Gateway effects and electronic cigarettes.

Abstract

E-cigarettes are alleged to be a gateway to cigarette smoking in non-smokers. This study examines whether the gateway theory has value, whether the criteria to establish causality have been met and what type of evidence is required to test this theory. Experiments are impractical, and we may not be able to test properly the gateway effects via observational studies that simply adjust for confounders. Multivariate models cannot eliminate all the variance in propensity to smoke captured by the variable 'vaping' because of the proximity of these two behaviours. It may be difficult to prove that vaping precedes smoking when product use co-occurs and when, in fact, smoking usually precedes vaping. The gateway theory is not compatible with either (1) the decrease in smoking prevalence observed in adolescents in countries where vaping increased or (2) an increase in smoking among teenagers after age restrictions were imposed on e-cigarette purchases. A spurious gateway effect can be produced artificially by mathematical models in which a propensity to use substances is correlated with opportunities to use substances. Finally, neither nicotine medications nor smokeless tobacco produce gateway effects. Available data are compatible with a common liability model in which people who are liable to use nicotine are more likely to use both e-cigarettes and cigarettes. Despite its weaknesses and scant empirical support, the gateway theory of smoking initiation has had enormous political influence. Policies based on this theory will not have the intended effects if the association between vaping and smoking is explained by common liabilities.