Associations of Demographics, Dependence, and Biomarkers With Transitions in Tobacco Product Use in a Cohort of Cigarette Users and Dual Users of Cigarettes and E-cigarettes

Abstract

Introduction It is uncertain whether e-cigarettes facilitate smoking cessation in the real world. We aimed to understand whether and how transitions among cigarette, e-cigarette, and dual use are associated with sociodemographics, dependence measures, and biomarkers. Methods We followed 380 adult daily cigarette users and dual users every 2 months for up to 2 years. We estimated transition rates between non-current, cigarette-only, e-cigarette-only, and dual use states using a multistate transition model. We estimated univariable hazard ratios (HR) for demographics, dependence measures for cigarettes and e-cigarettes, biomarkers, spousal/partner behaviors, and other measures. Results We estimated that participants transitioned from cigarette-only to e-cigarette-only through a period of dual use. Dual users ceased smoking (transitioning to e-cigarette-only use) at a greater rate than cigarette-only users did (HR 2.44, 95% CI: 1.49, 4.02). However, of the 60% of dual users estimated to transition to single product use in one year, 83% would transition to cigarette-only use and only 17% to e-cigarette-only use. E-cigarette dependence measures were generally associated with reduced e-cigarette cessation rather than enhanced cigarette cessation. E-cigarette users motivated by harm/toxicity reduction or because of restrictions on where/when they could smoke had reduced rates of smoking relapse. Cigarette dependence and spousal smoking were barriers to cigarette cessation for dual users, while using e-cigarettes first in the morning, motivation to quit smoking, and sensory/social/emotional enjoyment of e-cigarettes (secondary dependence motives) were facilitators smoking cessation among dual users. Conclusions Tobacco control policy and interventions may be informed by the barriers and facilitators of product transitions. Implications While e-cigarettes have the potential to promote smoking cessation, their real-world impact is uncertain. In this cohort, dual users were more likely to quit smoking than cigarette-only users, but the overall impact was small because most dual users returned to cigarette-only use. Moreover, e-cigarette dependence promoted continued dual use rather than smoking cessation. Yet, high motivation to quit smoking and the sensory, social, and emotional enjoyment of e-cigarettes facilitated smoking cessation in dual users. Better understanding the barriers and facilitators of transitions can help to develop regulations and interventions that lead to more effective use of e-cigarettes for smoking cessation.

Changing patterns of cigarette and ENDS transitions in the USA: a multistate transition analysis of youth and adults in the PATH Study in 2015–2017 vs 2017–2019

“…To do so, we use continuous-time multistate transition models, which are increasingly being used to analyse tobacco product transitions and to understand how sociodemographic factors impact these underlying rates. [17][18][19][20][21][22][23][24] In a previous multistate transition analysis, we found that ENDS use in 2013-2017 was less persistent than cigarette use in the USA, 19 but more recent work has suggested that ENDS use has become more persistent in recent years. 15 In this analysis, we build on that previous work, implementing a multistate transition model using data from the Population Assessment of Tobacco and Health (PATH) Study, distinguishing between waves 2-4 (2015-2017) and waves 4 and 5 (2017-2019) for adults and youth, and adjusting for demographics and frequency of product use.…”

“…Other limitations of the method include the lack of accounting for individuals' longer-term product trajectories (Markov assumption) and our inability to make causal inferences (eg, differences in cigarette initiation between those who never established use and those who use ENDS only may be caused by underlying demographic differences in the two populations, not the ENDS use itself). Additionally, while the model does account for the possibility of multiple transitions between follow-up points, yearly data (and, for adults in waves 4 and 5, every other year) are not sufficient to capture shorter-term transitions in past 30-day use 22 or the dynamics of experimentation with smoking or vaping. We also did not account for ENDS flavouring in this analysis.…”

Machine Learning-Based Sales Prediction Using Bayesian Optimized XGBoost Algorithms

“…The benchmark results of this study are those of a specific machine learning model from the XGBoost, which is widely regarded as a cutting-edge machine learning model. It is possible that the model's performance can be improved and it may have underperformed in the previous study due to suboptimal hyperparameter tuning procedures [25][26][27]. This paper extracts sampling data features for different data sources in different situations, and selects a portion of data samples from the overall dataset for the process of model training research.…”

Changing patterns of cigarette and ENDS transitions in the US: a multistate transition analysis of youth and adults in the PATH Study in 2015–17 vs 2017–2019

“…It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. use multistate transition models, which are increasingly being used to analyze tobacco product transitions and to understand how sociodemographic factors impact these underlying rates [15][16][17][18][19]. In a previous multistate transition analysis, we found that ENDS use in 2013-2017 was less persistent than cigarette use in the US [15], but more recent work has suggested that ENDS use has become more persistent in recent years [20].…”