New study: smoking may be harmless in the 21st century

by Carl V Phillips

The abstract of the new study, just published in the Journal of Abstracts that are More Educational than an Entire MPH Curriculum:

Title: The potential harmlessness of smoking

Background: While smoking was demonstrated to be a widespread and moderately-high-magnitude health threat in the 20th century, less is known about what effects it will have in the 21st.

Methods: Systematic review of journal articles reporting measures of disease risk from smoking during the 21st century.

Results: In contrast with the overwhelming evidence that smoking during the 20th century caused death and disease, no evidence was found to support a similar claim about the 21st century. No RCTs were found, and thus the effects cannot be proven to exist. Moreover, the observational studies that reported effects of smoking during the 21st century consistently failed to control for overwhelming confounding by the well-known risk factor, smoking in the 20th century, and thus the results are highly suspect.

Conclusions: Smoking during the present century may be harmless. All smoking-related public policies that are premised on smoking being harmful should be immediately revoked, in accordance with the precautionary principle and the requirement to do no harm.

Let’s unpack that. (If you were enjoying this as just a joke, pause for a second, because I am about the ruin the humor by explaining it. If you want to read a much longer piece — I actually wrote the whole paper — with hundreds of little jokes you have to find yourself, check out this one.)

As should be obvious to regular readers, the abstract mimics the tactics of a certain genre of anti-THR propaganda, often published in journals, of seeking to create ignorance from knowledge rather than the other way around. That is, instead of trying to optimally answer an important question (is smoking today harmful? does vaping pose very low risk?) by bringing to bear whatever information we have, they manufacture doubt by demanding a particular type  of information and then declaring nothing useful is known when they do not find it. This is basically the same anti-science that has been perfected by Creationists and UFOlogists. I often get the impression that most people in “public health” and medicine understand science so poorly — far less well than an inquisitive 6th-grader — that they actually do not realize how wrong this is. But, of course, they are also motivated by their extremist political views to act as if they think that way, so it is hard to be sure.

The abstract is based on a classic epistemology (basically, the philosophical study of how we know stuff) exercise about inductive logic: The example is based around the conclusion that “all swans are and will continue to be white” which we conclude because we have observed “every swan we have ever seen has been white” and the inductive reasoning is that this pattern will continue. This worked out fine until the people making that observation “discovered” Australia and its black swans. And then that very convincing bit of induction failed.

But inductive reasoning — which, roughly, means “making the best assessment we can about general rules based on the finite specifics we can observe” — is all science has. This contrasts with the deductive reasoning available in mathematics, where you start start with a postulate and then can prove its implications. There is no proof in science. Thus the joke in the abstract about clueless health researchers who seem to think that a particular type of research can prove conclusions while other types are epistemically different. The reality is that all useful evidence allows useful inference (but never proof) while bad evidence does not, and that a particular bit of evidence may either be good or bad regardless of the study design.

But, you are probably asking, what is good or bad? That is too big a question — sorry. But to take one little piece of it, the most obviously defensible bit of inductive reasoning is “what is suggested by the observations we have so far will persist in the future, given that the relevant conditions are the same.” So anyone with a modicum of scientific insight would conclude that smoking will continue to be unhealthy this century, based on the observations that cigarettes and human biology did not change substantially around Y2K, and progress in relevant medical technology is woefully slow (funny, it is almost like there is a concerted effort to avoid developing better screening and treatments for lung cancer). You would have to be a complete idiot, or from a certain subpopulation of philosophy professor, to suggest that the only possible basis for concluding that smoking right now is harmful is observing the effects of smoking right now.

But (I am going to see how many paragraphs I can begin that way) you then ask how do we know for sure whether “the relevant conditions are the same”? After all, everything seemed about the same, year after year, until that pesky black swan was discovered. The answer is we cannot. It is always possible that something will change. Indeed, in health science everything is always changing and is always different across different populations. Physicists know their “constants” might change and, indeed, imagine other universes and even parts of this universe where they are different. But in health, even these quasi-constants do not exist. You know you are reading the work of someone who understands little of science when they present the result of some health studies, down to the last decimal place, as if were a measure of some universal constant.

Using the same inductive logic that is pointedly ignored in the abstract, we can conclude that smoking is harmful for left-handed men in Tikrit, even though this has never been studied. (I assume, though given the billions of dollars wasted on pointless tobacco research, I suppose it is possible.) But we can also be quite confident that the harms are far less than they are for the populations that are typically studied, because those men are far more likely to die of something else before smoking causes them any problems. Similarly, despite the widespread desire to just let smokers die, we can be confident that medical technology will dramatically reduce the harm from smoking-caused diseases in the 21st century compared to the 20th. (Though the net loss of life-years due to smoking could go either way: By that measure, the harm from smoking increased during the course of the 20th century because many causes of early death were dramatically reduced. This relates to that “moderately” in the background section of the abstract, a subtle observation that despite the dramatic rhetoric, smoking was not a particularly important killer for most of the 20th century and that is still true for much of the world today. “First-world problems” and all that.)

