Monthly Archives: September 2018

Sunday Science Lesson: Calling vaping/tobacco use an “epidemic”: it’s even stupider than you might think

by Carl V Phillips

A correspondent suggested to me that those who are not population health experts have a gut feeling that all this rhetoric about “epidemics” — of tobacco use, of teen vaping, and such — is an innumerate misuse of the term. But few really understand why.

The first problem with these claims, which I think most people get, is that “epidemic” refers to a disease, and these behavioral choices are not diseases. Of course, words get used metaphorically, and it is a somewhat complicated technical term that thus has hundreds of pop-level dictionary definitions floating around. But keep in mind that the people misusing the word in this case are supposed health science experts: The WHO entitles its flagship semi-annual reports on anti-tobacco policies, “WHO report on the global tobacco epidemic 20xx”. The U.S. FDA has been banging on about an epidemic of teenage vaping. This misuse of the terms frequently appears in public health journals. This is not ok.

Yes there are colloquial common-language uses of the term that are not limited to actual diseases (and there are also those that are narrower still and use the word to refer to infectious disease outbreaks specifically). But health “experts” and officials should not be using sloppy colloquial definitions of technical scientific words. It would be like economists using “efficient” to mean “quick and effective” or geneticists using “fitness” to refer to someone’s cardio statistics. (In case you do not know, each of those is an important technical term in its field, with a particular meaning.) It it similar to a physician using “cancer” or “poisoning” metaphorically when talking to a patient — “I’m afraid that you have cancer…. Of your motivation, which is keeping you from really focusing on your physical therapy.” (I have a recollection of Dr. Hibbert on The Simpsons having a conversation in which he keeps saying things like this, alarming the family for a beat before he makes clear he is not really meaning the words. Anyone know how to find a clip of that?)

So that alone is a simple, obvious fatal error in this usage. Anyone misusing the word to refer to a behavior they dislike, rather than a disease, and doing so in the context of health science, is engaging in propaganda rather than an attempt at accurate communication. But even if we set that aside — ignore the inappropriate metaphor of calling a behavior a disease — the use of the term is still blatantly incorrect.

For a disease to be in an epidemic state, it needs to have an incidence rate that is not necessarily high, but that is spiking above the normal baseline. (It also needs to be affecting a fairly broad population and not have a single source of exposure, as often happens with foodborne disease, or we instead call it an “outbreak”. But that is not really relevant for present purposes.) So, even though there are always a lot more heart attacks and HIV infections compared to Zika infections, Zika has recently been epidemic in some populations, while the others were not. The big numbers for the others are (usually) just the normal incidence rate. Exactly what is enough of a spike to qualify as epidemic is not precisely defined, but it is safe to say that (genuine) experts would not call a sudden jump of merely 10 or 20%, let alone a steady upward trend with similar increases, an epidemic.

So is tobacco use a global epidemic (accepting the metaphorical non-disease use of the word), as suggested by the WHO? Clearly not. It is actually in decline. In almost every population, the prevalence of tobacco use and, more importantly, the incidence of initiation are declining. I am pretty sure that there is not a single country where the current increases in tobacco use would qualify as an epidemic, though I might have overlooked somewhere. If you drill down enough, undoubtedly there are some subpopulations with recent spikes that would qualify as epidemic. But you have to look hard, and it is clearly not global.

(Aside: The number of smokers in the world has continued to increase, despite the decline in incidence and prevalence. The population is increasing faster than smoking rates decrease in all but the substitution-miracle countries. And of course, “tobacco use” does not decrease when product substitution reduces smoking. In addition, in a few large populations — extremely poor people who finally have enough income to afford tobacco products — rates are increasing, though not at epidemic levels. Bottom line: do not get misled by tobacco controllers when they temporarily switch their rhetoric from “epidemic!!!” to “we are close to eliminating smoking!!!” They are not.)

But how about more specific claims like FDA’s “epidemic of vaping among U.S. teenagers”? For that claim, the metaphor is even more strained (and, again, clearly inappropriate coming from an ostensibly scientific health agency): At least smoking can be metaphorically likened to a disease because it causes a lot of disease outcomes and so is similar to not-yet-morbidity-causing cases of an often-harmful infection. But vaping and other smoke-free tobacco use are approximately harmless. Saying “vaping epidemic” is a lot like saying “Fortnite epidemic”; yes, I suppose you can metaphorically refer to a sharp increase in the initiation of any consumption choice, but when the serious disease risks are trivial, it seems like a pretty stupid choice. You should just go with “sharp increase”. Though in cases like these, where the sharp increase in consumption is inevitable because no one was using the product a few years ago, even that is kind of stupid to say.

Worse is that this is an example of the “look at just one entry in the ledger” game that tobacco control rhetoric is notorious for. Compare their game when they pretend that smoking costs society money by toting up costs of treating (frequently fatal) smoking-caused diseases, ignoring the (almost exactly offsetting) reduction in the cost of treatment for some later disease that never happened because the person died from smoking. They also ignore other foregone consumption (housing, food, etc.) that results from earlier deaths, which add up to meaning that smoking’s health effects cause an enormous net savings in social resources. It is the same game used by those who say we cannot afford single-payer healthcare because it would cost $X trillion, and we cannot afford that — never mind that we are currently spending 30% more than that on healthcare, and would save that cost. It is as if someone said “I would eat out less, but I could not afford the resulting increase in my grocery expenditures.”

