Predictors of attrition in a randomized controlled trial of an electronic nicotine delivery system among people interested in cigarette smoking reduction

Background: Mitigating attrition is a key component to reduce selection bias in longitudinal randomized controlled trials (RCTs). Few studies of electronic nicotine delivery systems (ENDS) allow for the examination of long-term retention. This analysis explores the relationship between attrition, baseline measures, and condition assigned for a RCT involving ENDS differing in nicotine delivery over a 24-week intervention period. Methods: Participants (N = 520) who smoked >= 10 cigarettes per day [CPD] for >= 1 year and reported interest in reducing but not quitting were randomized to 1 of 4 conditions: an ENDS containing 0, 8, or 36 mg/ml liquid nicotine (administered double-blind) or a cigarette-shaped plastic tube. Cox proportional hazards regression models were fit to examine attrition over time and predictors of attrition including baseline characteristics and condition. A stepwise approach was used to determine the final model; alpha was set at 0.05. Results: Attrition did not differ significantly by condition (223/520), and most (69%) were lost-to-follow-up. Only age, education level, and household income were significantly predictive of attrition. For every additional year of age, attrition risk fell by 3%. Holding a bachelor's degree or higher was associated with reduced attrition risk. Those with the lowest income (<$10 K) were more likely to be withdrawn compared to those earning $10 K-39 K, and those with the highest income ($100 K+) were more likely to be withdrawn compared with the latter bracket and those earning $70-99 K. Conclusion: ENDS nicotine content did not drive differential attrition in this trial, and targeted retention efforts are needed for specific subgroups.