So you noticed that… No, wait… But while you noticed that “conditions are the same” was a bit hard to pin down, you perhaps did not notice quite how hard “suggested by the observations we have so far” is to pin down. Returning to the swan exercise, consider a classic next step in examining that logic. While we had observed that all the evidence was consistent with “all swans are and will continue to be white” it was also equally consistent with “all swans are white up until the turn of the next century, and starting then all swans will be orange.” This leads to the realization that it is very difficult to make a rule that says that one of those conclusions — both equally well supported by all the evidence and by inductive logic — is somehow better. This is true even though the better conclusion is obvious for anyone who is seeking understanding of the world rather than trying to play logic games. I will leave it there and move on, but if you want to wander down that path further, search “grue” and “bleen” (yes, really — you just have to ignore the hits that relate to Zork).

But that observation is enough to point out that the inductive reasoning that smoking will continue to be harmful this century is not fundamentally more compelling than any other inductive reasoning. For example: “Consuming well-studied smoke-free sources of nicotine (smokeless tobacco, NRT) causes no measurable disease; inhaling lots of propylene glycol in workplaces causes no measurable disease; exposure to tiny bits of acrolein, TSNA, etc. causes no measurable disease. Therefore, inhaling these in the form of e-cigarette vapor causes no substantial risk of disease.” That is a bit more complicated than the logic of smoking continuing to be harmful, of course, but there is no fundamental difference. Therefore, those who claim we cannot conclude, based on available evidence, that e-cigarettes are low-risk are engaged in the same logic that there is no evidence that smoking in the 21st century is harmful. In both cases, it is a contrived exercise in assuming-away the knowledge we have, rather than the scientific exercise of trying to make use of the evidence as best we can.

Then, of course, those people write sweeping aggressive policy conclusions based on the contrived ignorance, as I did in the abstract. They make no attempt to assess what the net expected value of the real impacts of these policies would be, or even merely assess whether the policies would produce any benefits even if the ignorance were real. Even if we genuinely believed there were no evidence that smoking during the 21st century is harmful, that obviously would not be sufficient to recommend repealing every anti-smoking policy. Similarly, even if someone genuinely believes that we do not know that e-cigarettes are low-risk, it does not follow that a ban or other particular anti-ecig policy is recommended. Throwing in phrases like “public health concerns” or “precautionary principle” (which I riffed on to show that you can invoke it no matter what your policy biases are) do not contribute any substance that substitutes for that always-absent analysis.

To reiterate, the analogy I presented here is not merely a “they kinda sorta act like this” joke like my recent parody. The exact same failure to engage in proper scientific reasoning that is present in most anti-THR claims would lead to the conclusion that smoking right now is not known to be harmful.

4 responses to “New study: smoking may be harmless in the 21st century

  1. In a hundred years or so,people will be thanking us for our foresight in inventing vaping,thereby saving millions of smokers,who would have been visited by the Grim Reaper much earlier than they should have been.

  2. Excellent piece. You never fail to impress with your work, intelligence and honesty. It’s encouraging to know there are sensible people still alive and able to research like this and express it so well.

    Thank you.

  3. Jack Listerio

    7 October, the COT meeting on 26 October and the COC meeting on 18
    November 2004.

    “5. The Committees commented that tobacco smoke was a highly complex chemical mixture and that the causative agents for smoke induced diseases (such as cardiovascular disease, cancer, effects on reproduction and on offspring) was unknown. The mechanisms by which tobacco induced adverse effects were not established. The best information related to tobacco smoke – induced lung cancer, but even in this instance a detailed mechanism was not available. The Committees therefore agreed that on the basis of current knowledge it would be very difficult to identify a toxicological testing strategy or a biomonitoring approach for use in volunteer studies with smokers where the end-points determined or biomarkers measured were predictive of the overall burden of tobacco-induced adverse disease.”

    In other words … our first hand smoke theory is so lame we can’t even design a bogus lab experiment to prove it. In fact … we don’t even know how tobacco does all of the magical things we claim it does.

    • Carl V Phillips

      Maybe it doesn’t. Maybe it only did so up until yesterday.

      Note that even if the biological mechanisms for the harms caused by smoking were fully understood — they are substantially understood, but not fully — that would not change my point. It is the epidemiology that shows us what harm is really being caused (and then modeling those mechanisms merely fills in the “why”), and I emphasized that there is no epidemiology about the present or future. But it is still the case that even if we had the perfect biological understanding of what has been observed, we would never have any proof that it would still be true tomorrow. There is no biology about the future either. That same point applies to all of science, of course, not just the matters at hand.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s