by Carl V Phillips
Live by the sword….
A new study by Goniewicz et al. found that smoking and e-cigarette trialing[*] are both up in Poland. They conclude based on that (yes, just on that — my sentence fully sums up their results), “Observed parallel increase in e-cigarette use and smoking prevalence does not support the idea that e-cigarettes are displacing tobacco cigarettes in this population.” It turns out that simple sentence is wrong in its details (the trend was not remotely parallel) while right in its conclusion. But that is only because the conclusion is basically always true: There is no conceivable data from population usage trends that could either support or deny the conclusion that e-cigarettes are displacing cigarettes.
[*Important aside: The definition of “current smoker” in research is typically “have taken at least one puff in the last 30 days”. That is not a terrible simplification when the goal is to study smoking among adults. Most adults who have smoked once in the last 30 days are smokers by any reasonable definition; smoking is usually an established habit that is either done frequently or avoided entirely. It becomes rather more tenuous when we are talking about kids, because they are much more likely to dabble in such behaviors, so someone who has taken a puff recently may well not be a smoker by any natural definition. Still, there is a pretty decent chance he is because, again, it tends to be an established-or-never type behavior.
This contrasts sharply with e-cigarette usage, especially among kids. Survey designers have inappropriately copy-and-pasted the smoking question without learning enough to understand it is not a measure of the same phenomenon. Someone who has taken one puff on an e-cigarette in the last 30 days may be a vaper by any natural definition. Or she may be a smoker who keeps a cigalike in her purse and puffs on it once a week to relieve some of her suffering when she is trapped in a smoke-free venue for a long period. Or she may be transitioning off of cigarettes. Or she may not use any tobacco product and have approximately zero probability of ever doing so, but recently saw someone using an e-cigarette and asked to try a puff. The latter would seldom happen with a cigarette among adults. The contrast is that e-cigarettes are novel, interesting, and not unpleasant for nonusers to try, and informed people know that poses no more risk than trying a sip of someone’s coffee. We have all seen a nonuser sample an e-cigarette countless times. With kids — typically having more interest in novelty and broader social lives — that is going to happen much more.
Moreover the newness means that someone who takes a few puffs in their entire life, just to make the rational and normal human choice to try out the experience, is reasonably likely to have done it in the 30 days before the survey, unlike with cigarettes. Thus, to describe someone who has tried an e-cigarette in the last 30 days as “a current user” is clearly misleading.]
Back to the causal analysis of population data.
A disturbingly common trend among e-cigarette advocates is to claim that because smoking prevalence is going down where vaping prevalence or e-cigarette trialing is going up, it cannot be that e-cigarettes are a “gateway” to smoking. (I expect to be releasing a paper that covers that point within a few days.) Some take it even further and claim that because smoking rates are going down — as they have been for fifty years — it must be caused by e-cigarette use. This is wrong for the same reason that Goniewicz et al. are wrong: It is impossible, in such data, to sort out the effects of e-cigarettes from the many factors influencing smoking rates. Could e-cigarettes be a gateway to smoking, but smoking still decreases when vaping increases? Yes. Could it be that e-cigarettes do not actually reduce smoking, but smoking rates still go down? Yes. Could it be that e-cigarettes are making a big dent in would-be smoking, but smoking is still going up? Yes.
These are obviously not equally plausible claims. We have lots of evidence that shows that e-cigarettes are causing smoking cessation. But the population smoking prevalence tells us almost nothing. All those claims are equally wrong when the population prevalence data is cited as their basis, and for basically the same reasons.
Returning to Goniewicz paper itself, I have to chuckle at the first line of their press release, “Use of electronic cigarettes (e-cigarettes) has more than tripled among students in Poland, according to….” I wonder if they realize that any increasing trend that started at zero allows you to say “increased by a factor of X” where X is any positive number — you just have to pick the right starting and ending point. Ok, strictly speaking, because those numbers are quantum (the number of users must be a positive integer), you can only get X as high as the total number of users and you can only get a (large) finite subset of the rational numbers below it. Still, wouldn’t saying “…has increased 1000-fold…” be a much more impressive misleading statement? I wonder why they did not try that. (As long as I am enjoying math geek humor, I responded to a recent claim that e-cigarette use had doubled in the last decade with “ Inf”.)
Setting aside the innumeracy (and/or failure to notice they forgot to include the “in the last three years” or “among older teens” modifiers in their claim), the study found remarkably high rates in Poles, age 15-19. There are a few things that are quirky about the sampling methods and reasons to believe the populations sampled from 2010-11 and 2013-14 were not really the same, so I would not put too much stock in the exact numbers. That is something to keep in mind about most studies like this.
Smoking at least once in 30 days increased from 24% to 38%. It is really amazing what a sustained economic depression does to the use of all drugs by young people with no employment prospects.
The numbers for e-cigarette “current use” were 5.5% and 30%. That is the predictable trend, though the second figure is surprisingly high. Again, economic depression and all. Careful readers will have noticed that their describing that change as “more than tripled” is a bit odd. My three-year-old knows how to describe something as doubling or tripling but does not have the vocabulary for other multiples — maybe it was something like that.