In the present case, FDA et al. ignore the decline in teenage smoking that offsets (and is pretty clearly caused by) the increase in vaping. What they are saying is equivalent to breathlessly panicking that we are experiencing an epidemic of a specific influenza strain, even though we were having an unusually mild flu season. It just happens that the year’s dominant strain is a relatively new mutation and there had not been many inflections with this particular strain in previous years, even though influenza is almost always more common and more harmful than in the present year.

FDA’s carve-out logic also means we are also experiencing an epidemic of teens smoking Marlboros that were manufactured in 2018, even though smoking is way down. I mean, no one was smoking those just a few years ago, and now they are, like, everywhere! Something must be done!

In short, it makes no sense to talk about an epidemic of a single option within a category of competing diseases/products. The entire category should be considered. (Notice the trap here for the crowd who seeks to die on the hill of “e-cigarettes are not in the tobacco products category.” This is a way you could actually die on that hill.)

The final problem with the use of “epidemic” — even ignoring for the inappropriate strained metaphor, the full-on falsity of the WHO’s version the claim, and the misleading tricks behind FDA’s usage — is more subtle. It is a question of what counts as a population.

An epidemic occurs when there is a spike in cases, across time, within a particular population. We do not say that Congo has an epidemic of malaria because they have a much higher incidence rate than Canada, or vice versa for frostbite. The word someone is probably looking for there is “endemic”. We only say “epidemic” if there is an increase in the numbers within the country. However, this is not about the place, but the group of people. So who constitutes the population, the group of people to compare over time, for tobacco product use?

Unlike influenza, tobacco product use is all about cohort replacement. That is, flu incidence changes from year to year because a different portion of the (mostly) same population get the disease. By contrast, population smoking prevalence changes mostly because the new cohort that is being added to the count (e.g., those turning 18 that year) has a different prevalence than those dying that year. Yes, there is smoking uptake by 19- to 25-year-olds (though that is still really a matter of cohort replacement, and would be clearly that if we took those FDA et al. like to call “youth” out of the adult population and looked only at prevalence for age 26+). Yes, there is also quitting at all ages. But year-to-year changes are driven by an entirely different engine as compared to infections sweeping through a population.

Notice that I have had to distinguish incidence (rate of new cases occurring) from prevalence (portion of the population who have the disease/behavior), a distinction that seems to baffle tobacco controllers even though it is first-semester public health. They can never decide which one they consider to be important. Sometimes they whine about rate of trialing (incidence of first-trying a product). Sometimes they whine about ever-use (prevalence of ever having ever trialed the product; this, of course, can only increase for a cohort over time), without seeming to understand the difference. Currently FDA seems to be making their “epidemic” claims about “used at least once in the last 30 days” prevalence.

Among their and WHO’s more subtle crimes against the word “epidemic” is that that word refers to spikes in incidence rates, not in prevalence. Consider that the prevalence of HIV is far higher now than ever before (thanks to maintenance treatments that let people live with it). Is this an epidemic? Similarly, the population prevalence of HPV-16 will peak when the vaccines become sufficiently widely used (and thus in future years the new cohorts are immune while some of those with the virus are dying off). So does this mean that the epidemic will be at its height at a time when the incidence rate is hitting its lowest point since the start of the sexual revolution? The innumerate use of the word that FDA is employing would say exactly that.

But getting back to the question of populations, the problem is more subtle still. Consider the population “Americans born in 2000”. When looking for epidemic-level increases, can we look at that cohort’s own incidence rate of vaping or even (ignoring the fatal problem noted in the previous paragraph) prevalence of recent usage? That obviously does not work, because of course those will be higher in last year’s statistics than they were in recent years when that population were little kids. That is like saying we are experiencing a huge increase in knowledge of simple calculus, because so many more in that cohort know it now than a few years ago. Is someone also going to whine that there is currently an epidemic of premarital sex among Americans born in 2000? (We should certainly hope there is!)

Even though this is really the only comparison that makes sense for properly using the word epidemic, it is obviously dysfunctional and so is not the comparison that gets made. Instead, incidence rates (or, more likely, prevalences) are compared, year-to-year, among 17-year-olds (or for whatever age cohort). But this is not a comparison within a population. It is, in fact, an entirely disjoint population, like comparing Congo and Canada; those who are 17 on a particular date in 2017 include no one from the population who was 17 on that date in 2016. So you can say, e.g., that vaping among this year’s 17-year-olds is much higher than among last year’s (just as you can say malaria is much more common among Congolese than Canadians), but that is not an epidemic. That is cohort replacement.

