by Carl V Phillips
I have watched with dismay as the relationship with science in the e-cigarette advocacy world has continued to descend toward that found in tobacco control: embracing anything that supports ones political beliefs and rejecting anything that does not, regardless of the validity or scientific defensibility of the claims. While we are still a ways from parity — and it is certainly the case that pro-ecig misinformation tends to move the debate closer to reality even though it is wrong (because the anti-ecig misinformation is so widely believed) — it is hard to not extrapolate the trend. And even if it stops here rather than continuing downhill, we are not in a good place.
I am not talking so much about the chatter among the enthusiastic consumer community. It is inevitable that this is full of cheerleading and echo-chambers, and thus it is inevitable that a lot of misinformation is touted. The same is true for pretty much any issue of politics or lifestyle. It would be better if that were not the case, and one of the several purposes of this blog (along with others) is to try to reduce that tendency a bit. But this tendency to interpret scientific claims based primarily on the political implications of the conclusions is also common among ostensible experts. Going down that path is exactly how the movement that began with legitimate research and education about the harms from smoking turned into the fanatical and lie-based tobacco control industry we see now.
Consider the major anti-ecig broadside of the week, the claim that a new study shows a gateway effect from vaping to smoking among teenagers. This has been widely decried in pro-ecig circles. But just two days ago, those circles — in some cases, the exact same people — were touting the claim that another study had demonstrated there was no gateway effect. The thing is, the former claim is actually more legitimate than the latter.
They are both wrong. This would be obvious if we had a methodological analysis that points out what evidence would actually inform either of those claims. Oh, wait, we do have that. And it covers the fatal flaws in each. Yet I never saw it invoked in any of the commentary of these claims (though obviously I do not read everything). And, no, this is not about me wanting the citation — it would have been fine if someone offered such analysis by either borrowing my thoughts without attribution or re-deriving the points, but no one did.
Monday’s claim about there being no gateway effect was just dead wrong. There is literally nothing in the reported research that supports that claim. This is quite obvious on its face, but it also happens that I addressed the very fallacies employed in making the claim in my methodology paper. The short version is: The claim is based on two observations, that there are few teenage e-cigarette users who are even at risk of being gateway cases (i.e., they were not smokers in the first place) and that smoking rates are not going up. But the first of these obviously does not show there is no gateway effect, only that it has not had a chance to affect many people (yet). And this itself is the reason why the second is fallacious: even if, say, a quarter of the vapers in that at-risk group became gateway cases, it would be too few to show up in the smoking statistics.
By contrast, today’s claim is actually grounded in evidence that could be interpreted as supporting a gateway claim. Such a conclusion is still wrong, but it takes a bit of serious analysis to see why. The research found that teenagers who were exclusive e-cigarette users were more likely to later become smokers than teenagers who used no tobacco products. The pro-ecig commentators who dismissed the claim argued that this association could easily be explained by confounding: that the teenagers who chose to adopt e-cigarette use were more likely than average to become smokers for reasons that were not caused by their vaping. But the authors of the study, quite expectedly, preempted that simplistic response, pointing out that they controlled for some variables to try to correct for such confounding.
It turns out that “we controlled for that” is usually a bullshit claim in epidemiology; the standard statistical practice is inadequate. But simply crying “that association could have been caused by residual confounding!” is epistemic nihilism — you can always say that and thus use it as an excuse for refusing to accept a result you find objectionable. Thus, everyone is wrong at the simplistic level this conversation took place. Fortunately, we do not have to stop at the simplistic level. There are ways of assessing whether the “we controlled for that” claim is at all convincing — I cover them in my paper so will not repeat them here — and it turns out that the claim is very much not convincing in this case.
It is not that pro-ecig commentators were wrong in arguing the authors of that paper failed to rule out the obvious confounding as an explanation for the association. That statement is right. But it is not right because residual confounding is always possible and therefore you are free to ignore any study result you do not like. Such nihilism is exactly the behavior we observe from tobacco controllers when they want to dismiss a study result they do not like.
If this were explained by a culture of great scientific skepticism, it would be different. But pro-ecig advocates have also adopted the tobacco controllers’ habit of declaring that any study result that kinda sorta hints at support for their preferred hypothesis is sufficient for declaring it to be true. The pattern is that study results can be dismissed with hand-waving complaints about the methods being imperfect (as methods always are), unless the authors stated the right conclusion, in which case the methodology was just fine. Sound familiar?
