[Update, 21 May 2015: This paper now appears at International Journal of Environmental Research and Public Health (open access html version here, pdf version downloadable via link). Thanks to readers for the comments on the working paper that improved that version.]
My new working paper is available for download: Phillips – how to detect gateway effects (pdf).
This paper serves multiple purposes. It addresses the titular question, about how an analysis of empirical evidence could support a claim that e-cigarettes, smokeless tobacco, or other low-risk tobacco products are a gateway to smoking. Along the way, it points out that none of the studies to date that are purported to support that claim actually do so. This is also a methods paper (and is being submitted to a methods issue of a journal), and thus goes into some detail on generalizable methodological points. It does so — I believe — in a readable and expository way, such that interested readers can get a lot out of that even if they are not students of epidemiology methods.
This paper is quite different from, but subsumes most of the key content of, my earlier paper that debunks the Glantz et al. claims of finding evidence of a gateway effect. Because the previous paper is somewhat redundant now, I suspect I will not further update it, but will just leave the existing version as an extended appendix of the present paper.
It is often claimed that low-risk drugs still create harm because of “gateway effects”, in which they cause the use of a high-risk alternative. Such claims are popular among opponents of tobacco harm reduction, claiming that low-risk tobacco products (e.g., e-cigarettes, smokeless tobacco) cause people to start smoking, sometimes backed by empirical studies that ostensibly support the claim. However, these studies consistently ignore the obvious alternative causal pathways, particularly that observed associations may represent causation in the opposite direction (smoking causes people to seek low-risk alternatives) or confounding (the same individual factors increase the chance of using any tobacco product). Due to these complications, any useful analysis must to deal with simultaneity and confounding by common case. In practice, existing analyses provide almost cartoon examples of drawing naïve causal conclusions from observed associations. The present analysis examines what evidence and research strategies would be needed to empirically detect such a gateway effect, if there were one, explaining key methodological concepts including causation and confounding, examining the logic of the claim, identifying potentially useful data, and debunking common fallacies on both side of the argument, as well as presenting an extended example of proper empirical testing. The analysis demonstrates that none of the empirical studies to date that purport to show a gateway effect from tobacco harm reduction products actually does so. The observations and approaches can be generalized to other cases where observed association of individual characteristics in cross-sectional data can result from one or several causal relationships.
Comment are welcome. I am violating my usual rule of widely circulating a working paper for a while before submitting to a journal (as is standard in truth-seeking sciences) because I am already overdue on a promise to submit it. But I will endeavor to incorporate suggestions when I revise it.