I realize this is subtle, and its importance is probably lost on many readers. But believe me when I say for anyone literate in population health science, it stands out as a far larger error in the use of the term than the simple fact that vaping is not a disease, or even the “looking at only one line in the ledger” game.

To summarize how I believe we should respond to these innumerate “epidemic” claims: First, we should push back against the use of the word to refer to a behavior rather than a disease. This is not, however, because of some naive language purity urge, a failure to recognize that words get used metaphorically. Rather, it is because this usage is part of tobacco controllers’ game of trying to define the behavior as a disease. They are not actually trying to expand the definition of “epidemic” here, but that of “disease”, and we should push back.

Second, the strongest substantive replies are as follows: The FDA version of the claim is based on carving out one particular product, which is taking away market share from other products (which, oh by the way, are a hundred times worse for you). Always go back to the entire category, and perhaps consider noting the analogy of “Marlboros manufactured in 2018”, whose usage is up by infinity percent. As for the WHO version of the claim, it is simply factually false.

Third, the FDA version of the claim is actually something worse than false: it is nonsense. It is one thing to say something that could be true but happens to be false. It is another to utter a string of words that simply make no sense. There cannot be an epidemic of anything among 17-year-olds, based on year-to-year comparisons, because these are entirely disjoint populations. Again, you may have to take my word for it, but this is actually the clearest misuse of the word in this entire embarrassing mess. The supposed health scientists at FDA should understand this, though I expect they do not understand and are clearly immune to embarrassment.

Science lesson: The absurdity of “n deaths per year” and “leading preventable cause” claims about smoking.

by Carl V Phillips

Smoking is quite harmful. Lots of people choose to do it. Given these facts, you would think that people who warn/scold/fret about smoking, at the individual or population level, would see no reason to exaggerate. Yet they do. They lie constantly and habitually. Still, in spite of the lying, you might think that they would avoid making mantras of claims that are simply nonsense. Yet they do not.

I have covered most of this before, highlighting some of it as one of the six impossible things tobacco controllers believe, but have not pulled it all together before.

Consider first claims like “smoking causes 483,456.7 deaths per year in the U.S.” What does this even mean? It obviously does not mean what it literally says, that but for smoking, these individuals would not have died. Occasionally someone asserting these figures phrases the claim in a way that highlights the implicit suggestion of immortality, and is rightly ridiculed for it. But in fact, even the standard phrasing implies this if treated as natural language.

Understanding what this might(!) really mean requires understanding the epidemiology definition of causing a death, which, it is safe to say, few of those reciting the claims about smoking understand. This definition is actually, like much of epidemiology, fundamentally flawed, but it gets us closer to something meaningful. The textbook definition is that something is a cause of death if it made the death occur earlier than it otherwise would have. Notice that this means that every death (like every event) has countless causes. E.g., a particular death may have been caused (in this sense of it occurring when it did and not later) by all of: smoking, being born male, not eating perfectly, occupational exposures, and choosing a low-quality physician. (Notice that if we extend to a broader definition of causation, other causes include the evolution of life on Earth and the individual’s grandfather making it home from the war.)

This typical version of the definition is fairly useless because it includes exposures that caused the death to occur only a few seconds sooner than it would have. We are seldom interested in those. Indeed, by that definition, smoking is a cause of death for almost every smoker and former smoker. It is very likely that any smoker who who is not killed instantly by trauma would have survived longer because whatever disease killed her would have developed more slowly, or simply because the body would have functioned for a few more minutes. So more useful definitions of a cause of death would be something that we estimate caused the death to occur a month, or a year, or five years earlier than it would have. Note that a far more useful measure, in light of these problems, is “years of potential life lost” (YPLL).

So which of those definitions is the “X deaths per year” claim based on, given that it is clearly neither the literal meaning (with its implication of immortality) nor the faulty textbook epidemiology definition (which would include approximately all deaths among ever-smokers)? The answer is: none of them. Those statistics are actually a toting up of deaths attributed to a particular list of diseases, each multiplied by an estimate of the portion of those cases that were caused by smoking, in historical U.S. populations. That is, it is the number of lung cancer deaths among smokers, multiplied by the portion of such deaths that are attributed to smoking, plus the number among former smokers multiplied by the attributable fraction for former smokers, plus those for heart attacks, plus those for a few dozen other specific declared causes of death.

As you might guess, based on who is doing the toting, these numbers are biased upwards in various ways. Still, it would be possible to estimate that sum honestly (no one has tried to do so for a few decades, but it would be possible). But the resulting measure would obviously not properly be described “deaths caused by smoking.” It would not be that hard to identify what the figure really is, especially in serious written material like research papers or government statements: “each year in the U.S. smoking is estimated by the CDC to cause X fatal cases among 29 diseases.” Of course, most “researchers” and “experts” in the field do not even know this is what they are trying to say.

There are also several problems with the numbers themselves, not just the phrasing. First there is the noted, um, shading upward of the numbers. Second, as I alluded to in the third paragraph, the statistic is always presented with too much precision. Even two significant digits (e.g., 480,000) is too much precision. The estimates of the smoking-attributable fraction of cases of those diseases are not precise within tens of percent for smokers, let alone former smokers, a much more heterogeneous category, at best, making even one significant digit (e.g., 400,000) an overstatement of the precision.

