The concept of the causal pathway, the specific steps from an exposure or personal characteristic to an outcome, is critical in epidemiology and similar sciences. Best practice in modern epidemiology includes identifying the possible pathways before conducting research or analysis, often illustrating them with a flowchart called a causal diagram. Properly controlling for confounding, which I discussed in my previous Daily Vaper science lesson, is very difficult if this step is skipped.
But failing to think about causal pathways can also lead to simpler and even greater errors. An example is the paper by Robert West et al. that notoriously claimed that only 16,000 smokers quit because of e-cigarettes in England in 2014. The true number is more like 100,000. When I critiqued this recently, it became apparent that they ignored all but one of the four or five pathways via which vaping causes smoking cessation.
So how does vaping cause smoking cessation? None of this will come as news to anyone who understands vaping, but thinking about each pathway separately is useful for clarifying our understanding.
1. Trying to quit smoking by switching to vaping improves the chance of success. Some people would have tried to quit smoking using another method and failed, but instead try switching to vaping and succeed.
2. Vaping causes smoking cessation attempts that would not have otherwise occurred. Many smokers who would not have made a concerted attempt to quit during a particular month (or year, or perhaps ever) try because of vaping. So smoking cessation occurs not merely because switching to vaping works better than some alternative method, but because no method would have been tried without vaping.
2a. Some smokers become “accidental quitters” after they start vaping. They were not going to try to quit smoking, and also did not really plan to quit with vaping. Rather, they just start vaping and discover they are done with smoking. Arguably this is a version causing a cessation attempt (though it only becomes an “attempt” after it succeeds), but it is still useful to recognize it as a particular variation on that pathway.
3. Ongoing vaping prevents recent ex-smokers from returning to smoking. Most people who stop smoking for a week start smoking again within six months or a year. But by providing a satisfying substitute, not just a temporary push to stop, vaping dramatically reduces the benefits of returning to smoking.
4. Vaping can prevent relapse in established ex-smokers. Some ex-smokers start smoking again after a year of smoking abstinence. Some discover the option of taking up vaping, even though it did not contribute to them quitting in the first place. For them it can have the same effect as in point 3, causing temporary smoking abstinence to become permanent cessation.
Understanding the distinct causal pathways highlights the way many analyses understate the contribution of vaping to smoking cessation: They ignore some pathways. Almost all the focus in the academic literature is on the first pathway. Clinical trials measure only the first pathway, and sometimes a bit of the third. They ignore the effects of the the second pathway (both versions), though it probably accounts for most cases of vaping causing smoking cessation. Trials also ignore the fourth causal pathway, which is a smaller effect but not nothing. (Clinical trials of smoking cessations also have other major flaws that are not related to this).
The West et al. calculation assumed that the only causal pathway is the first on the list. It assumed that every cessation attempt that involved vaping would have happened in the absence of vaping, using some other method. It also assumed vapers were just as vulnerable to return to smoking as people using other methods, contrary to point 3. It excluded anyone who took up vaping after having already quit smoking, ignoring point 4. Those interested in the long-read version of my analysis of that paper, which also addresses other errors, can find it here.
The authors of that paper never acknowledge they are making these indefensible assumptions. Rather, it appears they did not even realize they were making them because they skipped the crucial step of identifying the causal pathways.
It is easy for vapers to overlook the usefulness of separating causal pathways when responding to detractors. Personally familiarity with all the pathways makes it tempting weave them together, arguing “and also…, and also…, and also…” in one figurative breath. It is sometimes more effective to carefully separate the pathways: “You underestimate how much vaping makes a cessation attempt more likely; but more important, you also ignore that vaping causes cessation attempts that would not have otherwise happened; also you ignore the value of vaping in preventing re-initiation of smoking.” This summarizes my criticism of the West paper.
Notice that none of this addresses the additional consumption value of vaping for ex-smokers, making them happier than they would be if completely abstinent. We can back up a step and see that there are multiple causal pathways from “vaping” to “improved welfare.” One of them – the only only one that tobacco control activists care about – is causing smoking cessation for people who prefer to be nonsmokers. But other pathways produce even more net benefits for many vapers. Once again, thinking about separate pathways can help improve our understanding and focus our arguments.
