Primary Submission Category: Applications in Health and Biology
ENDS and Cigarette Reduction: A Causal Bayesian Additive Regression Tree Analysis
Authors: Shu Xu, Jennifer Hill, Luchang Cui, Yang Feng, Raymond Niaura,
Presenting Author: Shu Xu*
There is ongoing debate about whether ENDS can help adult cigarette smokers quit smoking. One way to address this question is to examine the unbiased association between ENDS use and the frequency of cigarette smoking. However, this analysis faces challenges, including potential confounding in observational data, nonlinear and interactive relationships among study variables, and the difficulty of modeling the skewed distribution of cigarette frequency outcome.
We analyzed longitudinal data from Waves 4 through 6 of the Population Assessment of Tobacco and Health (PATH) Study. The analytic sample included 4193 adults who were current established cigarette smokers identified at Wave 4. We assessed the association between current ENDS use at Wave 5 and the number of days of cigarette smoking in the past 30 days at Wave 6 using Bayesian Additive Regression Tree analyses for causal inference. Models adjusted for a range of Wave 4 covariates (e.g., sociodemographic, nicotine dependence, and other relevant factors).
Assuming we have adjusted for all confounders, results indicate that Daily use of ENDS among established cigarette smokers is associated with a subsequent decline in days of cigarette smoking, compared to occasional or no use. These findings highlight the potential of ENDS as a harm reduction tool for adult smokers.