Third, and more important for most versions of the statistic is that “in historical U.S. populations” bit. The statistics you seen for other countries or the whole world are based on implicit assumptions that everyone share Americans’ health status and mix of exposures, because almost all the estimates come from U.S. studies (and those in the mix that do not are almost all from the countries that are most similar to the U.S.). At best, the estimated increase in risk for fatal cases of the disease ported to calculation for other populations, even though this varies across populations. That is, it is assumed that if the estimate is that half of all heart attacks among ever-smokers are caused by smoking, then that same multiplier is applied to heart attacks among ever-smokers in the other population. Worse, sometimes the attributable fraction itself is just ported, so if a quarter of all heart attacks in the U.S. are attributed to smoking, then that multiplier is applied to all heart attacks in other populations. That would mean, e.g., if a particular population has a lot of extra cases of cancers due to diet, the same fraction of those cancers that is due to smoking in the U.S. is attributed to smoking there.

Fourth, and worse still, the forward-looking versions of the statistics would be innumerate nonsense even if none of the other problems existed. These include the infamous prediction of a billion deaths from smoking in the 21st century, as well as assertions about the fate of cohorts who are taking up smoking now. The number of deaths from a list of diseases that are attributable to smoking is going to vary hugely not just across populations, but time. This is first-week Epidemiology 101 stuff. Population and time matter. There are no constants in epidemiology. The number of deaths from particular diseases will vary with technology. The attributable fraction will vary with the prevalence of other risk factors. Oh, and for those other changes, good news often makes things “worse”: An asteroid destroys higher life on Earth, and smoking stops causing any deaths. War, hunger, and infections are reduced and smoking causes a lot more cases of fatal diseases.

In summary, these statistics are: (a) not actually the number of deaths caused by smoking, (b) exaggerated, (c) far less precise than claimed, even setting aside the intentional bias, (d) only valid for a few populations, and (e) only applicable to the present (or, really, the recent past).

Moving on to the “leading preventable cause of death” claims, this mantra is equally absurd if you pause to actually look at the words. What does “preventable” mean? Typically in such contexts, it means “some obvious top-down action could have averted it.” So, for example, of the 3000 deaths from Hurricane Maria, a few score were hard to do much about. Every one of these was “preventable” in some sense (fly the particular person to Miami in advance of the storm) but this is meaningless; preventing someone, probably a few dozen someones, from getting killed was not a real option. But the vast majority of those deaths were meaningfully preventable — in the sense that an operationalizable action could have kept them from happening — with a competent relief operation.

So if this normal use of the word is what tobacco controllers mean when they recite this mantra, then they are basically testifying that they are horrifically incompetent. They spend their lives trying to prevent this from happening, and they fail even though it is doable. But while it is true that they are generally horrifically incompetent at what they do, it is clearly not doable. Smoking is not preventable by this standard sense of the word.

Perhaps they are saying it is theoretically preventable, in that sense, but no one has figured out how to do it. At least that is plausible, but then the full statement is clearly false. There are more important causes of death that are theoretically preventable. Deterioration with age of cellular repair mechanisms seems to pretty clearly top the list. Humanity will figure out how to largely prevent that. This bit of prevention (in the “we will figure it out eventually” sense) dwarfs preventing the deaths by smoking. Indeed, it will prevent a lot of the fatal disease cases that are caused by smoking. (I have a vision of one of my kids find this post in an archive 200 years from now, and being sad that this technology came a few decades too late for me. And for most of you too — sorry.)

Most likely, what they are not-quite-saying is that each individual who “dies from smoking” (i.e., has a fatal case of a disease that was caused by smoking) could have made a choice to not have that happen. In some sense, this suffers from the same problem that such a claim about hurricanes or earthquakes does: Yes, every death from a collapsed building could have been prevented by the person choosing to be in a different building. But it has a bit more legitimacy since it is obvious what the safer choice is and the risk is high enough probability to influence the decision. The problem here is to make this meaningful statement, tobacco controllers would have to acknowledge that smoking and other tobacco product use is an individual choice. They are not willing to say that out loud — and thus admit that their entire enterprise is devoted to keeping people from making the choices they want — so they hide it behind weasel words like “preventable”.

But just because the statement “the leading cause of death among individual behavioral choices” is meaningful does not mean it is right. Indeed, it is obviously wrong. Go back to the epidemiology textbook definition of a cause of death. Smoking is a cause of death, by that definition, for approximately everyone who smokes. But eating a less-than-optimal diet is, for the same reason, a cause of death for everyone who eats less than optimally. Two or three times as many deaths occur among people who ate less than optimally (i.e., basically everyone), as compared to those who smoked, so smoking is clearly not “leading”. Of course, no one really thinks in terms of that textbook definition. So how about if we limit it to deaths that occurred a year earlier than they would have. It is pretty difficult to imagine figuring out the numbers, but I would expect diet still has the edge. How about five years? At that level, smoking might really be leading. How about putting it in terms of YPLLs? Yes, it is probably true that smoking costs more YPLL than any other individual choice.

