E‐cigarette use is associated with subsequent cigarette use among young adult non‐smokers, over and above a range of antecedent risk factors: a propensity score analysis

Background and Aims There is a public health concern that the use of e‐cigarettes among non‐smoking young adults could be associated with transition to combustible cigarette use. The current study is a quasi‐experimental test of the relationship between e‐cigarette use and subsequent combustible cigarette use among young adult non‐smokers, accounting for a wide range of common risk factors. Design Logistic regression was used to predict combustible cigarette use on three or more occasions at age 23 years based on age 21 e‐cigarette use. Inverse probability weighting (IPW) was used to account for confounding variables. Setting Data were drawn from the Community Youth Development Study (CYDS), a cohort study of youth recruited in 2003 in 24 rural communities in seven US. states Participants Youth in the CYDS study (n = 4407) were surveyed annually from ages 11 to 16, and at ages 18, 19, 21 and 23 years (in 2016). The sample was gender balanced (50% female) and ethnically diverse (20% Hispanic, 64% white, 3% black and 12% other race or ethnicity). The current study was limited to participants who had never used combustible cigarettes by age 21 (n = 1825). Measurements Age 21 use of e‐cigarettes and age 23 use of combustible cigarettes (three or more occasions) were included in the regression analysis. Age 11–19 measures of 22 common predictors of both e‐cigarette and combustible cigarette use (e.g. pro‐cigarette attitudes, peer smoking, family monitoring) were used to create IPWs. Findings After applying IPW, e‐cigarette use at age 21 was associated with a twofold increase in odds of combustible cigarette use on three or more occasions 2 years later (odds ratio = 2.16, confidence interval 1.23, 3.79). Conclusions Among previously never‐smoking US young adults, e‐cigarette use appears to be strongly associated with subsequent combustible cigarette smoking, over and above measured preexisting risk factors.