/sustained abstinence rates, both in trajectories over time and in terms of relative abstinence. The few studies addressing relapse highlighted mixed demographic and behavioural characteristics associated with relapse rates. CONCLUSIONS: Smoking trajectories vary between trials of e-cigarettes for smoking cessation. Risk ratios may not be stable over time and may increase or decrease in favor of e-cigarettes depending on the study. Further data are needed, especially on relapse rates in early quitters who use e-cigarettes versus those who use other or no stop-smoking aids. IMPLICATIONS: While some have posited e-cigarettes might increase smoking relapse when used as a cessation tool, others have posited that combustible cigarette quit rates may increase over time in the same studies due to 'accidental quitting'. We set out to investigate this empirically and found considerable variation in smoking trajectories in e-cigarette trials. Data suggest risk ratios may not be stable over time and may increase or decrease in favor of e-cigarettes depending on the study. Further data are needed, especially on relapse rates in early quitters who use e-cigarettes versus those who use other or no stop-smoking aids.