Aha, so they are right!

Um, yeah. We just have to assume that these stupid phrases really represent deep and subtle thinking on the part of those using them. By “preventable” they actually mean resulting from individuals’ behavioral choices. By “cause of death” they actually mean cause of YPLLs. And their declaration that it is true, rather than speculation, is based on valid estimates of the comparative number of YPLLs from different behavioral choices, even though they never cite such evidence. Also, by “n deaths” they mean “n cases of particular fatal diseases attributed to smoking, if you believe our numbers, and assuming that future looks exactly like the past and all population are like the U.S.” Giving someone the benefit of the doubt is sometimes noble, but it would just be silly in this case.

The bottom line are that these mantras are just as false as much of the rest of what tobacco controllers claim. Moreover, they are not just factually wrong, but are a demonstration of just how thinking-free the whole endeavor is. At least things like “second-hand smoking causes 30% of all heart attacks” or “vaping is causing more kids to take up smoking” are meaningful claims. They are obviously false, but they are valid hypotheses and are only false because empirical evidence shows they are false, not because it is impossible for them to be true based on some simple fundamentals of how we know the world works.

Sure, people say things all the time such that, if anyone paused to think and ask the question, would not stand up to a “what does that even mean?” query. We are not always precise in all our thinking, let alone how it translates into words. But the claims in question are not fleeting thoughts or ad hoc word choices. They are mantras that getting written or said a thousand times per day by supposedly credible people in supposed credible contexts. The fact that they cannot pass a “what does that even mean?” test is one of the greatest overlooked testaments to the fundamental lack of seriousness in public health. The fact that they get repeated by others is a testament to how influential sloppy public health thinking is, even over those who are attempting to position themselves as opponents of it.

A subtle tobacco control self-contradiction lie, re FDA pumping cigarette stock prices

by Carl V Phillips

Tobacco controllers contradict themselves all the time. That is the inevitable result of them saying whatever seems expedient at the time, without any concern for whether the evidence supports it, or even even flatly contradicts it. When someone is sociopathic enough to do this (*cough* Trump *cough*), they will not only contradict the evidence, but (unless they have incredible discipline and intelligence, which they do not) also inevitably contradict themselves. Many of tobacco control’s self-contradictions are quite simple, with patently contradictory statements appearing in the same document, or even the same paragraph. It hardly seems worth searching out the contradictions that require analysis and observations across multiple threads. But this one is kind of interesting.

A classic tobacco control trope, which you still see a fair bit, is that tobacco companies have to recruit new generations of smokers to replace their current customers. Most of those reciting this probably actually believe it, which reflects public health people’s fundamental lack of awareness about how the world works. Anyone familiar with business (and I mean at just the level of reading the newspaper) will know that markets these days hardly look beyond the next quarter’s earnings. There is not the slightest interest in future generations, and barely any interest in two years from now. C-suite executives respond mainly to these very-short-term incentives. Even stakeholders in the company – medium- to high-level employees who are planning on working there for another couple of decades — do not care about selling products to future generations.

Similarly, an economic theory or long-term shareholder perspective says companies should not care about future generations. Even the most modest discounting of future profits makes sales to upcoming generations approximately worthless in present value terms. Shareholders would rather the companies buy back shares rather than investing in “recruiting” future smokers. If you believe in an anthropomorphic view that cigarette companies “want” to stay in business – rather than making the economically rational choice of maximizing profits as far as they go and sunsetting – then they would be better off expanding horizontally into other logistics businesses (including other tobacco products) rather than worrying about whether anyone is smoking in 2060.

But let’s assume that tobacco controllers are sufficiently innumerate about business that they actually believe that companies “need” to recruit new generations. Then it cannot possibly be that they also believe this:

FDA’s attacks on the vapor product industry drove up cigarette company stock prices because they portend less competition for cigarettes over the time horizon that actually matters to the markets, a couple of years. The market cap increase reflects the expectation that the companies will be able to sell cigarettes for a higher per-unit profit and sell more cigarettes (the former is more important in terms of profits — another simple market fact that tobacco controllers do not understand — but the latter is what matters most in terms of social impacts).

Here’s the thing: If someone believes that the companies (i.e., their shareholders) actually care about selling to future generations, as they have claimed for decades, and believes that vaping will cause future generations to smoke, as they claim in this tweet and frequently, they they would have to predict that threatening to shut down the vapor product industry would depress share prices, or at least not send them through the roof.

As I said, tobacco control lies of self-contradiction are typically so blatant that there is no reason we have to dig this deeply. I certainly do not want to give tobacco controllers too much credit, by implying that they ever actually assess their hypotheses against the evidence. Still, it does not hurt to run through the scientific implications of what they say to illustrate the layers of their dishonesty.

