Science Lesson: Clinical Trial Evidence Debunks The ‘Vaping Impedes Smoking Cessation’ Claim
One of the most damaging anti-vaping lies being told by tobacco controllers is that vaping or trying vaping causes smokers to become less likely to quit smoking. This is absurd on its face, given the low background rate of smoking cessation and the millions of former smokers who have switched to vaping. The evidence that supposedly supports the claim clearly does not do so. Rather the observed associations are caused by temporality errors (study subjects started vaping because they failed to quit smoking, not vice versa) and simple confounding (smokers who try vaping to quit or as a partial substitute are more devoted to smoking than average). But despite these errors being clear to experts, it can be worthwhile to look at what simple experiments tell us. This is especially true since tobacco controllers are intent on pretending they do not understand, and their goal is to trick people who genuinely do not understand.
Clinical trials offer an excellent way to test the hypothesis that vaping impedes smoking cessation. Before you check to make sure you read the byline correctly, read on.
Yes, I have previously explained that epidemiology is a science of measurement, not for answering yes-no hypothesis tests. It is not interesting to know that a chemical in vapor sometimes causes cancer; what matters is how large the cancer risk from vaping is. (As far as we know, it is zero. It is undoubtedly not exactly zero. But knowing it is minuscule is what matters, not which side of zero it falls on.) I have also explained that randomized clinical trials (i.e., medical experiments) are a lousy way to assess the effectiveness of smoking cessation via product switching, and of behavioral questions more generally. They are too rigid to replicate the real-world switching experience and the clinical setting does not mimic the social support that is usually involved in switching. Moreover, randomization eliminates the self-selection by people who find a particular alternative attractive; while eliminating self-selection is useful for some research, in this case it eliminates part of the real-world consumer choice we are trying to replicate.
The false claims that clinical trials are best and that we should focus on hypothesis testing are common ways of simplifying scientific concepts for nonscientists. Unfortunately, they morphed into claims that the simplifications are actual scientific rules.
We have all heard the myth that clinical trials are better for answering any question than other study methods. This myth was created to try to discourage physicians from arrogantly basing their assessments on their “professional expertise” (i.e., their nonsystematic observations of the few dozen people they have treated). The idea of statistical hypothesis testing was developed to overcome the problem of drawing yes-no conclusions from samples that were too small to provide reliable answers. But in both cases, the lesson “you need to pay attention to X, which would fix a mistake you are making” got morphed into “X is always the right thing to do.” Put another way, someone observed a persistent bad practice in scientific inference and created a “rule” to combat it. However, when this correction is applied in other situations, as if it really were a rule, it creates more bad scientific inference.
Previous science lessons explained that these are not actually rules. So is this science lesson saying those “rules” offer good advice after all? Yes. It is complicated. Doing scientific studies often involves a well-worn path where we can follow a recipe. But figuring out what research results to look at to draw scientific conclusions has no such recipes. It is, well, more art than science. It would be an ironic error to interpret a lesson that says “it is a myth that clinical trials are always better; there are no such universal rules” as offering a universal rule that “clinical trials are never useful for drawing conclusions about behavior.”
In this case, the claim is that if smokers who are considering quitting start vaping, it lowers their probability of quitting. To explore this hypothesis, we need to propose a pathway from the exposure (vaping) to the outcome (someone who would have otherwise quit smoking failing to do so). Tobacco controllers do not propose such causal pathways — both because they are in the business of creating propaganda rather than seeking knowledge, and because they do not understand how science works — so we need to do it for them.
The most obvious pathway would involve a smoker who was ready to quit because she suffered repeated periods of temporary withdrawal when smoking was prohibited. She was inclined to just finish the withdrawal once and for all, but discovered she was happy to vape when she could not smoke and thus eliminate the problem without quitting smoking. This scenario means that vaping still improved her welfare, perhaps quite a lot, but it did cause her to fail to quit smoking. (It is worth noting the hypocrisy of tobacco controllers who cheerlead for NRT, which is primarily purchased for this purpose, but believe this scenario represents a bad outcome. It is also worth noting that they are angry that someone made a choice that made her happier.)
The other possible causal pathways stray into the realm of magic. Perhaps something in vapor — or the way it delivers nicotine or something else about the experience of vaping — makes a smoker like smoking more than she already did. Thus she changes her mind about quitting. It is difficult to concoct a plausible scenario here, but this is the only other apparent causal pathway. Under these “magic” scenarios or under the withdrawal relief scenario, if smokers are effectively forced to try vapor products — as in a clinical trial — then it will impede smoking cessation.
Keep in mind that actually reducing smoking cessation on net — which is the claim — means that the rate of causing would-be quitters to not quit must exceed the rate of would be non-quitters switching (a rate we know is fairly high). But now we can take advantage of one of the failures of clinical trials: Because they do a poor job of mimicking people’s actual switching behavior, they do a poor job of encouraging switching. Thus, if vaping actually causes a net reduction of quitting in the real world, it will do so to an even greater extent in clinical trials.
No. Clinical trials in which some smokers are assigned to vape consistently show that they become smoking abstinent at a higher rate than the comparison group, so the hypothesis should be rejected. This includes the latest such study, as well as previous ones. Clinical trials are still a lousy way to try to estimate the real-world effectiveness of switching as a method of quitting smoking. But they are excellent at eliminating reverse-causation or temporality errors (which are responsible for a lot of the junk claims in question) and confounding biases (which are responsible for the rest of them). In addition, it happens that they are biased in the direction of supporting the “reduces cessation” hypothesis, so when they reject it — as they do — the evidence is even more definitive. There is simply no way to honestly claim vaping reduces quitting in light of these results.
When I was teaching courses, my students would often ask me questions like “is X a good method?” or “is Y the best measure to consider?” outside the context of a specific scientific question. I would always tell them “the quality of an answer is highly dependent on the question that is being asked.”
Is three a good answer? If the question is “what is twelve divided by four?” then it is an impeccable answer. If the question is “what is two-thirds of five cups?”, it is a good enough answer if someone is making soup but probably not good enough if he is baking. If the question is “are we there yet?”, it is a terrible answer, however tempting it might be to give.
In that spirit, if the question is “how do we estimate the probability that someone trying to switch to vaping will succeed?”, then a clinical trial is a lousy answer. But if the question is “how can we effectively test the claim that vaping impedes smoking cessation, and thus demonstrate that those who are claiming it are lying?”, then the clinical trials are the perfect tool.