The “we do not know the population effects” myth

by Carl V Phillips

Some anti-THR activists are not willing to repeat the most blatant anti-THR lies, but they still need to concoct some reason for not supporting THR without admitting their true motives.  One tactic is to claim that even though it is clear that THR is beneficial[*] for individual smokers, it still might be harmful at “the population level”.  There is really no such thing “population level” health apart from individual health.  Health (unlike social phenomena such as social cohesion or inequality) exists only as an individual phenomenon, and what happens at the population level exists only as an emergent phenomenon of what happens to individuals (thus the advantage of using agent-based modeling to study THR, as I posted recently).

[*Actually, I am not sure that tobacco control industry people, even those who are not willing to blatantly lie, would use a proper term like “beneficial”.  They tend to stick to weasel words like “a reduced risk alternative for those who cannot quit” while working hard to avoid any phrasing that admits that people must be choosing the option because it offers them the greatest welfare among the available alternatives.]

If we unpack the “population level” rhetoric a bit, it is clear that it is really just a way of obscuring the statement, “we are afraid that more people will use tobacco/nicotine when they find out they can do so with very little risk”.  This is undoubtedly true (that it will happen) — people are not morons, after all, despite what the ANTZ claim.  But is it really the case that we are unsure whether the average/aggregate population health effects will be positive or negative?  No.  It is obvious that the effects will be positive.

A good way to recognize someone who really thinks like a scientist and is seeking the truth, as opposed to someone who either does not understand how to do science or is intentionally trying to obscure the truth, is how he deals with orders-of-magnitude differences.  A scientist, as a matter of second nature, sets aside considerations that are too small to matter (after doing whatever it takes to ensure this is the case).  Typically some of these smaller considerations will be included in a scientific model because there is no reason to leave them out, but there will be no suggestion that they might substantially alter the main phenomenon of interest.

There is a classic illustrative joke.  A museum visitor asks a guard how old a particular fossil is.  The guard replies “two million and seven years”.  When asked by the baffled visitor how such precision is possible, the guard replies, “Well, when I started working here they told me it was two million years old, and that was seven years ago, so….”  We might hope that even a security guard, after hanging around researchers for seven years, would understand the concept of rounding error in this context.  But since it is pretty clear that most of the “scientists” who dominate public health and health reporters never seem to learn this grade-school-level science lesson, maybe this is too optimistic.

A scientist, when addressing the population average effects of promoting THR, would quickly recognize that what matters is how the THR effort affects the smoking rate.  Almost everything else is a rounding error.  Anyone who worries about finicky details — like exactly how many people will be using THR who would have otherwise been abstinent — is either (a) not actually concerned about health, and is using it to hide her real motives, or (b) is functionally innumerate.

Smokeless tobacco is about 99% (+/-1%) less harmful than smoking, and other smoke-free alternatives (e-cigarettes, pharma products) are similar (they might be a bit worse, but it would be shocking if any of them turned out to cause even 5% of the risk from smoking).  Thus, the number of people using THR rather than being abstinent that would be needed to make a dent in the net benefits from one person switching from smoking is quite large.  In fact, it is so large that the count of such people does not really even matter.  It is a rounding error.  Mis-estimating the reduction in smoking by 1% (which is pretty much inevitable) swamps the implications of that other estimate, so what really matters is improving the accuracy of the smoking reduction estimate.

Notice that I said this is almost the only thing that matters.  The only other number that has effects that are not swamped by uncertainty and rounding errors in the smoking rate (i.e., the only number that is worth not ignoring) is the impact of THR on causing some smokers to reduce but not eliminate their smoking.  Reducing smoking intensity reduces risk, though less than linearly (i.e., cutting down by half reduces your risk, but the reduction is by less than half).  So the number of smokers partially adopting THR, and how much they reduce their smoking, could have an effect worth measuring.

But notice that this points in the same direction as the main effect:  a reduction in smoking and thus a substantial reduction in average risk in the population.  The rate of use of THR alternatives is not even worth measuring when asking the question “are we sure the net effect is positive?”

The ANTZ rhetoric surrounding this issue often includes claims about how THR will cause some people to take up smoking, but there is no apparent way in which promoting something as a better alternative to smoking could lead to smoking.  The existence of smoke-free products might keep more people smoking — it tends to interfere with the efforts to torment smokers with place restrictions (prohibitions against smoking in certain places) because they can temporize with the smoke-free alternative.  Setting aside the unethical nature of such tactics, they undoubtedly work to some extent (they are akin to The Inquisition or criminalizing consensual sex — it is possible to torture or threaten someone enough that they will give up something that is very important to them).  But the actual promotion of THR, rather than just the existence of the products, has no such effect.  Besides, since no one has ever suggested banning all smoke-free nicotine products (including pharma products, which are often used to minimize the torment of place restrictions), it is difficult to believe that anyone is really motivated by this.

The context where this “population level” myth shows up most importantly is in US FDA policy, which will not allow a manufacturer to make claims about something being lower risk than smoking until the “population effect” is researched.  Thus, it is THR — not the product itself — that the “population level” myth is being used to interfere with.

With all that in mind, there is one apparent way in which education about THR could cause someone to smoke more than he otherwise would:  It could be that someone’s utility from abstinence would have been higher than that from smoking (so he would have quite entirely given only those options) but that the utility from reducing smoking to take advantage of some THR but then continuing to smoke is higher still.  Is that a plausible scenario, something other than an occasional curiosity?  Addressing that requires thinking about the utility function that measures the costs and benefits from smoking, something that the ANTZ and those dependent on their funding will probably never do (because they like to pretend there are no benefits).  I will let you know when I make some progress on the question.