Add new comment


Reverse causality: Another excuse to ignore the evidence falls

So far, most of the evidence on long-term adverse effects of e-cigarettes has been from cross-sectional studies, as our recent paper showing that people who used e-cigarettes were more likely to have had heart attacks, which represent a “snapshot in time.”

As a result one can only talk about “associations” not “cause” because there is always the possibility of so-called “reverse causation.”  If this case, reverse causation would exist if people who had heart attacks were subsequently more likely to start using e-cigarettes than people who had not had heart attacks.  In such a situation the association that we found would be due to this effect rather than e-cigarettes causing heart attacks.

Indeed, the possibility of reverse causation is one argument that e-cigarette enthusiasts have been using to dismiss  a wide range of studies implicating e-cigarettes on everything from health effects to the gateway effect they have for kids (starting with e-cigarettes, then adding or switching to cigarettes).

Holly Middlekauff and Jeffrey Gornbeing wrote a letter to the editor of American Journal of Preventive Medicine, where we published our paper.  Our response cites longitudinal data from the PATH study showing that people who have had a heart attack are no more likely to start using e-cigarettes than people who have not had a heart attack.  Our cross-sectional data also showed that e-cigarette use was not associated with making a quit attempt among people who had a heart attack.

Thus, reverse causality is unlikely to explain the association we found between e-cigarette use and having had a heart attack.  Significantly, this risk is on top of any risks of smoking cigarettes, which means that dual use – when people use both products at the same time (the most common pattern) – is more dangerous than using either one alone.

PS: Another word game that e-cigarette enthusiasts have use to criticize cross-sectional studies is use of the term “risk factor,” which they say implies causality.  It does not.  Indeed, the term “risk factor” was developed specifically to refer to associations in cross-sectional studies.

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.