Almost three-quarters of those who had recently tried an e-cigarette were also smokers, up from about 60%. This fits the downward trend in novelty I mentioned earlier: Over time, the portion of people who try an e-cigarette once or twice in their lives who have done that in the last 30 days will decrease, so more of them will be actual tobacco product users. From this we can conclude the trend is toward e-cigarettes being used almost exclusively by smokers and ex-smokers. Of course, the authors did not mention that obvious point nor, perhaps, even understand it. Also, it will probably surprise no one that they did not report the rate of former smoking among e-cigarette users, to avoid demonstrating that inconvenient fact.
The one good thing I can say about this paper is that they did not ramble on for pages about unrelated points, though the Discussion section is longer than either the Methods or Results, so that is still a fail. They discuss why their e-cigarette trialing rates were higher than those in another country where e-cigarettes are popular, the US. They nailed the obvious major source of the contrast (because the latter figures are based on ages 11-18), and yet go on to talk about other possible explanations.
So, what should the anti-THR liars have done with the main study result? If they were smart, they would have spun it (honestly) as a clear demonstration that it is invalid to use population-level trends to make claims about e-cigarettes helping people quit. After all, this is likely to be a rare or even unique case of smoking increasing in a population with a lot of e-cigarette use. Their best tactic would have been to use it as a demonstration that the many cases where smoking decreased while vaping increased do not provide evidence that e-cigarettes work.
That is both valid and tactical. It would be a serious mistake to get on record as saying “this shows that e-cigarettes are not causing cessation” — not just because it is a blatant lie, but because it provides a handy reference for everyone who wants to make the much more common opposite misleading statement.
Any guess what they did?
In an editorial accompanying the research paper, which is basically as long as the original paper, Dutra and Glantz wrote (…you can already guess where this is going…), “If adolescents were adopting e-cigarettes as alternatives to conventional cigarettes, conventional cigarette smoking…would have declined.” And a hearty thanks from everyone who wants to use declining smoking as evidence that e-cigarettes work to their pals Lauren and Stan. I cannot condone making arguments that are as bad as something Glantz would write, which that claim is. But if you are going to claim that declining smoking shows that people are adopting e-cigarettes as alternatives, be sure to cite Dutra and Glantz as agreeing with you.
I am not going to waste my time dissecting random rambles from Glantz. Down that path lies madness. But one point: The ellipses in the above quote stood for “…and dual use of e-cigarettes and cigarettes….” This is utter nonsense. There is not even a bad naive argument to be made that dual use should go down when e-cigarettes are substituted for smoking, and they do not even try to back that silly assertion. The reality, of course, is that dual use goes up sharply when substitution takes place because those engaging in partial substitution are necessarily dual users, while those transitioning off of cigarettes are also dual users during the transition period.
Ok, just one more, because if I am going to have to read Glantz we might as well get something useful from it:
Indeed, in its 2014 policy statement on e-cigarettes, the American Heart Association stated, “Among never smokers, 0.7% were currently users (past 30 days), which indicates that few never smokers who try e-cigarettes continue their use”. Results from Poland challenge this assumption (Table 1).
So apparently neither Dutra nor Glantz understands the difference between a simple observation about what is in a dataset and an assumption. Ok, we knew that. But if you look at their Table 1, in which they report a few other numbers based on the Goniewicz survey, they report that even for this population and its rather extreme numbers, only 2% of never-smokers tried an e-cigarette in the last 30 days, hardly a stunning contrast with the “assumption” they claim to deny.
But here is where the conspiracy gets really interesting. Recall how I mentioned that Goniewicz et al. fail to report what portion of the e-cigarette “users” were former smokers. They did not mention anything about former vs. never smoking in their results. So unless Dutra and Glantz were just making up numbers out of whole cloth (which seems unlikely — why commit that bright-line crime when it is so easy to get away with just lying about what the numbers mean), they were given access to data that was intentionally not shared with the readers of the original paper (nor the peer reviewers who signed off on it).
If only we could figure out what Goniewicz et al. did not tell us. Oh wait.
I thought it might take a bit of back calculating, but it turns out Dutra and Glantz accidentally provided new data that made it just a matter of simple subtraction. Of the population who puffed an e-cigarette in the last 30 days, from the 2013-2014 survey, 72% were smokers (reported by Goniewicz) and 21% were former smokers (and thus were probably using e-cigarettes for THR), with only 7% (reported by Glantz) being never smokers. The horror.
Of course, there is nothing terrible, nor even bad, about nonsmokers trying or even really using e-cigarettes. Some are would-be smokers who settled on a low-risk alternative instead (that number will grow steadily over time). Those who are would-be abstinent include a lot of people who will only try a few puffs ever, but happened to have done so just before the survey. But even if a would-be abstinent decides that using a low-risk product is better than being abstinent, where is the problem? They are making a rational consumption choice. Therein lies the fundamental problem with all of the rabid ranting from all these people.