All people like better products. Teenagers are people. Therefore….

by Carl V Phillips

So today FDA Commissioner Gottlieb is pumping cigarette company stock prices by threatening to ban flavors in vapor products (or something — not entirely clear), unless the manufacturers magically get teenagers to switch back to smoking instead (or something — not entirely clear). I wanted to address one aspect of this rhetorical game that does not get talked about enough. I doubt there is any serious observer of this space who does not get this, but much of what is said seems to overlook it rather than drilling down to it as it should.

The prohibitionist’s simplest rhetorical game here is to confuse “this product feature is appealing to teenagers” with “this product feature is particularly or uniquely appealing to teenagers.” But there is a deeper game, trying to cement the premise that intentionally lowering product quality is a good thing. This applies not just to interesting flavors of e-liquid, but also everything from attractive packaging to convenient unit quantities. The standard response to the “teenagers like flavors” rhetoric is to counter that adults like them too, and thus they seem to be critical for smoking cessation. Both systematic data and a deluge of testimonials make this point. It is a great point, and those making it are doing a great job.

However, the prohibitionists at FDA and elsewhere are obviously not unaware that adults also like and buy interesting flavors. Similarly, adults and teenagers both like it that e-cigarettes are less than five kilograms and come in colors other than day-glo orange. They like it that they are affordable, that cartridges last for a while, and that the devices do not burn your lips. They like it that there is no regulation that says tobacco products must be smeared with feces before they are packaged. All of these are aspects of product quality. The same features that make a product appealing to people (and thus, the banning of which would make them less appealing to people) make it appealing to teenagers. It turns out that teenagers are very similar to people, and many would argue that they are people. Lower the quality of the product, and fewer teenagers will choose to consume it. Fewer adults too. This works for food, movies, and pens also. There is no magic here.

The magic exists entirely in the rhetoric, in which the prohibitionists trick people into endorsing (or at least not actively pushing back against) their underlying premise: Intentionally lowering product quality is a good thing because it discourages teenage use. Never mind that intentionally lowering people’s welfare is a phenomenally radical action for a government to take, one that ought to be based on a lot of open and honest analysis, not sneaky rhetoric. I find it is a useful clarifying thought to replace whatever quality-lowering regulation is being debated with “mandatory smearing with feces” (assume the feces are sterilized so they are not a health hazard): If it is okay to intentionally lower the product quality by doing X (flavor bans, “plain packs”, punitive taxes, etc.), then it must be okay to mandate feces smears.

Consider the usual scientific response to flavor ban proposals, that there is no evidence that particular flavors or categories are particularly appealing to teenagers. This is accurate; there is no such evidence and no reason to believe it is true. If someone wanted to lower vapor product quality in a way that particularly affected teenagers, perhaps the orange coloration or increased mass options would be the better bet. After all, isn’t the usual claim that teenagers are taking advantage of the products being so subtle that they can hide them from parents and teachers? Adults would not like ugly heavy products, but they could deal with them.

The thing is that FDA et al. are not actually claiming that the flavors are particularly appealing to teenagers, just that they are appealing. This is obviously true (see above observation that teenagers are very much like people). A casual reader might conclude they are claiming that this is a targeted lowering of quality that affects teenagers but not adults. In fact, the serious actors in the space seldom actually claim that, and when they do it seems usually to be a matter of sloppy word choice. They do not actually consider it a problem that a regulation lowers the appeal of a product for everyone (and thus hurts all consumers). To them, this is a feature, not a bug. They want to ruin the products for everyone.

In getting opponents to go along with their fiction that this is not their motive, they win their greatest victory. One of the important skills of a conman like Scott Gottlieb is to get people to adopt his hidden premises without him ever stating them, let alone defending them. When the arguments hinge on “but adults like flavors just as much as teenagers do”, they effectively concede a key prohibitionist premise: If there were a way to intentionally lower product quality, such that it hurt teenage consumers more than adult consumers, then doing it would be fine. Not just fine, but good or even clearly the right thing to do. No doubt there are some vape advocates who accept that, but presumably most are not ready to agree that their e-cigarettes should have to look like traffic cones. But by just fighting the empirical claim (which is not actually even being claimed), they are often implicitly endorsing the normative premise.

Some advocates lead with the message that there are already laws about teenage access and these just need to be enforced. This is good in that it does not endorse the premise that it goes without saying that harming adults for the good of the chiiiildren is  justified (though usually this is not explicitly stated). The problem is that Gottlieb has cleverly turned this on its head, and threatens to hurt adults if they do not somehow better enforce the government’s laws, magically figuring out how to do what the government has never been able to do with cigarettes. Today’s rhetoric was mostly threatening the industry (though it is consumers who would suffer, of course), but he has directed that same demand at vapers themselves. Those who have been tricked into endorsing the underlying premises are cornered by this. They have effectively already conceded that destroying product quality is acceptable if minor bans cannot be enforced.