There are various aphorisms along the lines of “choose your enemies carefully because you will eventually end up acting like them.” Frankly I find this fairly silly as a general statement. It is obviously not true that everyone ends up acting like their enemies, and when they do it is likely to be for organic reasons that have nothing to do with what the enemies did (e.g., soldiers do not commit war crimes because their enemy have done so, but because they are all put in similar horrific circumstances that bring out the worst devils of human nature). But it actually does seems to be causal in the present case.
Non-scientists in the e-cigarettes space are exposed to “science” mostly by reading tobacco control junk science and similar bad “public health” research. While they might reject the conclusions, their understanding of how scientific conclusions should be reached is poisoned. More important, many of the ostensible experts who are pro-ecig are tobacco controllers. They may be schismatic regarding the political party line on this one point, but their training and experience are still based in junk science. They simply were not taught any better in school, did not learn by example because they did not read outside their field, and never suffered any penalty for doing junk work because there was no critical review. And all this remains true after they went pro-ecig.
This brings us to the other big e-cigarette news of the week, today’s report published by Public Health England, which is being touted as The Best Thing Ever by many e-cigarette proponents. Really? My reading finds this a not-very-veiled stalking horse for medicalizing THR, which has always been a goal of tobacco controllers (and the authors very much tobacco controllers, despite being schismatics). In addition, the content makes a good argument for discouraging or even forbidding e-cigarette use as a reasonable lifestyle choice, relegating to merely being a lesser evil than smoking. This is tangential to the point of the present post (see my next post, which I had been working on for most of two weeks when this came along, for more on this point), but some specifics are exactly on point.
In particular, the report is presented under the headline, “E-cigarettes around 95% less harmful than tobacco estimates landmark review”. Just skip over the overblown self-assessment of this made-for-government-palatability review, whose main messages were already obvious to anyone who had been paying attention, to the number: 95%?? Where did that come from? Well it turns out that it was just made up from whole cloth, tracing to an exercise in making up numbers which is often mis-described as a study. (I excoriated that silly exercise here.) Yet the number is declared not merely to be a fact, but to be exactly right (there is no hint of an uncertainty interval around it).
The magnitude is absurd. If vaping really were really 5% as harmful as smoking, it would be the most harmful lifestyle choice almost any nonsmoker ever made, short of bundling all of “not eating optimally” into a single choice. It would be more harmful/risky than all of someone’s transport and travel, or being substantially overweight, or eating tons of sugar, or eating lots of fried foods, or dabbling with drug use, or most levels of heavy drinking, and so on. If I believed vaping really was that harmful, I would be encouraging vapers to quit, so long as the alternative was not returning to smoking or being very unhappy. If drinking coffee were 5% as harmful as smoking, I would certainly give it up, as much as I like it.
For vaping to be 5% as harmful as smoking, there would have to be particular diseases that produced that much harm. There is no such thing as “just harm”, without a specific disease pathway. But no one has made any plausible suggestions about what diseases those are, presumably because they would clearly be absurd. Still, supposed expert commentators made this claim and others seem unbothered by it. I am not sure whether to say this scientific failure should be attributed to pro-ecig advocates (who have gushed about the report and the original paper) or tobacco controllers (who are the ones who ultimately benefit from such a claim being accepted). I suppose it is not necessary to sort that out, because my point is that those groups demonstrate similar willingness to accept absurd scientific-sounding claims.
I have observed some pushback against this claim from the pro-ecig side, trying to rewrite it to say that 5% is the maximum possible level of risk. But this is not true either. The reasonable point estimate is a small fraction of that 5%, but we still cannot rule out that the risk is greater than 5%. It is unlikely, based on the knowledge we have, that inhaling the chemicals in e-cigarette vapor for a decade or two is all that harmful, but it is certainly not impossible, given that we have never observed these exposure levels before. Again, we are growing a culture of just making up and uncritically accepting scientific claims that quite easy to debunk.
It seems rather unlikely that this will work out well. When people look in at a squabble where both sides are, um, stretching the truth, those who are 90% honest do not actually look a lot different from those who are 40% honest.
It does not have to go this way. I noted that because the anti-ecig junk claims are so widely believed, it turns out that pro-ecig junk claims tend to push the messaging closer to the truth. But the fully accurate and defensible claims are also quite sufficient to do that. It is really not all that hard to get the science right. The obstacle is not a lack of ability to do so, but a distressing lack of an ethic that getting it right is the right and proper thing to do — ultimately better for the consumers whose welfare this is supposed to be about — and a lack of concern that flouting getting it right is ultimately toxic.