Recurrent event analysis of lapse and recovery in a smoking cessation clinical trial using bupropion

We report a reanalysis of data from a prior study describing the event history of quitting smoking aided by bupropion, using recurrent-event models to determine the effect of the drug on occurrence of lapses and recoveries from lapse (resumption of abstinence). Data were collected on 1,070 subjects across two similar double-blind randomized clinical trials of bupropion versus placebo and fitted with separate Cox regression models for lapse and recovery. Analyses were split using discrete time-varying covariates between the treatment (weeks 1-10) and follow-up phases (end of treatment to 12 months). Bupropion was associated with slower lapse during treatment for both sexes, and being female was associated with faster lapse across both phases. Drug did not affect time to recovery for males but was associated with faster recovery among females, allowing women to recover as quickly as men. High levels of nicotine dependence did not affect time to lapse but were associated with slower recovery from lapse across treatment and follow-up phases. During the treatment phase, higher levels of baseline depression symptoms had no effect on time to lapse but were associated with slower recovery from lapse. Results highlight the asymmetry in factors preventing lapse versus promoting recovery. Specifically, dependence, depression symptoms, and a sex x drug interaction were found to affect recovery but not lapse. Further research disentangling lapse and recovery events from summary abstinence measures is needed to help us develop interventions that take advantage of bupropion at its best and that compensate where it is weak.