Advocates need to do a better job of backing a few steps up the prohibitionists’ chain of reasoning, rather than being tricked into conceding so much ground. Every argument should begin with the observation, “this policy is about intentionally harming people (vapers, smokers, other product users).” This should always be pointed out, because in itself that is a radical use of government power that should not pass without comment. It should be followed with a demand for an answer to, “by what right do you harm me/adult consumers/your citizens, even if it is true that this harms others more and harming them is a good thing because it changes their behavior?” Only after making those observations, and trying to never let the audience forget them, is it time to add “discouraging teenage vaping probably encourages teenage smoking”, “there the evidence does not support your implicit claim that teenagers like flavors better than adults do”, and other arguments about the scientific facts.

Let’s try to get our criticisms right, shall we? (More on the recent “vaping causes heart attack” study)

by Carl V Phillips

Sigh. We are supposed to be the honest and scientific ones in the tobacco wars. But we won’t be if we are not, well, scientific. Case in point are the criticisms of the recent paper with Glantz’s name on it that has been erroneously said to suggest that vaping doubles the risk of heart attack.

Incidentally, the meaningless statistic in the paper is a RR of 1.8, which is not double. Also, when the paper was originally written as a student class project (not by science students, mind you, but by medical students), that statistic was 1.4. That was when Glantz heard about it, managed to get the kids to put his name on the paper, and taught them how to better cook their numbers. That “contribution” has him being called the lead author.

The paper is junk science. So are most of the criticisms of it. If only someone with expertise in these methods had written a critique of it that people could look to. Oh, wait, here’s one in The Daily Vaper from February. That was based on a poster version of the paper, but as I noted in the article, “It has not yet appeared in a peer-reviewed journal, but it will, and the peer-review process will do nothing to correct the errors noted here.” I wish I could claim this was an impressive prediction, but it is about the same as predicting in February that the sun will rise in August.

You can go read that if you just want a quick criticism of the paper, and also look at the criticism on this page of some hilarious innumeracy Glantz piled on top of it. In the present post I am mostly criticizing the bad criticisms, though at the end I go into more depth about the flaws in the paper.

About half the critiques I have seen say something along the lines of “it was a cross-sectional study, and therefore it is impossible to know whether the heart attacks occurred before or after someone started vaping.” No. No no no no no. This is ludicrous.

Yes, the data was from a cross-sectional survey (the 2014 and 2016 waves of NHIS, mysteriously skipping 2015). And, yes, we do not know the relative timing (as discussed below). But “therefore it is impossible to know” (or other words along those lines)? Come on. A cross-sectional survey is perfectly capable of measuring the order of past events. Almost every single cross-sectional survey gives us a pretty good measure of, for example, whether someone’s political views were formed before or after the end of the Cold War. Wait! what kind of wizardry is this? How can such a thing be known if we do not have a cohort to follow? Oh, yeah, we ask them their age or what year they were born. Easy peasy.

Almost every statistic you see about average age of first doing something — a measure of the order in which events occurred (e.g., that currently more Americans become smokers after turning 18 than before, but most extant smokers started before they were 18) — is based on cross-sectional surveys that ask retrospective questions. It is perfectly easy to do a survey that asks heart attack victims the order in which events occurred. Indeed, any competent survey designed to investigate the relationship in question would ask current age, age of smoking initiation and quitting, age of vaping initiation and quitting, and age at the time of heart attack(s), ideally drilling down to whether smoking cessation was just before or just after the heart attack if they occurred the same year. We would then know a lot more than the mere order. But NHIS does not do that because, as I noted in the DV article, it is a mile wide and an inch deep. It is good for a lot of things, but useless for investigating this question. It can be used, as it was here, for a cute classroom exercise to show you learned how to run (not understand, but run) the statistical software from class. But only an idiot would think this paltry data was useful for estimating the effect.

(A variation on these “therefore it is impossible” claims is the assertion that because it is a cross-sectional study, it can only show correlation and not causation. I am so sick of debunking that particular bit of epistemic nonsense that I am not even going to bother with it here.)

So, we do not know the order of events. We can be confident that almost all the smokers or former smokers who had heart attacks smoked before that event. We do not know whether subjects quit smoking and/or started vaping before their heart attacks. Given that vaping was a relatively new thing at the time of the surveys, whereas heart attacks were not, it seems likely that most of the heart attacks among vapers occurred before they started vaping. This creates a lot of noise in the data.

A second, and seemingly more common, erroneous criticism of the analysis is that this noise has a predictable direction: “Smokers had heart attacks and then, desperate to quit smoking following that event, switched to vaping, thereby creating the association.” Again, no no no. Heart attacks do cause some smokers to become former smokers, but there is little reason to believe they are much more likely than other former smokers to have switched to vaping. Some people will have heart attacks and quit smoking unaided or using some other method. Indeed, I am pretty sure (not going to look it up, though because it is not crucial) that most living Americans who have ever had a heart attack experienced that event before vaping became a thing. So if they quit smoking as a result of the event, they did not switch to vaping. Also it seems plausible that the focusing event of a heart attack makes unaided quitting more likely than average, as well as making “getting completely clean” more appealing.

Of course, an analysis of whether behavior X causes event Y should not be based on data that includes many Y that occurred before X started. That much is obviously true. NHIS data is not even a little bit useful here, which is the major problem. There is so much noise from the heart attacks that happened before vaping this that the association in the data is utterly meaningless for assessing causation.

But there is no good reason to assume that this noise biases the result in a particular direction. If asked to guess the direction of the bias it creates, a priori, I probably would go in the other direction (less vaping among those who had heart attacks compared to other former smokers). The main reason we have to believe that the overall bias went in a particular direction is that the result shows an association that is not plausibly causal. We know the direction of the net bias. But this is not the same as saying we had an a priori reason to believe this particular bit of noise would create bias in a particular direction. When we see a tracking poll with results that are substantially out of line with previous results, it is reasonable to guess that random sampling error pushed the result in a particular direction. But we only conclude that based on the result; there was not an a priori reason to predict random sampling error would go in a particular direction.

Moreover, we do not have any reason to believe that the net bias was caused by this particular error, because it has a rather more obvious source (see below).

Sometimes we do have an a priori reason to predict the direction of bias caused by similar flaws in the data, as with the previous Glantz paper with an immortal person-time error (explained here, with a link back to my critique of the paper). If the medical students had engaged in a similar abuse of NHIS data to compare the risks of heart attack for current versus former smoking, then the direction of bias would be obvious: Heart attacks cause people to become former smokers, which would make former smoking look worse than it is compared to current smoking. I suspect that people who are making the error of assuming the direction of bias from the “Y before X” noise are invoking some vague intuition of this observation. They then mistranslate it into thinking that former smokers who had a heart attack are more likely to be vapers than other former smokers.

This brings up a serious flaw in the analysis that I did not have space to go into in my DV article: The analysis is not just of former smokers who vape, but includes people who both smoke and vape, as well as the small (though surprisingly large) number of never-smokers who vape. If vaping does cause heart attacks, it would almost certainly do so to a different degree in each of these three groups. For reasons I explored in the previous post, different combinations of behaviors have different effects on the risk of an outcome. Vaping probably is protective against heart attack in current smokers because they smoke less than they would on average. If a smoker vapes in addition to how much she would have smoked anyway, the increased risk from adding vaping to the smoking is almost certainly less than the (hypothesized) increased risk from vaping alone. Whatever it is about vaping that increases the risk (again, hypothetically), the smoking is already doing that. Thus any effect from adding vaping to smoking would be small compared to the effect from vaping compared to not using either product. Most likely the effect on current smokers would be nonexistent or even protective.

Indeed, this is so predictable that if you did a proper study of this topic (using data about heart attacks among vapers, rather than vaping among people who sometime in the past had a heart attack; also with a decent measure of smoking intensity — see below), and your results showed a substantial risk increase from vaping among current smokers, it would be a reason to dismiss whatever result appeared for former smokers. This is especially true if the estimated effect was substantial in comparison to the estimate for former- or never-smokers. If you stopped to think, you would realize that your instrument produced an implausible result, and thus it would be fairly stupid to believe it got everything else right. This is a key part of scientific hypothesis testing. Of course, such real science is not part of the public health research methodology. Nor is stopping to think.

It is a safe bet that the students who did this analysis understand none of that, having never studied how to do science and lacking subject-matter expertise. Glantz and the reviewers and editors of American Journal of Preventive Medicine neither understand nor care about using fatally flawed methods. So the analysis just “controls for” current and former smoking status as a covariate rather than separating out the different smoking groups as it clearly should. This embeds the unstated — and obviously false — assumption that the effect of vaping is the same for current, former, and never smokers. Indeed, because “the same” in this case means the same multiplicative effect, it actually assumes that the effect for current smokers is higher than that for former smokers (because their baseline risk is higher and this larger risk is being multiplied by the same factor).

Though they did not stratify the analysis properly, it is fairly apparent their results fail the hypothesis test. The estimate is driven by the majority of vapers in the sample who are current smokers, so they must have had a substantially greater history of heart attacks.

There is a good a priori reason to expect this upward bias, as I noted in the DV article, but it is not the reason voiced in most of the critiques. It is because historically vapers had smoked longer and more than the average ever-smoker. This is changing as vaping becomes a typical method for quitting smoking, or a normal way to cut down to having just a couple of real cigarettes per day as a treat, rather than a weird desperate attempt to quit smoking after every other method has failed. Eventually the former-smoking vaper population might look just like the average former-smoker population, with lots of people who smoked lightly for a few years and quit at age 25, and so on. But in the data that was used, the vapers undoubtedly smoked more than average and so were more likely to have a heart attack (before or after they started vaping).

Controlling for smoking using only “current, former, never” is never adequate if the exposure of interest is associated with smoking history and smoking causes the outcome, both of which are obviously true here. If there are no such associations then there is no reason to control for smoking, of course. Thus basically any time you see those variables in a model, you can be pretty sure there is some uncontrolled confounding due to unmeasured smoking intensity. In this case, you can be pretty sure that its effect is large and it biases the association upward.

In short, the results are clearly invalid. There are slam-dunk criticisms that make this clear. So let’s try to stick to those rather than offering criticisms that are as bad as the analysis itself. Ok?